A Sliding Mode Observer for State Estimation and Reconstruction of Unknown Inputs with Arbitrary Relative Degree: Application to Chemical Processes
|
|
- Francine Walsh
- 6 years ago
- Views:
Transcription
1 A Sliding Mode Observer for State Estimation and Reconstruction of Unknown Inputs with Arbitrary Relative Degree: Application to Chemical Processes Esteban López Aguirre Chemical Engineer Universidad Nacional de Colombia Facultad de Minas Departamento de Procesos y Energía Medellín, Colombia 2015
2
3 A Sliding Mode Observer for State Estimation and Reconstruction of Unknown Inputs with Arbitrary Relative Degree: Application to Chemical Processes Un Observador por Modos Deslizantes para la Estimación de Estado y la Reconstrucción de Entradas Desconocidas con Grado Relativo Arbitrario: Aplicación a Procesos Químicos Esteban López Aguirre Chemical Engineer Thesis presented as partial requirement for the degree of: Master of Engineering - Chemical Engineering Advisor: Héctor Antonio Botero Castro, Ph.D. Grupo de Investigación en Procesos Dinámicos - KALMAN Universidad Nacional de Colombia Facultad de Minas Departamento de Procesos y Energía Medellín, Colombia 2015
4
5 To Ligia Henao and Heroína Delgado: Two shining stars in the sky above, Two true examples of strength and love.
6 Acknowledgments Firstly, I would like to express my gratitude to my advisor, Professor Héctor Botero, for his constant support and guidance; without his valuable insight and his kind disposition to help me through his expertise, this work would not have been possible. Also, I am in great debt to my other mentors, Professors Lina Gómez and Hernán Álvarez, and my fellow researchers at the Dynamical Processes Research Group - KALMAN ; their encouragement, constructive comments, and assessment were fundamental in the development of this thesis. In addition, I wish to thank my dear friends Cristian, Lina, Jorge, Mateo, Julián, and Felipe for the gift of their company, which allowed me to clear my mind when necessary and to keep motivated. Finally, I am immensely grateful to my parents María Leticia and Marino and to my sister Carolina; they are the foundation upon which I have constructed all the achievements in my life, and their unconditional love was always a source of fortitude throughout the educational process that has led to the formulation of this thesis.
7 vii Abstract This work proposes a sliding mode observer for the estimation of the state variables and the reconstruction of the unknown inputs of nonlinear dynamical systems regardless of whether or not these systems satisfy relative-degree-related conditions. A chronological review of the evolution of sliding mode techniques for state estimation is presented and the two main approaches for the design of sliding mode observers are described. The mathematical tools, including a Lyapunov-like theorem for fixed-time attraction, a high order sliding mode differentiator, and an original method for the dynamical solution of time-varying algebraic equations, that are necessary to carry out this work are appropriately described. In addition, two approximate methods to assess if the proposed observer produces accurate estimates of the variables of a given system are presented. The use of the proposed estimator is illustrated by simulating its application to several chemical and biochemical systems, to which an extended Kalman filter is also applied for comparison purposes. This comparison leads to the conclusion that, in general, the devised sliding observer presents a significantly better performance. Finally, simulations involving situations of particular interest for the developed observer such as non-smoothness of the unknown inputs, presence of noise in the output, and coupling with a controller are presented. Keywords: Sliding mode observers, unknown input reconstruction, fixed-time stability, zero-finding methods.
8 viii Resumen Este trabajo propone un observador por modos deslizantes para la estimación de las variables de estado y la reconstrucción de las entradas desconocidas de sistemas dinámicos no lineales independientemente de si estos sistemas satisfacen o no condiciones relacionadas con el grado relativo. Se presenta una revisión cronológica de la evolución de las técnicas por modos deslizantes para la estimación de estado y se describen los dos enfoques principales para el diseño de observadores deslizantes. Las herramientas matemáticas, incluyendo un teorema tipo Lyapunov para la atracción en tiempo fijo, un diferenciador por modos deslizantes de alto orden y un método original para la solución dinámica de ecuaciones algebraicas variantes en el tiempo, que son necesarias para llevar a cabo este trabajo son descritas apropiadamente. Además, se presentan dos métodos aproximados para evaluar si el observador propuesto produce estimativos exactos de las variables de un sistema dado. El uso del estimador propuesto se ilustra mediante la simulación de su aplicación a varios procesos químicos y bioquímicos, a los cuales también les es aplicado un filtro de Kalman extendido con propósitos comparativos. Dicha comparación lleva a la conclusión de que, en general, el observador deslizante ideado presenta un desempeño significativamente superior. Finalmente, simulaciones que involucran situaciones de particular interés para el observador desarrollado tales como entradas desconocidas no suaves, presencia de ruido en la salida y acoplamiento con un controlador son presentadas. Palabras clave: Observadores por modos deslizantes, reconstrucción de entradas desconocidas, estabilidad en tiempo fijo, métodos para hallar raíces.
9 Contents List of Figures List of Tables List of Symbols and Abbreviations xii xiii xiv 1 Introduction Motivation Research Problem Objectives General objective Specific objectives Notation Publications Thesis Outline Sliding Modes in State Estimation A Historical Review Most Common Types of Sliding Mode Observers Luenberger-like sliding mode observer Sliding mode observer based on transformation to triangular form Chapter Summary Mathematical Tools Extended Kalman Filter Finite-Time Stability Sliding Mode Differentiator Sliding Mode Solver for Systems of Time-Varying Algebraic Equations Chapter Summary Proposed Sliding Mode Observer Description of The Proposed Observer Methods for Assessment of Convergence A method based on linearization ix
10 x Contents A method based on multiple simultaneous implementation Chapter Summary Simulations Batch Reactor Continuous Bioreactor Continuous Stirred Reactors in Series Simulations Under Adverse Circumstances Non-smooth unknown input Noisy output Application to Automatic Control Chapter Summary Conclusions and Future Work Conclusions Future Work References 78
11 List of Figures 3.1 Differentiation of φ(t) = sin(t) + cos(t) Suggested value of γ as a function of the initial conditions Solution of (3.24) Proposed observer scheme Value of the index (4.26) as a function of time during the application of the proposed SMO to system (4.27) State estimation and unknown input reconstruction for system (4.27) using the proposed SMO Value of the index (4.26) as a function of time during the application of the proposed SMO to system (4.29) State estimation and unknown input reconstruction for system (4.29) using the proposed SMO State estimation and unknown input reconstruction for system (4.31) using the proposed SMO with three different sets of initial conditions State estimation and unknown input reconstruction for system (4.36) using the proposed SMO with three different sets of initial conditions Illustration of a jacketed batch reactor State esatimation and unknown input reconstruction for a batch reactor Illustration of a continuous bioreactor Value of the index (4.26) as a function of time during the application of the proposed SMO to system (5.6) State esatimation and unknown input reconstruction for a bioreactor Illustration of two continuous stirred reactors in series Value of the index (4.26) as a function of time during the application of the proposed SMO to system (5.12) State esatimation and unknown input reconstruction for a system of two continuous stirred reactors in series State esatimation and unknown input reconstruction for a CSTR affected by a non-smooth unknown input Temperature inside a CSTR and its noisy measurement xi
12 xii List of Figures 5.11 State esatimation and unknown input reconstruction for a CSTR with noisy temperature measurements ITAE index for the state estimation of a CSTR with different standard deviations for the output noise Concentraction control, state esatimation, and unknown input reconstruction for a CSTR Control action for the concentraction control in a CSTR
13 List of Tables 5.1 Parameters for the batch reactor model ITAE index for the state estimation of a batch reactor Parameters for the bioreactor model ITAE index for the state estimation of a continuous bioreactor Parameters for the model of two continuous stirred reactor in series ITAE index for the state estimation of a system of two continuous stirred reactors in series ITAE index for the state estimation of a CSTR affected by a non-smooth unknown input ITAE index for the state estimation of a CSTR with noisy temperature measurements IAE index for the regulation error of several simulations of the CSTR system 69 xiii
14 List of Symbols and Abbreviations Latin Letters Symbol A C d f F g G h h j H Hj i I n I L L i f m M n p q P P 0 Description Linearized system matrix Linear or linearized output matrix Vector of unknown inputs Function that defines a dynamical system Function that defines a system of algebraic equations Input distribution matrix function Function of actual derivatives of the outputs Output function j-th component of the output function Function of theoretical derivatives of the outputs i-th time derivative of the j-th output as a function of the state and the time derivatives of the unknown inputs Identity matrix of size n Index for assessment of convergence Parameter in sliding mode differentiator Lie derivative operator of the i-th order along f Number of outputs Linear input distribution matrix Order of a dynamical system Number of unknown inputs Number of algebraic equations in a system Estimation error covariance matrix Initial condition of the estimation error covariance matrix xiv
15 xv Symbol P c P i q Q Q N r j rj, k R N t t 0 T T c T v v N V V 0 w ŵ w w N x X ˆx ˆx 0 y y j ŷ z Description IAE index for regulation error ITAE index for the estimation error of the i-th state variable Number of equations/variables in a system of algebraic equations Linearized error dynamics matrix for the degrees of freedom Process noise covariance matrix Relative degree of the j-th output Relative degree of the j-th output with respect to the k-th unknown input Measurement noise covariance matrix Time Initial time Settling-time function Upper bound on convergence time State transformation Vector of switching terms Measurement noise Lyapunov function candidate Initial condition of the Lyapunov function Vector of degrees of freedom Estimate of w Estimation error for w Process noise State vector Set of admissible states Estimate of the state vector Initial condition of the estimated state vector Output vector j-th output Estimate of the output vector Vector of unknown variables in a system of algebraic equations
16 xvi List of Symbols and Abbreviations Symbol z ẑ z 0 Z Description Solution to a system of algebraic equations Estimate of z Initial condition for the sliding mode algebraic solver Set of admissible values for unknown variables Greek Letters Symbol γ δ k λ i ν j ξ σ ζ i Description Parameter in fixed-time attraction theorem Highest derivative of the k-th unknown input considered in the estimation Tuning parameter in sliding mode differentiator Number of derivatives of the j-th output used in the state estimation Transformed state vector Sliding variable Estimate of the i-th derivative in the sliding mode differentiator Subindices Symbol eq Description Equivalent signal Abbreviations Abbreviation CSTR EKF IAE Meaning Continuous stirred tank reactor Extended Kalman filter Integral of absolute error
17 xvii Abbreviation ITAE MIMO PI SISO SMO Meaning Integral of time-weighted absolute error Multiple-inputs multiple-outputs Proportional-integral Single-input single-output Sliding mode observer
18
19 1 Introduction 1.1 Motivation In the particular case of Control Theory applied to chemical processes, there are plenty of important variables that are very difficult to measure in real-time; some notable examples are concentrations inside chemical reactors, crystal size in crystallization operations, average molecular weight in polymerization reactions, moisture content in drying processes, and phase compositions in distillation towers. If these variables are not measured online, they cannot be monitored or controlled; because of this, the desired products may fail to satisfy the necessary quality specifications. Moreover, if these unsupervised variables get outside their safety ranges, damages and accidents may occur (Isermann, 2011). This problem may be approached by using conventional state estimators that assume the available system model is perfect and faithfully represents the real behavior of the process. Nonetheless, this assumption is rarely close to be valid in practice. In most cases, during the modeling process necessary simplifications are made in order to produce a model that is feasible and practical. In addition to this conscious simplifications, there often exist unidentified phenomena that are not taken into account while modeling and generate uncertainty. Another situation than hinders the correspondence of the model with reality is the presence of external disturbances, which are scenarios outside the foreseen operating conditions or changes in variables that affect the dynamical behavior of the process and cannot be measured. Finally, parametric changes, which are common in chemical processes due to variations of the properties of substances and materials over time, represent an additional difficulty for a model to predict the real evolution of a system. In the presence of situations like the ones described above, a software sensor based on a model that does not consider uncertainties usually delivers incorrect estimates for the state variables of the system; therefore, the development of state estimators that are robust against unknown inputs is of great interest. These unknown inputs are variables that affect the system (generally modifying the model through a structure that is assumed to be known) and may group different kinds of uncertainties. Besides, estimating the value of these unknown inputs is useful for fault detection, model assessment, parameter fitting, and identification of unknown dynamics. 1
20 2 1 Introduction Being aware of these facts and the advantages of sliding mode techniques, the scientific community has proposed and studied sliding mode observers with the ability to estimate the state variables of a dynamical system in the presence of uncertainties (Floquet and Barbot, 2007; Fridman et al., 2008; Martínez-Guerra et al., 2004; Walcott and Zak, 1986; Xiong and Saif, 2001). However, continuing to advance in the overcoming of the obstacles faced when handling unknown inputs in state estimation is of great academic and practical interest to Chemical Engineering and Control Theory. 1.2 Research Problem Information is certainly a valuable asset in the practice of engineering; it is fundamental for making decisions, generating reports, detecting possible problems, and developing solutions. Therefore, every engineer constantly faces the necessity of handling, finding, and processing information. Lately, this concept has acquired such a level of significance that the classic paradigm of industrial processes has been modified to include information processing in addition to matter and energy transformations (Álvarez et al., 2009). In particular, knowing the online values of the important variables in a process is highly desirable and, in many cases, indispensable for Chemical Engineering. This knowledge is used for monitoring (Botero and Álvarez, 2009; Martínez-Guerra et al., 2004; Dochain et al., 2009), fault detection and isolation (Floquet et al., 2004; Yan and Edwards, 2005; Nagesh and Edwards, 2014; Menon and Edwards, 2014; Ríos et al., 2014), and feedback control (Wang et al., 1997; Loza et al., 2013; Chalanga et al., 2014). In the best case scenario, all the variables of interest in an industrial process are measured online by means of very accurate physical sensors. Nonetheless, this ideal situation is usually precluded by several factors. In some cases, the cost of necessary sensors is prohibitive, while in other occasions the sensors for the properties of interest do not exist. It is also possible that the required measurement instruments exist but they are unreliable, are not suitable for the operating conditions, or provide delayed information (de Assis and Maciel-Filho, 2000; Oliveira et al., 2006; Mohd Ali et al., 2015). Although the impossibility to measure important variables is particularly prominent in chemical and biological processes, this obstacle is not exclusive to these fields but may affect any dynamical system. Consequently, Control Theory has approached this problem by implementing algorithms that seek to infer the behavior of the unknown variables based on the knowledge of the variables whose online values can be measured appropriately. These algorithms are called state estimators; furthermore, when the systems they are applied to are deterministic, they are termed state observers (Besançon, 2007; Botero, 2010). The union of the available physical sensors and the state estimation algorithm is called software sensor
21 1.2 Research Problem 3 (de Assis and Maciel-Filho, 2000). Some of the state observers reported in the literature are the Luenberger observer, the asymptotic observer, the adaptable observer, the high gain observer, and the sliding mode observer (SMO) (Botero, 2010). However, in many cases, these observers are based on models that are assumed to be perfect, which cause estimation errors when model uncertainties or disturbances are present. Among the aforementioned kinds of estimators, sliding mode observers have acquired great interest in the last decades due to their insensitivity to some uncertainties and their possibility to converge in a finite time interval (Spurgeon, 2008). Sliding mode applications to state estimation and automatic control are characterized by the use of discontinuous or non-smooth correction terms; although these terms are the key to the advantages of sliding mode techniques, they are also the source of their main disadvantage, which consists of small amplitude, high frequency oscillations in the response of the treated system known as chattering (Edwards and Spurgeon, 1994; Walcott and Zak, 1986). Current research efforts on this topic focus on chattering reduction and improvement of robustness (Ginoya et al., 2014; Utkin, 2013; Yang et al., 2013). Generally, the methods intended for robust state estimation based on sliding mode theory make use of uncertain models which are affected by unknown inputs through a known structure. The implemented observers are expected to be independent of these unknown inputs or to be able to estimate their values, thus achieving robustness against some uncertainties. However, there are considerable structural restrictions to the kinds of unknown inputs conventional sliding mode techniques can deal with. Some of the proposed siding mode observers for systems with unknown inputs require the satisfaction of the matching condition (Shtessel et al., 2014). This condition demands that the unknown inputs act upon the outputs in such a way that it is possible to match them to the output error injection term in order to cancel their effect on the estimation error dynamics. This condition is mathematically defined for linear systems and, in the singleinput single-output (SISO) case, it is equivalent to a requirement that asks for the relative degree of the unknown input with respect to the output to equal 1. Another tendency, focused on nonlinear systems, has developed sliding mode observers for dynamical systems in triangular observable form (Barbot et al., 1996; Sánchez-Torres et al., 2011a), which, in the absence of known inputs, corresponds to a cascade of integrators where the output is the first state variable. When an unknown input is considered, it can only affect the last differential equation (Floquet and Barbot, 2007). These methods are ideal for simple mechanical systems since they take this triangular form naturally; nevertheless, to
22 4 1 Introduction apply these ideas to other kinds of systems, it is necessary to employ a state transformation to put them into triangular observable form, which requires the relative degree to equal the order of the system. This imposes a restriction as strong as the matching condition; furthermore, if a state transformation is used, the estimation is not carried out explicitly for the original state variables and it is necessary to apply the inverse transformation to obtain them, something that is rarely an easy task. Although efforts to relax the previously described conditions have been carried out (Barbot et al., 2009; Floquet and Barbot, 2006; Fridman et al., 2008; Xiong and Saif, 2001), the resulting methods are based on state transformations that may be complex, difficult to find, and difficult to invert. 1.3 Objectives General objective To develop a sliding mode observer for deterministic nonlinear dynamical systems that, besides estimating the the value of the unmeasured state variables, permits to reconstruct their unknown inputs regardless of their relative degree or whether they satisfy or not the matching condition Specific objectives 1. To identify the most used techniques for the state estimation of dynamical systems. 2. To describe the advantages and disadvantages of the use of sliding mode techniques in state estimation. 3. To characterize the sliding mode observers that have been applied to uncertain dynamical systems in the literature. 4. To describe the mathematical tools useful in the formulation of the proposed technique. 5. To propose a nonlinear sliding mode observer for state estimation and reconstruction of unknown inputs. 6. To validate the performance of the proposed technique through the simulation of its application to chemical systems and the comparison with a conventional state estimator.
23 1.4 Notation Notation When reading this dissertation, the following items regarding notation should be considered: Vector and matrix variables are displayed using bold face type, although regular mathematical type is used for vector and tensor fields. The operator corresponds to the Euclidean norm. The symbol 0 represents either the zero vector or the zero matrix of dimensions that are easily determined from the context, and the symbol 0 m n corresponds to the zero matrix with a specified size of m n. Given any function f of the form f(z 1, z 2,..., z 3 ) that may be evaluated at the estimated variables ẑ 1, ẑ 2,..., ẑ 3, the notation for the Jacobian matrix f z i is reduced to z1 f ẑ =ẑ 1,z 2 =ẑ 2,...,z 3 =ẑ i for the sake of simplicity Publications The following article, based on preliminary results obtained in the process of devising this thesis, has been submitted to an international journal: López, E. and Botero, H. Finite-time sliding mode observer for uncertain nonlinear systems based on a tunable algebraic solver. International Journal of Robust and Nonlinear Control. (Pending submission of revised version). 1.6 Thesis Outline This thesis is structured as follows: Chapter 2 aims to provide a brief chronological review of state estimation through sliding mode techniques along with a general description of the most common sliding mode observers for systems with unknown inputs. The mathematical tools necessary to fulfill the objectives of this work, including an original sliding mode method for the solution of systems of time-varying algebraic equations, are presented in Chapter 3. Chapter 4 introduces the main contribution of this dissertation, which is a sliding mode observer for the state estimation and unknown input reconstruction of systems that do not necessarily satisfy the conditions usually requested by other observers in the literature. The application of the proposed observer to chemical and biochemical processes is simulated in Chapter 5, where a comparison with a conventional state estimation technique is also carried out. In Chapter 6 the conclusions of this work are presented and some recommendations for possible future work are provided.
24 2 Sliding Modes in State Estimation The purpose of this chapter is to provide a review of the the use of sliding mode techniques in state estimation, creating the background necessary to characterize the disadvantages of the current methods whose overcoming will eventually become the the contributions of this work. Section 2.1 consists of a brief chronological review of the evolution of sliding mode observers. Section 2.2 describes the two most common sliding mode techniques used to estimate the state of uncertain dynamical systems along with the restrictions they impose. 2.1 A Historical Review In the Mid-Twentieth Century, a control strategy that used variable structure systems was started to be investigated in the Soviet Union (Utkin, 1992). This strategy implemented control laws that were discontinuous with respect to the state variables so that controllers would switch among different structures. In the decade of 1960 a particular kind of variable structure control system, known as sliding mode control, emerged (Spurgeon, 2008). Sliding mode techniques for automatic control acquired great interest due its insensitivity to a certain class of uncertainties and disturbances (Utkin, 1977; Draženović, 1969). When such techniques are used, once the dynamical system has reached the region of the state space where the control law is discontinuous, the control signal switch among structures with a theoretically infinite frequency and forces the state trajectory to remain in this region, which is called the sliding manifold. This ability to confine the system to a particular subset of its state space provides significant advantages (Sabanovic et al., 2004), which have been widely exploited and continue to be researched (Shtessel et al., 2014). At first, the most studied application of sliding mode techniques was automatic control. However, in the decade of 1980 the pioneer work of Utkin (1981), Slotine et al. (1986, 1987), and Walcott and Zak (1986) paved the way for a promising research field by inquiring into the application of sliding modes to state estimation. These first explorations considered systems with a known linear part and possible nonlinearities or disturbances grouped as uncertainties (Edwards and Spurgeon, 1994); nevertheless, sliding mode observers did not take long to be applied to certain classes of nonlinear systems, even achieving convergence in finite time for all the estimates of the state variables (Drakunov, 1992; Barbot et al., 1996; Drakunov and Utkin, 1995). 6
25 2.2 Most Common Types of Sliding Mode Observers 7 With the passage of time, the study of sliding observers became popular among a part of the scientific community and, although some applications to chemical processes exist Wang et al. (1997); Martínez-Guerra et al. (2004), most researchers focused on mechanical and electromechanical systems (Kao and Moskwa, 1995; Chen and Moskwa, 1997; Bartolini et al., 2003; Aurora and Ferrara, 2007). The low interest in the application of these techniques to chemical systems was inherited from sliding mode control, which did not find a niche among systems of this type since their typical actuators would not be able to reproduce discontinuous control signals. Just like sliding mode controllers, sliding mode observers may be designed to be insensitive to certain types of unknown inputs so that they deliver the true values of the state variables even in the presence of some uncertainties or disturbances. Due to these robustness qualities, sliding mode observers were, and continue to be, used for fault detection and isolation (Edwards and Spurgeon, 2000; Edwards et al., 2000; Yan and Edwards, 2005; Nagesh and Edwards, 2012, 2014). Nonetheless, in order for sliding mode observers to possess this insensitivity to unknown inputs, the treated uncertain systems must routinely satisfy significantly restrictive requirements such as the observer matching condition (Floquet and Barbot, 2006) or relative degree conditions (Barbot et al., 2009). With the objective of bypassing these restrictions, Xiong and Saif (2001) and Fridman et al. (2008) proposed to use a state transformation defined by the Lie derivatives of the outputs that do not involve the unknown inputs and some other function that can be proven to exist, but that are not constructively given. In addition, Floquet and Barbot (2006) and Barbot et al. (2009) developed constructive algorithms that, in some cases, permit to find a state transformation that converts the system into a set of triangular blocks which are suitable for the application of previously designed observers intended for systems with this structure. However, the only way to find whether these algorithms work or not for a particular system is to follow them until the corresponding conclusion is reached at some step. 2.2 Most Common Types of Sliding Mode Observers In this section, the two main currents of sliding mode observer design are described. The first one adds discontinuous correction terms to the model of the system and requires the satisfaction of a matching condition to be able to be insensitive to the unknown inputs, while the second one is based on the transformation of the system into a series of triangular blocks and imposes conditions on the relative degree of the outputs in order for these blocks to be independent of the unknown inputs. Both kinds of observers are designed for a system of the form
26 8 2 Sliding Modes in State Estimation ẋ = f(x) + g(x) d y = h(x) (2.1) where x X R n is the state vector, y R m is the output vector, d R p is the vector of unknown inputs, f : X R p R n, g : R n R n p, and h : R n R m are sufficiently smooth functions, and X is the set of admissible states. The objective of the observers is to produce an estimated state vector ˆx that converges to the real state x of the system Luenberger-like sliding mode observer One of the first observers for deterministic systems was proposed by Luenberger (1964, 1966). Its working principle consists of adding output error injection terms to the model of the system evaluated at the state estimates. The most simple sliding mode observers mimic this principle but they use injection terms that are discontinuous functions of the output error which seek to drive the sliding variable σ = ŷ y to zero. Here, ŷ = h(ˆx) is the estimated output vector. The application of of this kind of observer to system (2.1) takes the following form: ˆx = f(ˆx) + K v (2.2) where K is a gain matrix which may be variable and v is a vector of switching terms. A possible choice for v is ρ 1 sign(ŷ 1 y 1 ) ρ 2 sign(ŷ 2 y 2 ) v =. ρ m sign(ŷ m y m ) (2.3) where ŷ j and y j are the j-th components of ŷ and y respectively and the gains ρ j are chosen so that the system satisfies some reachability condition that ensures sliding motion on the manifold σ = 0 is achieved (Shtessel et al., 2014). Sometimes, a linear output error injection term is also included in observer (2.2) with the purpose of enlarging the set of initial conditions for which convergence to the sliding manifold is guaranteed (Shtessel et al., 2014; Spurgeon, 2008); however, such linear feedback has no effect on the observer dynamics once sliding mode is attained and is not considered in this discussion for the sake of simplicity. Let x = ˆx x be the estimation error for the state vector. Then, subtracting (2.1) from (2.2) yields the following error dynamics:
27 2.2 Most Common Types of Sliding Mode Observers 9 x = f(ˆx) f(x) g(x) d + K v (2.4) Once the sliding manifold is reached, the action of v becomes equivalent to that of a certain signal v eq (Shtessel et al., 2014) and the derivative of the sliding variable σ = ẏ ŷ vanishes; then it follows that h ˆx ˆx h x ẋ = 0 (2.5) Consequently, the equivalent output error injection signal v eq must satisfy h ˆx (f(ˆx) + K v eq) h (f(x) + g(x) d) = 0 (2.6) x Assuming h/ ˆx is full rank for all ˆx X, an appropriate selection of the gain matrix K allows ( h/ ˆx) K to be invertible so that (2.6) can be solved for v eq to produce v eq = ( ) 1 [ ] h h ˆx K h (f(x) + g(x) d) x ˆx f(ˆx) (2.7) Substituting this expression for v eq into (2.4) and replacing f(ˆx) f(x) by f to save notation yields ( ) 1 [ ] h h x = f g(x) d + K ˆx K h (f(x) + g(x) d) x ˆx f(ˆx) (2.8) Equation (2.8) corresponds to the estimation error dynamics when the system is in sliding motion and it can be rearranged as ( ) 1 [ ] [ ( ) ] 1 h h x = f + K ˆx K h h f(x) x ˆx f(ˆx) I n K ˆx K h g(x) d (2.9) x where I n is the n n identity matrix. The conventional approach to robustness in sliding mode techniques requires the error dynamics to become independent of the uncertainties on account of the equivalent correction signal that acts while the system sustains sliding mode. In this case, for the error dynamics (2.9) to be insensitive to the unknown inputs, the following is necessary: From (2.10) it follows that [ ( ) ] 1 h I n K ˆx K h g(x) = 0 x X (2.10) x
28 10 2 Sliding Modes in State Estimation ( ) 1 ( ) h h g(x) = K ˆx K x g(x) x X (2.11) It is a well-known property of matrix product that if A, B, and P are matrices such that P = A B, then rank(p) rank(a) and rank(p) rank(b). According to this, if (2.11) is satisfied the following must be true: [ ] h rank [g(x)] rank x g(x) x X (2.12) Moreover, ( h/ x)g(x) is in turn the product of some matrix and g(x), which implies [ ] h rank x g(x) rank [g(x)] (2.13) From (2.12) and (2.13) it is obvious that a necessary condition for the estimation error dynamics of this kind of observer to be invariant with respect to the unknown inputs is [ ] h rank [g(x)] = rank x g(x) x X (2.14) Equation (2.14) constitutes the so-called matching condition for state estimation. Roughly, this condition requires the unknown inputs to act in the channels of the outputs so they can be matched by the discontinuous error injection terms and canceled by their equivalent signals. If an uncertain dynamical system meets this condition, its unknown inputs are said to be matched; on the other hand, if this requirement is not satisfied, the unknown inputs are called unmatched. In the SISO case, the matching condition is equivalent to requesting the relative degree of the system to equal one. Nevertheless, when there are multiple inputs and outputs (MIMO), if less than p outputs have a relative degree equal to one, this condition cannot be satisfied, while if at least p of the m outputs possess a relative degree equal to one, it may or may not be fulfilled. When system (2.1) is such that there are constant matrices, C R m n, and M R n p that satisfy y = h(x) = C x and g(x) = M, the matching condition takes the form rank (M) = rank (C M) (2.15) which is the most common form to express it since this technique is mainly applied to linear systems.
29 2.2 Most Common Types of Sliding Mode Observers 11 Although the matching condition is necessary to apply this kind of observer to uncertain systems, it is worth noting that its satisfaction does not guarantee convergence to the true state variables; even if the observer can be designed so that (2.10) is true, the rest of Equation (2.9) still has to constitute an asymptotically stable system for the error x to converge to zero Sliding mode observer based on transformation to triangular form More recent works have proposed sliding mode observers whose estimates converge to the actual state variables in finite time. These observers are usually based on a state transformation that permits to express the treated system as a set of subsystems in triangular form (Floquet et al., 2004; Floquet and Barbot, 2007; Shen et al., 2010). This paradigm is described in this subsection. Let the output and the unknown inputs distribution matrix of system (2.1) be such that y 1 h 1 (x) y 2 h 2 (x) y = = h(x) =.. h m (x) y m (2.16) g(x) = [g 1 (x), g 2 (x),..., g p (x)] (2.17) where g k (x) R n for all k {1, 2,..., p} and y j = h j (x) R for all j {1, 2,..., m}. Assume each output y j has a well-defined relative degree r j for all x X, that is, for all j {1, 2,..., m} L gk L i fh j (x) = 0 x X k {1, 2,..., p} i < r j 1 L gk L r j 1 f h j (x) 0 x X for at least one k {1, 2,..., p} (2.18) where L i f h j(x) is the i-th Lie derivative of h j (x) along f(x), defined by L 0 fh j (x) = h j (x) L i fh j (x) = Li 1 f h j f(x) x (2.19) These Lie derivatives are also used to define the following state transformation:
30 12 2 Sliding Modes in State Estimation ξ 1 T 1 (x) ξ 2 T 2 (x) ξ = = T (x) =.. T m (x) ξ m (2.20) where the ξ j = T j (x) are defined by ξ 1 j ξ ν j j ξ 2 ξ j = = T j (x) = j. h j (x) L f h j (x). L ν j 1 f h j (x) j {1, 2,..., m} (2.21) for some positive integers ν 1, ν 2,..., ν m such that ν 1 + ν ν m = n. It is desired that T (x) does not depend on d so that, in the new coordinates, the system takes the form of a series of m triangular blocks where the unknown input may only affect the last differential equation of each block. In that case, the j-th triangular block would be given by ξ j = y j = ξ 1 j ξ 1 j ξ 2 j. ξ ν j 1 j ξ ν j j ξ 3 = j. ξ ν j j L ν j f h j(t 1 (ξ)) + p k=1 L g k L ν j 1 f h j (T 1 (ξ)) d k ξ 2 j (2.22) For this to be possible, that is for T (x) to be independent of d, the following condition must be met: ν j r j j {1, 2,..., m} (2.23) This restriction favors high relative degrees, as opposed to the matching condition. In particular, for SISO systems, condition (2.23) asks for the relative degree to equal n. Notice that, for the same system, some sets of integers ν 1, ν 2,..., ν m may meet this specification and some others may not. However, there are systems for which no set ν 1, ν 2,..., ν m that satisfies (2.23) exist; the state variables of such systems cannot be accurately estimated through this
31 2.2 Most Common Types of Sliding Mode Observers 13 method. In addition to this relative degree restriction, it is clear that the transformation T must be invertible in X in order to recover estimates for x from estimates for ξ. Then, assuming (2.23) is fulfilled, the Jacobian matrix T / x is required to be such that ( ) T rank = n x X (2.24) x This actually corresponds to an observability condition, since T / x is clearly the nonlinear observability matrix. If conditions (2.23) and (2.24) are satisfied, the following sliding mode observer, based on the work by Drakunov (1992), can be applied to the transformed system: ˆξ j 1 kj 1 sign(ψj 1 ˆξ j 1 ) ˆξ j = ˆξ 2 kj = j. 2 sign(ψj 2 ˆξ j 2 ). ˆξ ν j k ν j j j sign(ψ ν j j ˆξ ν j j ) j {1, 2,..., m} (2.25) In this expression, ˆξ j = [ˆξ 1 j, ˆξ 2 j,..., ˆξ ν j j ]T is the estimate of ξ j, the k i j are scalar gains and the ψ i j are given by ψ 1 j = y j = ξ 1 j ψj i = {k i 1 j sign(ψ i 1 i 1 j ˆξ j )} eq i 2 (2.26) where the { } eq operator produces the equivalent signal (Shtessel et al., 2014) of the switching term given as its argument. If the necessary equivalent signals are ideally known and the k i j gains, which may be variable, are such that kj i > ξ i+1 j = L i f h j (x) i {1, 2,..., νj 1} j {1, 2,..., m} k ν j j > Lν j f h j(x) + p k=1 L gk L ν j 1 f (2.27) h j (x)) d k j {1, 2,..., m} then, through arguments similar to those by Drakunov (1992), it can be proved that ˆξ j converges to ξ j in finite time for all j {1, 2,..., m}. If ˆx and ˆξ correspond to the estimated state vectors of the original and the transformed systems respectively, doing
32 14 2 Sliding Modes in State Estimation ˆx = T 1 ( ˆξ) (2.28) makes ˆx converge to x in finite time. However, in practice, the equivalent signals are usually extracted through low-pass filters, which converge asymptotically. Although this precludes a rigorous finite time convergence of the whole observer scheme, the effect of this non-ideality is commonly negligible due to quick convergence of the low-pass filters. 2.3 Chapter Summary In this chapter, a historical and practical review of state estimation via sliding mode techniques has been provided. The major milestones in the evolution of sliding mode observers have been pointed out, indicating the fundamental advantages and disadvantages of the resulting methods. In addition, two observers that are representative of the main paradigms in sliding mode estimation have been described in more detail, formally identifying the restrictions they impose on the structure of the systems they can be applied to. One of this restrictions ask for the outputs to have a small relative degree, while the other one requires them to have a large relative degree. The next chapter introduces some mathematical tools that will be useful in achieving the central objectives of this work.
33 3 Mathematical Tools This chapter describes the mathematical tools necessary to devise the sliding mode observer this thesis intends to propose. Section 3.1 introduces the conventional state estimator that will be compared with the sliding mode observer to be developed. The main purpose of Section 3.2 is to provide a Lyapunov-like theorem for fixed-time stability. Section 3.3 describes a robust high order sliding mode differentiator. Finally, Section 3.4 presents an original sliding mode technique for the solution of systems of time-varying algebraic equations, which is one of the contributions of this work and will constitute a central part of the pursued observer. 3.1 Extended Kalman Filter The specific objectives of this work express that the sliding mode observer that will be proposed must be compared with a conventional state estimator. The selected conventional estimator is the extended Kalman filter (EKF), whose continuous version is described in this section. This algorithm was chosen because it is extensively used in academic literature and it may possess a certain degree of robustness through the use of fictitious process noise (Simon, 2006). Consider a stochastic dynamical system of the form ẋ = f(x, w N ) y = h(x) + v N (3.1) where x R n is the state vector, y R m is the output vector, f : X R p R n and h : R n R m are known functions, w N R p and v N R m are assumed to be Gaussian white noises with zero mean and with covariance matrices given by Q N R p p and R N R m m respectively. Let the variable matrices A R n n, C R m n, and L R n p be defined as A = f ˆx C = h ˆx 15 (3.2) (3.3)
34 16 3 Mathematical Tools L = f w N (3.4) x=ˆx Then the continuous extended Kalman filter for system (3.1) is given by (Simon, 2006) ˆx = f(ˆx, 0) + P C T R N 1 [y h(ˆx)] Ṗ = A P + P A T + L Q N L T P C T R N 1 C P (3.5) where P is the estimation error covariance matrix. The initial conditions ˆx(t 0 ) = ˆx 0 and P(t 0 ) = P 0 for estimator (3.5) should be chosen so that ˆx 0 is the expected value of x(t 0 ) and P 0 is the expected value of [x(t 0 ) ˆx 0 ][x(t 0 ) ˆx 0 ] T. 3.2 Finite-Time Stability One of the essential characteristics of sliding mode techniques is that the sliding manifold is reached in finite time. Because of this, Lyapunov-like theorems have been developed in order to evaluate finite-time stability (Haddad et al., 2008). These theorems are commonly applied as reachability conditions to determine whether a system enters sliding motion or not or to design it in a way that ensures it does (Shtessel et al., 2014). In this work, the stronger notion of fixed-time stability is of particular interest. This kind of stability implies the existence of a bound on the convergence time that is independent of the initial conditions of the system. This section focuses on the definitions of finite-time attractive and fixed-time attractive sets and on a theorem that provides sufficient conditions to verify the latter for a dynamical system of the form ẋ = f(x, t) (3.6) Definition 3.1. (Polyakov, 2012) Let S be a non-empty subset of R n. S is said to be globally finite-time attractive for system (3.6) if, for any initial condition x(t 0 ) = x 0 R n, the solution x(t) reaches S at some finite time t = T (x 0 ) and remains there for all t T (x 0 ), where T : R n [t 0, ) is the settling-time function. Definition 3.2. (Polyakov, 2012; Sánchez-Torres et al., 2015) Let S be a non-empty subset of R n. S is said to be globally fixed-time attractive for system (3.6) if it is globally finite-time attractive for the system with a settling-time function T : R n [t 0, ) that is bounded by a constant; that is, there exists a positive constant T max such that T (x 0 ) T max for all x 0 R n.
35 3.2 Finite-Time Stability 17 Theorem 3.1. (Sánchez-Torres et al., 2015) If there exists a continuous radially unbounded function V : R n R such that (i) V (x) 0 for all x R n (ii) V (x) = 0 if and only if x S (iii) For some T c > 0 and some γ (0, 1], its time derivative V = ( V/ x)f(x) satisfies V 1 γ T c exp (V γ ) V 1 γ x R n (3.7) Then, S is globally fixed-time attractive for (3.6) with T max = t 0 + T c. Proof. Equation (3.7) is easily solved through separation of variables; letting V (x(t 0 )) = V 0, its solution is V (x(t)) [ ( )] 1/γ t t0 ln + exp( V γ 0 ) (3.8) T c Assume τ is such that [ ( )] 1/γ τ t0 ln + exp( V γ 0 ) = 0 (3.9) T c Solving (3.9) for τ produces τ = t 0 + T c (1 exp( V 0 γ )) (3.10) Then, the right-hand side of (3.8) reaches zero at t = τ = t 0 + T c (1 exp( V 0 γ )). Therefore, there exists a time t c such that t 0 t c τ and V (x(t)) reaches zero exactly at t = t c. Since V only takes non-negative values, the expression [ exp (V γ ) V 1 γ ]/(γ T c ) can only take non-positive values. In consequence, once V (x(t)) reaches zero at t = t c, it remains zero thereafter. Equivalently, x(t) S for all t t c. In addition, given that V (x) is radially unbounded and (3.7) holds for all x R n, this is true for any initial condition x(t 0 ) = x 0. From the previous discussion, it is clear that S is globally finite-time attractive for (3.6) with a settling-time function given by T (x 0 ) = t 0 + T c (1 exp( V (x 0 ) γ )) (3.11) Furthermore, since 0 < exp( V 0 γ ) 1, T (x 0 ) is such that
36 18 3 Mathematical Tools T (x 0 ) t 0 + T c (3.12) Then, S is globally fixed-time attractive for (3.6) with a settling-time bound given by T max = t 0 + T c. This theorem, which was developed by Sánchez-Torres et al. (2015), presents a convenient condition for fixed-time stability if the parameter T c appears explicitly in system (3.6) and can be modified, since this provides a straightforward way to tune the system by setting a limit for its convergence time. 3.3 Sliding Mode Differentiator Real-time differentiation of a signal is a well-known problem that has found a satisfactory solution via sliding modes. There are two main factors that encouraged the development of differentiators within the scientific community that studies sliding mode techniques. In the first place, the implementation of higher order sliding mode control requires knowledge of the time derivatives of the sliding variable. Secondly, through transformation to a triangular form like the one described in Section 2.2, the state estimation problem is reduced to a differentiation problem. The main inconvenient of numerical differentiation is the difficulty to simultaneously achieve accuracy of the estimated derivatives and robustness in the presence of noise. However, the work by Levant (1998, 2003) overcomes this problem and introduces a high order sliding mode differentiator that delivers the exact derivatives under ideal conditions and possesses robustness with respect to noise and sampling. Such differentiation takes the following form (Levant, 2003): ζ 0 = v 0 v 0 = λ k L 1 k+1 ζ0 φ(t) k k+1 sign(ζ0 φ(t)) + ζ 1 ζ 1 = v 1 v 1 = λ k 1 L 1 k ζ1 v 0 k 1 k sign(ζ 1 v 0 ) + ζ 2. (3.13) ζ k 1 = v k 1 v k 1 = λ 1 L 1 2 ζk 1 v k sign(ζk 1 v k 2 ) + ζ k ζ k = λ 0 L sign(ζ k v k 1 )
State and Parameter Estimation of a CSTR
Instituto Tecnológico y de Estudios Superiores de Occidente Repositorio Institucional del ITESO rei.iteso.mx Departamento de Matemáticas y Física DMAF - Artículos y ponencias con arbitrae 2012-10 State
More informationEvery real system has uncertainties, which include system parametric uncertainties, unmodeled dynamics
Sensitivity Analysis of Disturbance Accommodating Control with Kalman Filter Estimation Jemin George and John L. Crassidis University at Buffalo, State University of New York, Amherst, NY, 14-44 The design
More informationNonlinear Observers. Jaime A. Moreno. Eléctrica y Computación Instituto de Ingeniería Universidad Nacional Autónoma de México
Nonlinear Observers Jaime A. Moreno JMorenoP@ii.unam.mx Eléctrica y Computación Instituto de Ingeniería Universidad Nacional Autónoma de México XVI Congreso Latinoamericano de Control Automático October
More informationDynamic Integral Sliding Mode Control of Nonlinear SISO Systems with States Dependent Matched and Mismatched Uncertainties
Milano (Italy) August 28 - September 2, 2 Dynamic Integral Sliding Mode Control of Nonlinear SISO Systems with States Dependent Matched and Mismatched Uncertainties Qudrat Khan*, Aamer Iqbal Bhatti,* Qadeer
More informationMin-Max Output Integral Sliding Mode Control for Multiplant Linear Uncertain Systems
Proceedings of the 27 American Control Conference Marriott Marquis Hotel at Times Square New York City, USA, July -3, 27 FrC.4 Min-Max Output Integral Sliding Mode Control for Multiplant Linear Uncertain
More informationMathematical Theory of Control Systems Design
Mathematical Theory of Control Systems Design by V. N. Afarias'ev, V. B. Kolmanovskii and V. R. Nosov Moscow University of Electronics and Mathematics, Moscow, Russia KLUWER ACADEMIC PUBLISHERS DORDRECHT
More informationChaos suppression of uncertain gyros in a given finite time
Chin. Phys. B Vol. 1, No. 11 1 1155 Chaos suppression of uncertain gyros in a given finite time Mohammad Pourmahmood Aghababa a and Hasan Pourmahmood Aghababa bc a Electrical Engineering Department, Urmia
More informationDesigning Stable Inverters and State Observers for Switched Linear Systems with Unknown Inputs
Designing Stable Inverters and State Observers for Switched Linear Systems with Unknown Inputs Shreyas Sundaram and Christoforos N. Hadjicostis Abstract We present a method for estimating the inputs and
More informationRiccati difference equations to non linear extended Kalman filter constraints
International Journal of Scientific & Engineering Research Volume 3, Issue 12, December-2012 1 Riccati difference equations to non linear extended Kalman filter constraints Abstract Elizabeth.S 1 & Jothilakshmi.R
More informationState estimation of uncertain multiple model with unknown inputs
State estimation of uncertain multiple model with unknown inputs Abdelkader Akhenak, Mohammed Chadli, Didier Maquin and José Ragot Centre de Recherche en Automatique de Nancy, CNRS UMR 79 Institut National
More informationON CHATTERING-FREE DISCRETE-TIME SLIDING MODE CONTROL DESIGN. Seung-Hi Lee
ON CHATTERING-FREE DISCRETE-TIME SLIDING MODE CONTROL DESIGN Seung-Hi Lee Samsung Advanced Institute of Technology, Suwon, KOREA shl@saitsamsungcokr Abstract: A sliding mode control method is presented
More informationResearch Article Convex Polyhedron Method to Stability of Continuous Systems with Two Additive Time-Varying Delay Components
Applied Mathematics Volume 202, Article ID 689820, 3 pages doi:0.55/202/689820 Research Article Convex Polyhedron Method to Stability of Continuous Systems with Two Additive Time-Varying Delay Components
More informationA sub-optimal second order sliding mode controller for systems with saturating actuators
28 American Control Conference Westin Seattle Hotel, Seattle, Washington, USA June -3, 28 FrB2.5 A sub-optimal second order sliding mode for systems with saturating actuators Antonella Ferrara and Matteo
More informationState-norm estimators for switched nonlinear systems under average dwell-time
49th IEEE Conference on Decision and Control December 15-17, 2010 Hilton Atlanta Hotel, Atlanta, GA, USA State-norm estimators for switched nonlinear systems under average dwell-time Matthias A. Müller
More informationCHATTERING-FREE SMC WITH UNIDIRECTIONAL AUXILIARY SURFACES FOR NONLINEAR SYSTEM WITH STATE CONSTRAINTS. Jian Fu, Qing-Xian Wu and Ze-Hui Mao
International Journal of Innovative Computing, Information and Control ICIC International c 2013 ISSN 1349-4198 Volume 9, Number 12, December 2013 pp. 4793 4809 CHATTERING-FREE SMC WITH UNIDIRECTIONAL
More informationIEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY 1
IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY 1 Robust Control With Exact Uncertainties Compensation: With or Without Chattering? Alejandra Ferreira, Member, IEEE, Francisco Javier Bejarano, and Leonid
More informationMinimum-Phase Property of Nonlinear Systems in Terms of a Dissipation Inequality
Minimum-Phase Property of Nonlinear Systems in Terms of a Dissipation Inequality Christian Ebenbauer Institute for Systems Theory in Engineering, University of Stuttgart, 70550 Stuttgart, Germany ce@ist.uni-stuttgart.de
More informationGramians based model reduction for hybrid switched systems
Gramians based model reduction for hybrid switched systems Y. Chahlaoui Younes.Chahlaoui@manchester.ac.uk Centre for Interdisciplinary Computational and Dynamical Analysis (CICADA) School of Mathematics
More informationCHAPTER 5 ROBUSTNESS ANALYSIS OF THE CONTROLLER
114 CHAPTER 5 ROBUSTNESS ANALYSIS OF THE CONTROLLER 5.1 INTRODUCTION Robust control is a branch of control theory that explicitly deals with uncertainty in its approach to controller design. It also refers
More informationNavigation and Obstacle Avoidance via Backstepping for Mechanical Systems with Drift in the Closed Loop
Navigation and Obstacle Avoidance via Backstepping for Mechanical Systems with Drift in the Closed Loop Jan Maximilian Montenbruck, Mathias Bürger, Frank Allgöwer Abstract We study backstepping controllers
More informationEML5311 Lyapunov Stability & Robust Control Design
EML5311 Lyapunov Stability & Robust Control Design 1 Lyapunov Stability criterion In Robust control design of nonlinear uncertain systems, stability theory plays an important role in engineering systems.
More informationDesign of Sliding Mode Control for Nonlinear Uncertain System
Design of Sliding Mode Control for Nonlinear Uncertain System 1 Yogita Pimpale, 2 Dr.B.J.Parvat ME student,instrumentation and Control Engineering,P.R.E.C. Loni,Ahmednagar, Maharashtra,India Associate
More informationCopyrighted Material. 1.1 Large-Scale Interconnected Dynamical Systems
Chapter One Introduction 1.1 Large-Scale Interconnected Dynamical Systems Modern complex dynamical systems 1 are highly interconnected and mutually interdependent, both physically and through a multitude
More informationDisturbance Attenuation for a Class of Nonlinear Systems by Output Feedback
Disturbance Attenuation for a Class of Nonlinear Systems by Output Feedback Wei in Chunjiang Qian and Xianqing Huang Submitted to Systems & Control etters /5/ Abstract This paper studies the problem of
More informationObservability for deterministic systems and high-gain observers
Observability for deterministic systems and high-gain observers design. Part 1. March 29, 2011 Introduction and problem description Definition of observability Consequences of instantaneous observability
More informationOutput Feedback Bilateral Teleoperation with Force Estimation in the Presence of Time Delays
Output Feedback Bilateral Teleoperation with Force Estimation in the Presence of Time Delays by John M. Daly A thesis presented to the University of Waterloo in fulfilment of the thesis requirement for
More informationObserver-based quantized output feedback control of nonlinear systems
Proceedings of the 17th World Congress The International Federation of Automatic Control Observer-based quantized output feedback control of nonlinear systems Daniel Liberzon Coordinated Science Laboratory,
More informationHIGHER ORDER SLIDING MODES AND ARBITRARY-ORDER EXACT ROBUST DIFFERENTIATION
HIGHER ORDER SLIDING MODES AND ARBITRARY-ORDER EXACT ROBUST DIFFERENTIATION A. Levant Institute for Industrial Mathematics, 4/24 Yehuda Ha-Nachtom St., Beer-Sheva 843, Israel Fax: +972-7-232 and E-mail:
More informationLessons in Estimation Theory for Signal Processing, Communications, and Control
Lessons in Estimation Theory for Signal Processing, Communications, and Control Jerry M. Mendel Department of Electrical Engineering University of Southern California Los Angeles, California PRENTICE HALL
More informationHow to Implement Super-Twisting Controller based on Sliding Mode Observer?
How to Implement Super-Twisting Controller based on Sliding Mode Observer? Asif Chalanga 1 Shyam Kamal 2 Prof.L.Fridman 3 Prof.B.Bandyopadhyay 4 and Prof.J.A.Moreno 5 124 Indian Institute of Technology
More informationModel-Based Linear Control of Polymerization Reactors
19 th European Symposium on Computer Aided Process Engineering ESCAPE19 J. Je owski and J. Thullie (Editors) 2009 Elsevier B.V./Ltd. All rights reserved. Model-Based Linear Control of Polymerization Reactors
More informationThe ϵ-capacity of a gain matrix and tolerable disturbances: Discrete-time perturbed linear systems
IOSR Journal of Mathematics (IOSR-JM) e-issn: 2278-5728, p-issn: 2319-765X. Volume 11, Issue 3 Ver. IV (May - Jun. 2015), PP 52-62 www.iosrjournals.org The ϵ-capacity of a gain matrix and tolerable disturbances:
More informationEE C128 / ME C134 Feedback Control Systems
EE C128 / ME C134 Feedback Control Systems Lecture Additional Material Introduction to Model Predictive Control Maximilian Balandat Department of Electrical Engineering & Computer Science University of
More informationComparison of four state observer design algorithms for MIMO system
Archives of Control Sciences Volume 23(LIX), 2013 No. 2, pages 131 144 Comparison of four state observer design algorithms for MIMO system VINODH KUMAR. E, JOVITHA JEROME and S. AYYAPPAN A state observer
More informationDesign and Implementation of Piecewise-Affine Observers for Nonlinear Systems
Design and Implementation of Piecewise-Affine Observers for Nonlinear Systems AZITA MALEK A Thesis in The Department of Electrical and Computer Engineering Presented in Partial Fulfillment of the Requirements
More informationSliding Mode Control à la Lyapunov
Sliding Mode Control à la Lyapunov Jaime A. Moreno Universidad Nacional Autónoma de México Eléctrica y Computación, Instituto de Ingeniería, 04510 México D.F., Mexico, JMorenoP@ii.unam.mx 4th-8th September
More informationOutput Input Stability and Minimum-Phase Nonlinear Systems
422 IEEE TRANSACTIONS ON AUTOMATIC CONTROL, VOL. 47, NO. 3, MARCH 2002 Output Input Stability and Minimum-Phase Nonlinear Systems Daniel Liberzon, Member, IEEE, A. Stephen Morse, Fellow, IEEE, and Eduardo
More informationA Generalization of Barbalat s Lemma with Applications to Robust Model Predictive Control
A Generalization of Barbalat s Lemma with Applications to Robust Model Predictive Control Fernando A. C. C. Fontes 1 and Lalo Magni 2 1 Officina Mathematica, Departamento de Matemática para a Ciência e
More informationA. LEVANT School of Mathematical Sciences, Tel-Aviv University, Israel,
Chapter 1 INTRODUCTION TO HIGH-ORDER SLIDING MODES A. LEVANT School of Mathematical Sciences, Tel-Aviv University, Israel, 2002-2003 1.1 Introduction One of the most important control problems is control
More informationThe Rationale for Second Level Adaptation
The Rationale for Second Level Adaptation Kumpati S. Narendra, Yu Wang and Wei Chen Center for Systems Science, Yale University arxiv:1510.04989v1 [cs.sy] 16 Oct 2015 Abstract Recently, a new approach
More informationCodes for Partially Stuck-at Memory Cells
1 Codes for Partially Stuck-at Memory Cells Antonia Wachter-Zeh and Eitan Yaakobi Department of Computer Science Technion Israel Institute of Technology, Haifa, Israel Email: {antonia, yaakobi@cs.technion.ac.il
More informationOutput Regulation of Uncertain Nonlinear Systems with Nonlinear Exosystems
Output Regulation of Uncertain Nonlinear Systems with Nonlinear Exosystems Zhengtao Ding Manchester School of Engineering, University of Manchester Oxford Road, Manchester M3 9PL, United Kingdom zhengtaoding@manacuk
More informationRobust Output Feedback Stabilization of a Class of Nonminimum Phase Nonlinear Systems
Proceedings of the 26 American Control Conference Minneapolis, Minnesota, USA, June 14-16, 26 FrB3.2 Robust Output Feedback Stabilization of a Class of Nonminimum Phase Nonlinear Systems Bo Xie and Bin
More informationScientific Computing: An Introductory Survey
Scientific Computing: An Introductory Survey Chapter 5 Nonlinear Equations Prof. Michael T. Heath Department of Computer Science University of Illinois at Urbana-Champaign Copyright c 2002. Reproduction
More informationTrajectory planning and feedforward design for electromechanical motion systems version 2
2 Trajectory planning and feedforward design for electromechanical motion systems version 2 Report nr. DCT 2003-8 Paul Lambrechts Email: P.F.Lambrechts@tue.nl April, 2003 Abstract This report considers
More informationHigher order sliding mode control based on adaptive first order sliding mode controller
Preprints of the 19th World Congress The International Federation of Automatic Control Cape Town, South Africa. August 4-9, 14 Higher order sliding mode control based on adaptive first order sliding mode
More informationStability Analysis Through the Direct Method of Lyapunov in the Oscillation of a Synchronous Machine
Modern Applied Science; Vol. 12, No. 7; 2018 ISSN 1913-1844 E-ISSN 1913-1852 Published by Canadian Center of Science and Education Stability Analysis Through the Direct Method of Lyapunov in the Oscillation
More informationAlexander Scheinker Miroslav Krstić. Model-Free Stabilization by Extremum Seeking
Alexander Scheinker Miroslav Krstić Model-Free Stabilization by Extremum Seeking 123 Preface Originating in 1922, in its 95-year history, extremum seeking has served as a tool for model-free real-time
More informationRobust multivariable pid design via iterative lmi
Robust multivariable pid design via iterative lmi ERNESTO GRANADO, WILLIAM COLMENARES, OMAR PÉREZ Universidad Simón Bolívar, Departamento de Procesos y Sistemas. Caracas, Venezuela. e- mail: granado, williamc,
More information1 The Observability Canonical Form
NONLINEAR OBSERVERS AND SEPARATION PRINCIPLE 1 The Observability Canonical Form In this Chapter we discuss the design of observers for nonlinear systems modelled by equations of the form ẋ = f(x, u) (1)
More informationAn Approach to Constructing Good Two-level Orthogonal Factorial Designs with Large Run Sizes
An Approach to Constructing Good Two-level Orthogonal Factorial Designs with Large Run Sizes by Chenlu Shi B.Sc. (Hons.), St. Francis Xavier University, 013 Project Submitted in Partial Fulfillment of
More informationAutomatic Control II Computer exercise 3. LQG Design
Uppsala University Information Technology Systems and Control HN,FS,KN 2000-10 Last revised by HR August 16, 2017 Automatic Control II Computer exercise 3 LQG Design Preparations: Read Chapters 5 and 9
More informationON OPTIMAL ESTIMATION PROBLEMS FOR NONLINEAR SYSTEMS AND THEIR APPROXIMATE SOLUTION. A. Alessandri C. Cervellera A.F. Grassia M.
ON OPTIMAL ESTIMATION PROBLEMS FOR NONLINEAR SYSTEMS AND THEIR APPROXIMATE SOLUTION A Alessandri C Cervellera AF Grassia M Sanguineti ISSIA-CNR National Research Council of Italy Via De Marini 6, 16149
More informationAdaptive Dynamic Inversion Control of a Linear Scalar Plant with Constrained Control Inputs
5 American Control Conference June 8-, 5. Portland, OR, USA ThA. Adaptive Dynamic Inversion Control of a Linear Scalar Plant with Constrained Control Inputs Monish D. Tandale and John Valasek Abstract
More informationOptimal control and estimation
Automatic Control 2 Optimal control and estimation Prof. Alberto Bemporad University of Trento Academic year 2010-2011 Prof. Alberto Bemporad (University of Trento) Automatic Control 2 Academic year 2010-2011
More informationChapter III. Stability of Linear Systems
1 Chapter III Stability of Linear Systems 1. Stability and state transition matrix 2. Time-varying (non-autonomous) systems 3. Time-invariant systems 1 STABILITY AND STATE TRANSITION MATRIX 2 In this chapter,
More informationEvent-Triggered Decentralized Dynamic Output Feedback Control for LTI Systems
Event-Triggered Decentralized Dynamic Output Feedback Control for LTI Systems Pavankumar Tallapragada Nikhil Chopra Department of Mechanical Engineering, University of Maryland, College Park, 2742 MD,
More informationUNIVERSITY OF CALIFORNIA. Los Angeles. Distributed Model Predictive Control of Nonlinear. and Two-Time-Scale Process Networks
UNIVERSITY OF CALIFORNIA Los Angeles Distributed Model Predictive Control of Nonlinear and Two-Time-Scale Process Networks A dissertation submitted in partial satisfaction of the requirements for the degree
More informationROBUST STABLE NONLINEAR CONTROL AND DESIGN OF A CSTR IN A LARGE OPERATING RANGE. Johannes Gerhard, Martin Mönnigmann, Wolfgang Marquardt
ROBUST STABLE NONLINEAR CONTROL AND DESIGN OF A CSTR IN A LARGE OPERATING RANGE Johannes Gerhard, Martin Mönnigmann, Wolfgang Marquardt Lehrstuhl für Prozesstechnik, RWTH Aachen Turmstr. 46, D-5264 Aachen,
More informationCHBE320 LECTURE XI CONTROLLER DESIGN AND PID CONTOLLER TUNING. Professor Dae Ryook Yang
CHBE320 LECTURE XI CONTROLLER DESIGN AND PID CONTOLLER TUNING Professor Dae Ryook Yang Spring 2018 Dept. of Chemical and Biological Engineering 11-1 Road Map of the Lecture XI Controller Design and PID
More informationContents. 1 State-Space Linear Systems 5. 2 Linearization Causality, Time Invariance, and Linearity 31
Contents Preamble xiii Linear Systems I Basic Concepts 1 I System Representation 3 1 State-Space Linear Systems 5 1.1 State-Space Linear Systems 5 1.2 Block Diagrams 7 1.3 Exercises 11 2 Linearization
More informationFinite-Time Converging Jump Observer for Switched Linear Systems with Unknown Inputs
Finite-Time Converging Jump Observer for Switched Linear Systems with Unknown Inputs F.J. Bejarano a, A. Pisano b, E. Usai b a National Autonomous University of Mexico, Engineering Faculty, Division of
More informationMOST control systems are designed under the assumption
2076 IEEE TRANSACTIONS ON AUTOMATIC CONTROL, VOL. 53, NO. 9, OCTOBER 2008 Lyapunov-Based Model Predictive Control of Nonlinear Systems Subject to Data Losses David Muñoz de la Peña and Panagiotis D. Christofides
More informationAdaptive fuzzy observer and robust controller for a 2-DOF robot arm Sangeetha Bindiganavile Nagesh
Adaptive fuzzy observer and robust controller for a 2-DOF robot arm Delft Center for Systems and Control Adaptive fuzzy observer and robust controller for a 2-DOF robot arm For the degree of Master of
More informationSTABILIZATION FOR A CLASS OF UNCERTAIN MULTI-TIME DELAYS SYSTEM USING SLIDING MODE CONTROLLER. Received April 2010; revised August 2010
International Journal of Innovative Computing, Information and Control ICIC International c 2011 ISSN 1349-4198 Volume 7, Number 7(B), July 2011 pp. 4195 4205 STABILIZATION FOR A CLASS OF UNCERTAIN MULTI-TIME
More informationFault Estimation for Single Output Nonlinear Systems Using an Adaptive Sliding Mode Observer
Proceedings of the 7th World Congress The International Federation of Automatic Control Seoul, Korea, July 6-, 28 Fault Estimation for Single Output Nonlinear Systems Using an Adaptive Sliding Mode Observer
More informationA fuzzy logic controller with fuzzy scaling factor calculator applied to a nonlinear chemical process
Rev. Téc. Ing. Univ. Zulia. Vol. 26, Nº 3, 53-67, 2003 A fuzzy logic controller with fuzzy scaling factor calculator applied to a nonlinear chemical process Alessandro Anzalone, Edinzo Iglesias, Yohn García
More informationComputation of CPA Lyapunov functions
Computation of CPA Lyapunov functions (CPA = continuous and piecewise affine) Sigurdur Hafstein Reykjavik University, Iceland 17. July 2013 Workshop on Algorithms for Dynamical Systems and Lyapunov Functions
More informationCONTROL SYSTEMS, ROBOTICS AND AUTOMATION - Vol. XIII - Nonlinear Observers - A. J. Krener
NONLINEAR OBSERVERS A. J. Krener University of California, Davis, CA, USA Keywords: nonlinear observer, state estimation, nonlinear filtering, observability, high gain observers, minimum energy estimation,
More informationA Model-Free Control System Based on the Sliding Mode Control Method with Applications to Multi-Input-Multi-Output Systems
Proceedings of the 4 th International Conference of Control, Dynamic Systems, and Robotics (CDSR'17) Toronto, Canada August 21 23, 2017 Paper No. 119 DOI: 10.11159/cdsr17.119 A Model-Free Control System
More information1 Lyapunov theory of stability
M.Kawski, APM 581 Diff Equns Intro to Lyapunov theory. November 15, 29 1 1 Lyapunov theory of stability Introduction. Lyapunov s second (or direct) method provides tools for studying (asymptotic) stability
More informationA Robust Extended State Observer for the Estimation of Concentration and Kinetics in a CSTR
Instituto Tecnológico y de Estudios Superiores de Occidente Repositorio Institucional del ITESO rei.iteso.mx Departamento de Matemáticas y Física DMAF - Artículos y ponencias con arbitrae 05- A Robust
More informationMulti-Robotic Systems
CHAPTER 9 Multi-Robotic Systems The topic of multi-robotic systems is quite popular now. It is believed that such systems can have the following benefits: Improved performance ( winning by numbers ) Distributed
More informationDelay-Dependent Stability Criteria for Linear Systems with Multiple Time Delays
Delay-Dependent Stability Criteria for Linear Systems with Multiple Time Delays Yong He, Min Wu, Jin-Hua She Abstract This paper deals with the problem of the delay-dependent stability of linear systems
More informationLyapunov Stability Analysis of a Twisting Based Control Algorithm for Systems with Unmatched Perturbations
5th IEEE Conference on Decision and Control and European Control Conference (CDC-ECC) Orlando, FL, USA, December -5, Lyapunov Stability Analysis of a Twisting Based Control Algorithm for Systems with Unmatched
More informationComputing Optimized Nonlinear Sliding Surfaces
Computing Optimized Nonlinear Sliding Surfaces Azad Ghaffari and Mohammad Javad Yazdanpanah Abstract In this paper, we have concentrated on real systems consisting of structural uncertainties and affected
More informationAN INVESTIGATION OF TECHNIQUES FOR NONLINEAR STATE OBSERVATION
AN INVESTIGATION OF TECHNIQUES FOR NONLINEAR STATE OBSERVATION DEAN CHRISTIAAN TAIT MCBRIDE A dissertation submitted to the Faculty of Engineering and the Built Environment, University of the Witwatersrand,
More informationDESIGN OF OBSERVERS FOR SYSTEMS WITH SLOW AND FAST MODES
DESIGN OF OBSERVERS FOR SYSTEMS WITH SLOW AND FAST MODES by HEONJONG YOO A thesis submitted to the Graduate School-New Brunswick Rutgers, The State University of New Jersey In partial fulfillment of the
More informationConstrained State Estimation Using the Unscented Kalman Filter
16th Mediterranean Conference on Control and Automation Congress Centre, Ajaccio, France June 25-27, 28 Constrained State Estimation Using the Unscented Kalman Filter Rambabu Kandepu, Lars Imsland and
More informationOverview of Control System Design
Overview of Control System Design General Requirements 1. Safety. It is imperative that industrial plants operate safely so as to promote the well-being of people and equipment within the plant and in
More informationOn Sontag s Formula for the Input-to-State Practical Stabilization of Retarded Control-Affine Systems
On Sontag s Formula for the Input-to-State Practical Stabilization of Retarded Control-Affine Systems arxiv:1206.4240v1 [math.oc] 19 Jun 2012 P. Pepe Abstract In this paper input-to-state practically stabilizing
More informationLecture Note 7: Switching Stabilization via Control-Lyapunov Function
ECE7850: Hybrid Systems:Theory and Applications Lecture Note 7: Switching Stabilization via Control-Lyapunov Function Wei Zhang Assistant Professor Department of Electrical and Computer Engineering Ohio
More informationA Soft Sensor for Biomass in a Batch Process with Delayed Measurements
Instituto Tecnológico y de Estudios Superiores de Occidente Repositorio Institucional del ITESO rei.iteso.mx Departamento de Matemáticas y Física DMAF - Artículos y ponencias con arbitraje - A Soft Sensor
More informationDESIGN OF PROBABILISTIC OBSERVERS FOR MASS-BALANCE BASED BIOPROCESS MODELS. Benoît Chachuat and Olivier Bernard
DESIGN OF PROBABILISTIC OBSERVERS FOR MASS-BALANCE BASED BIOPROCESS MODELS Benoît Chachuat and Olivier Bernard INRIA Comore, BP 93, 692 Sophia-Antipolis, France fax: +33 492 387 858 email: Olivier.Bernard@inria.fr
More informationOutput Adaptive Model Reference Control of Linear Continuous State-Delay Plant
Output Adaptive Model Reference Control of Linear Continuous State-Delay Plant Boris M. Mirkin and Per-Olof Gutman Faculty of Agricultural Engineering Technion Israel Institute of Technology Haifa 3, Israel
More informationState observers for invariant dynamics on a Lie group
State observers for invariant dynamics on a Lie group C. Lageman, R. Mahony, J. Trumpf 1 Introduction This paper concerns the design of full state observers for state space systems where the state is evolving
More information1 Kalman Filter Introduction
1 Kalman Filter Introduction You should first read Chapter 1 of Stochastic models, estimation, and control: Volume 1 by Peter S. Maybec (available here). 1.1 Explanation of Equations (1-3) and (1-4) Equation
More informationMathematical formulation of restrictions for the design of low voltage bandgap references
Rev Fac Ing Univ Antioquia N 60 pp 165-170 Septiembre, 2011 Mathematical formulation of restrictions for the design of low voltage bandgap references Formulación matemática de restricciones para el diseño
More informationBook review for Stability and Control of Dynamical Systems with Applications: A tribute to Anthony M. Michel
To appear in International Journal of Hybrid Systems c 2004 Nonpareil Publishers Book review for Stability and Control of Dynamical Systems with Applications: A tribute to Anthony M. Michel João Hespanha
More informationAdvanced Adaptive Control for Unintended System Behavior
Advanced Adaptive Control for Unintended System Behavior Dr. Chengyu Cao Mechanical Engineering University of Connecticut ccao@engr.uconn.edu jtang@engr.uconn.edu Outline Part I: Challenges: Unintended
More informationA ROBUST ITERATIVE LEARNING OBSERVER BASED FAULT DIAGNOSIS OF TIME DELAY NONLINEAR SYSTEMS
Copyright IFAC 15th Triennial World Congress, Barcelona, Spain A ROBUST ITERATIVE LEARNING OBSERVER BASED FAULT DIAGNOSIS OF TIME DELAY NONLINEAR SYSTEMS Wen Chen, Mehrdad Saif 1 School of Engineering
More informationStability theory is a fundamental topic in mathematics and engineering, that include every
Stability Theory Stability theory is a fundamental topic in mathematics and engineering, that include every branches of control theory. For a control system, the least requirement is that the system is
More informationEstimation for state space models: quasi-likelihood and asymptotic quasi-likelihood approaches
University of Wollongong Research Online University of Wollongong Thesis Collection 1954-2016 University of Wollongong Thesis Collections 2008 Estimation for state space models: quasi-likelihood and asymptotic
More informationA NONLINEAR TRANSFORMATION APPROACH TO GLOBAL ADAPTIVE OUTPUT FEEDBACK CONTROL OF 3RD-ORDER UNCERTAIN NONLINEAR SYSTEMS
Copyright 00 IFAC 15th Triennial World Congress, Barcelona, Spain A NONLINEAR TRANSFORMATION APPROACH TO GLOBAL ADAPTIVE OUTPUT FEEDBACK CONTROL OF RD-ORDER UNCERTAIN NONLINEAR SYSTEMS Choon-Ki Ahn, Beom-Soo
More informationState Estimation using Moving Horizon Estimation and Particle Filtering
State Estimation using Moving Horizon Estimation and Particle Filtering James B. Rawlings Department of Chemical and Biological Engineering UW Math Probability Seminar Spring 2009 Rawlings MHE & PF 1 /
More informationOutline. Scientific Computing: An Introductory Survey. Nonlinear Equations. Nonlinear Equations. Examples: Nonlinear Equations
Methods for Systems of Methods for Systems of Outline Scientific Computing: An Introductory Survey Chapter 5 1 Prof. Michael T. Heath Department of Computer Science University of Illinois at Urbana-Champaign
More informationASIGNIFICANT research effort has been devoted to the. Optimal State Estimation for Stochastic Systems: An Information Theoretic Approach
IEEE TRANSACTIONS ON AUTOMATIC CONTROL, VOL 42, NO 6, JUNE 1997 771 Optimal State Estimation for Stochastic Systems: An Information Theoretic Approach Xiangbo Feng, Kenneth A Loparo, Senior Member, IEEE,
More informationNonlinear Observer Design for Dynamic Positioning
Author s Name, Company Title of the Paper DYNAMIC POSITIONING CONFERENCE November 15-16, 2005 Control Systems I J.G. Snijders, J.W. van der Woude Delft University of Technology (The Netherlands) J. Westhuis
More informationOutput Feedback and State Feedback. EL2620 Nonlinear Control. Nonlinear Observers. Nonlinear Controllers. ẋ = f(x,u), y = h(x)
Output Feedback and State Feedback EL2620 Nonlinear Control Lecture 10 Exact feedback linearization Input-output linearization Lyapunov-based control design methods ẋ = f(x,u) y = h(x) Output feedback:
More informationRobust Stabilization of Non-Minimum Phase Nonlinear Systems Using Extended High Gain Observers
28 American Control Conference Westin Seattle Hotel, Seattle, Washington, USA June 11-13, 28 WeC15.1 Robust Stabilization of Non-Minimum Phase Nonlinear Systems Using Extended High Gain Observers Shahid
More information