Output-feedback Dynamic Surface Control for a Class of Nonlinear Non-minimum Phase Systems

Similar documents
Output Regulation of Uncertain Nonlinear Systems with Nonlinear Exosystems

Robust Stabilization of Non-Minimum Phase Nonlinear Systems Using Extended High Gain Observers

UDE-based Dynamic Surface Control for Strict-feedback Systems with Mismatched Disturbances

A NONLINEAR TRANSFORMATION APPROACH TO GLOBAL ADAPTIVE OUTPUT FEEDBACK CONTROL OF 3RD-ORDER UNCERTAIN NONLINEAR SYSTEMS

IMECE NEW APPROACH OF TRACKING CONTROL FOR A CLASS OF NON-MINIMUM PHASE LINEAR SYSTEMS

MULTI-AGENT TRACKING OF A HIGH-DIMENSIONAL ACTIVE LEADER WITH SWITCHING TOPOLOGY

THE DESIGN OF ACTIVE CONTROLLER FOR THE OUTPUT REGULATION OF LIU-LIU-LIU-SU CHAOTIC SYSTEM

Design of Observer-based Adaptive Controller for Nonlinear Systems with Unmodeled Dynamics and Actuator Dead-zone

Dynamic backstepping control for pure-feedback nonlinear systems

arxiv: v2 [nlin.ao] 5 Jan 2012

OUTPUT REGULATION OF THE SIMPLIFIED LORENZ CHAOTIC SYSTEM

OUTPUT REGULATION OF RÖSSLER PROTOTYPE-4 CHAOTIC SYSTEM BY STATE FEEDBACK CONTROL

OVER THE past 20 years, the control of mobile robots has

Research Article Convex Polyhedron Method to Stability of Continuous Systems with Two Additive Time-Varying Delay Components

ACTIVE CONTROLLER DESIGN FOR THE OUTPUT REGULATION OF THE WANG-CHEN-YUAN SYSTEM

Output Regulation of the Arneodo Chaotic System

High-Gain Observers in Nonlinear Feedback Control. Lecture # 3 Regulation

Output Regulation of the Tigan System

Disturbance Attenuation for a Class of Nonlinear Systems by Output Feedback

Adaptive Predictive Observer Design for Class of Uncertain Nonlinear Systems with Bounded Disturbance

Tracking Control of a Class of Differential Inclusion Systems via Sliding Mode Technique

THE nonholonomic systems, that is Lagrange systems

An homotopy method for exact tracking of nonlinear nonminimum phase systems: the example of the spherical inverted pendulum

Adaptive Nonlinear Control A Tutorial. Miroslav Krstić

Nonlinear Tracking Control of Underactuated Surface Vessel

IN THIS paper, we study the problem of asymptotic stabilization

Dynamic Integral Sliding Mode Control of Nonlinear SISO Systems with States Dependent Matched and Mismatched Uncertainties

Robust Output Feedback Stabilization of a Class of Nonminimum Phase Nonlinear Systems

Semi-global Robust Output Regulation for a Class of Nonlinear Systems Using Output Feedback

State and Parameter Estimation Based on Filtered Transformation for a Class of Second-Order Systems

EL 625 Lecture 10. Pole Placement and Observer Design. ẋ = Ax (1)

On Convergence of Nonlinear Active Disturbance Rejection for SISO Systems

Neural adaptive control for uncertain nonlinear system with input saturation Gao, Shigen; Dong, Hairong; Ning, Bin; Chen, Lei

Adaptive Robust Control for Servo Mechanisms With Partially Unknown States via Dynamic Surface Control Approach

Anti-synchronization of a new hyperchaotic system via small-gain theorem

UDE-based Robust Control for Nonlinear Systems with Mismatched Uncertainties and Input Saturation

Lyapunov Stability of Linear Predictor Feedback for Distributed Input Delays

Multi-Robotic Systems

High-Gain Observers in Nonlinear Feedback Control. Lecture # 2 Separation Principle

Min-Max Output Integral Sliding Mode Control for Multiplant Linear Uncertain Systems

Multiple-mode switched observer-based unknown input estimation for a class of switched systems

IEEE TRANSACTIONS ON AUTOMATIC CONTROL, VOL. 58, NO. 5, MAY invertible, that is (1) In this way, on, and on, system (3) becomes

ADAPTIVE FEEDBACK LINEARIZING CONTROL OF CHUA S CIRCUIT

Global output regulation through singularities

High-Gain Observers in Nonlinear Feedback Control

Dynamical analysis and circuit simulation of a new three-dimensional chaotic system

Further results on global stabilization of the PVTOL aircraft

Robust Gain Scheduling Synchronization Method for Quadratic Chaotic Systems With Channel Time Delay Yu Liang and Horacio J.

Adaptive backstepping for trajectory tracking of nonlinearly parameterized class of nonlinear systems

Output Feedback Control for a Class of Nonlinear Systems

Robust Tracking Control of Uncertain MIMO Nonlinear Systems with Application to UAVs

Global Practical Output Regulation of a Class of Nonlinear Systems by Output Feedback

arxiv: v2 [math.oc] 14 Dec 2015

Control design using Jordan controllable canonical form

Robust Output Feedback Control for a Class of Nonlinear Systems with Input Unmodeled Dynamics

Further Results on Adaptive Robust Periodic Regulation

Set-based adaptive estimation for a class of nonlinear systems with time-varying parameters

A Recurrent Neural Network for Solving Sylvester Equation With Time-Varying Coefficients

Output Regulation of Non-Minimum Phase Nonlinear Systems Using Extended High-Gain Observers

Adaptive Dynamic Inversion Control of a Linear Scalar Plant with Constrained Control Inputs

NONLINEAR PATH CONTROL FOR A DIFFERENTIAL DRIVE MOBILE ROBOT

Research Article Stabilization Analysis and Synthesis of Discrete-Time Descriptor Markov Jump Systems with Partially Unknown Transition Probabilities

IN recent years, controller design for systems having complex

The Rationale for Second Level Adaptation

Nonlinear Controller Design of the Inverted Pendulum System based on Extended State Observer Limin Du, Fucheng Cao

A ROBUST ITERATIVE LEARNING OBSERVER BASED FAULT DIAGNOSIS OF TIME DELAY NONLINEAR SYSTEMS

Robust Stabilization of Jet Engine Compressor in the Presence of Noise and Unmeasured States

Robust Internal Model Control for Impulse Elimination of Singular Systems

Aerospace Science and Technology

RECENTLY, the study of cooperative control of multiagent

Event-Triggered Decentralized Dynamic Output Feedback Control for LTI Systems

Finite-time hybrid synchronization of time-delay hyperchaotic Lorenz system

Adaptive Nonlinear Control Allocation of. Non-minimum Phase Uncertain Systems

A Discrete Robust Adaptive Iterative Learning Control for a Class of Nonlinear Systems with Unknown Control Direction

Time-delay feedback control in a delayed dynamical chaos system and its applications

Modeling and Analysis of Dynamic Systems

Discontinuous Backstepping for Stabilization of Nonholonomic Mobile Robots

Output Feedback Stabilization with Prescribed Performance for Uncertain Nonlinear Systems in Canonical Form

CONTROL DESIGN FOR SET POINT TRACKING

NEURAL NETWORKS (NNs) play an important role in

Generalized projective synchronization of a class of chaotic (hyperchaotic) systems with uncertain parameters

Adaptive and Robust Controls of Uncertain Systems With Nonlinear Parameterization

Trajectory tracking & Path-following control

Problem 1: Ship Path-Following Control System (35%)

An Adaptive LQG Combined With the MRAS Based LFFC for Motion Control Systems

Terminal Sliding Mode Control for Cyber Physical System Based on Filtering Backstepping

CONTROLLING CHAOTIC DYNAMICS USING BACKSTEPPING DESIGN WITH APPLICATION TO THE LORENZ SYSTEM AND CHUA S CIRCUIT

An asymptotic ratio characterization of input-to-state stability

CHATTERING-FREE SMC WITH UNIDIRECTIONAL AUXILIARY SURFACES FOR NONLINEAR SYSTEM WITH STATE CONSTRAINTS. Jian Fu, Qing-Xian Wu and Ze-Hui Mao

Chaos Suppression in Forced Van Der Pol Oscillator

Design and Stability Analysis of Single-Input Fuzzy Logic Controller

H 2 Adaptive Control. Tansel Yucelen, Anthony J. Calise, and Rajeev Chandramohan. WeA03.4

SLIDING MODE FAULT TOLERANT CONTROL WITH PRESCRIBED PERFORMANCE. Jicheng Gao, Qikun Shen, Pengfei Yang and Jianye Gong

Prediction-based adaptive control of a class of discrete-time nonlinear systems with nonlinear growth rate

Converse Lyapunov theorem and Input-to-State Stability

On the Convergence of Extended State Observer for Nonlinear Systems with Uncertainty

FINITE HORIZON ROBUST MODEL PREDICTIVE CONTROL USING LINEAR MATRIX INEQUALITIES. Danlei Chu, Tongwen Chen, Horacio J. Marquez

STABILIZABILITY AND SOLVABILITY OF DELAY DIFFERENTIAL EQUATIONS USING BACKSTEPPING METHOD. Fadhel S. Fadhel 1, Saja F. Noaman 2

Is Monopoli s Model Reference Adaptive Controller Correct?

Passivity-based Stabilization of Non-Compact Sets

Transcription:

96 IEEE/CAA JOURNAL OF AUTOMATICA SINICA, VOL. 3, NO., JANUARY 06 Output-feedback Dynamic Surface Control for a Class of Nonlinear Non-minimum Phase Systems Shanwei Su Abstract In this paper, an output-feedback tracking controller is proposed for a class of nonlinear non-minimum phase systems. To keep the unstable internal dynamics bounded, the method of output redefinition is applied to let the stability of the internal dynamics depend on that of redefined output, thus we only need to consider the new external dynamics rather than internal dynamics in the process of designing control law. To overcome the explosion of complexity problem in traditional backstepping design, the dynamic surface control DSC method is firstly used to deal with the problem of tracking control for the nonlinear non-minimum phase systems. The proposed outputfeedback DSC controller not only forces the system output to asymptotically track the desired trajectory, but also drives the unstable internal dynamics to follow its corresponding bounded and causal ideal internal dynamics, which is solved via stable system center method. Simulation results illustrate the validity of the proposed output-feedback DSC controller. Index Terms Non-minimum phase system, output-feedback, trajectory tracking, internal dynamics, dynamic surface control DSC. I. INTRODUCTION It is well known that nonlinear systems with asymptotically unstable zero dynamics or internal dynamics are called non-minimum phase [ ]. This feature cannot be removed by feedback, and restricts the straightforward application of the powerful nonlinear control techniques such as feedback linearization, sliding mode control and backstepping method [3 4], which work well in minimum phase systems. Under this circumstance, the control problem of nonlinear nonminimum phase systems is more challenging and has been payed more attention by the control community. The existing research work concerning the nonlinear nonminimum phase systems can be divided into two main branches: stabilization control and tracking control. Firstly we give a brief description on stabilization control for nonlinear non-minimum phase systems. Based on backstepping control Manuscript received January 5, 05; accepted July 7, 05. This work was supported by National Natural Science Foundation of China 640303 and the Aero-Science Foundation of China 05ZA5009. Recommended by Associate Editor Shengwei Mei. Citation: Shanwei Su. Output-feedback dynamic surface control for a class of nonlinear non-minimum phase systems. IEEE/CAA Journal of Automatica Sinica, 06, 3: 96 04 Shamwei Su is with the Research Institute of Unmanned Aerial Vehicle, Beihang University, Beijing 009, China e-mail: sushanwei@aliyun.com. and inverse design with the combination of neural network [5], implemented the state-feedback adaptive stabilization for a class of non-affine single-input single-output SISO nonminimum phase systems [6]. achieved stabilization control for a class of nonlinear non-minimum phase systems in general output-feedback form via standard backstepping control and small-gain technique. In [7 0], different stabilization methods for non-minimum phase systems can be found, such as robust observer, neural network, and high-gain observer. Though the backstepping method cannot be directly used in non-minimum phase systems as described in [4], the aforementioned work properly introduced the backstepping into the stabilization control for those non-minimum phase systems under proper assumptions, and attained control aims. The common feature of these referred papers is that the outputs and the unstable internal dynamics are all stabilized to zero. As we all know, stabilization control is the basis of output tracking control, and output tracking can transformed into stabilization control problem of tracking errors [3], so the aforesaid stabilization control methods can lead to the creation of output tracking control methods for nonlinear non-minimum phase systems. Whereas the work such as [3, ] focus on another ambitious problem - trajectory tracking control of nonlinear nonminimum phase systems. When we design tracking control laws for minimum phase systems, the control aim is to let the system outputs to follow the desired output signals, and the internal dynamics are generally disregarded, because it will get stable when the external dynamics attains stability. However, this case is not applicable to non-minimum phase systems. For the tracking control of nonlinear non-minimum phase systems, the controller is designed to meet the following two demands: the output tracking errors asymptotically converge to zero; the unstable internal dynamics is rendered acceptable, that is, stabilized to zero [3 4] or kept bounded [5]. The output tracking problem of a nonlinear non-minimum phase VTOL aircraft was solved in [3] by a Lyapunov-based technique and a minimum-norm strategy. However, the internal dynamics which stands for the actual roll attitude of aircraft was directly stabilized to zero. In fact, this method is unfeasible for aircraft since it is impossible for its roll attitude to keep unchanged when it performs trajectory maneuver. Reference [5] pointed

SU: OUTPUT-FEEDBACK DYNAMIC SURFACE CONTROL FOR A CLASS OF NONLINEAR NON-MINIMUM PHASE SYSTEMS 97 out that the proper method for presenting acceptable internal dynamics lies in finding a bounded solution for the unstable internal dynamics to follow, rather than directly stabilizing the internal dynamics to zero. The bound solution to the internal dynamics is originally called the ideal internal dynamics IID in [ ] proposed noncausal stable inversion NSI method to construct the bounded IID for non-minimum phase systems. To obtain the causal IID for the unstable internal dynamics [3], originated the stable system center SSC method, which was used to solve the IID of non-minimum phase VTOL aircraft in [6]. The output regulation OR method in [7] can be used to find IID of non-minimum phase systems by solving partial differential algebraic equations. As a survey on the control of nonlinear non-minimum phase systems [8], discussed the features of three methods for the IID solution as follows: NSI is an iterative solution method, the desired trajectories and any of their changes must be exactly known in advance, the offline pre-computing procedure is conducted backward in time, thus it can only get numerical and noncausal solutions, so it is of limited practical use; Compared with NSI approach, SSC method does not necessarily require the future information of the desired trajectories which must be generated by a exosystem, the online solving procedure for bounded IID is performed forward in time, so the obtained numerical solutions are causal; OR method is applied to tackle the system with linear internal dynamics, and can provide accurate and analytical solutions. Until now, there is few work about the output tracking control for nonlinear non-minimum phase systems in outputfeedback form. This paper aims to solve this problem, and the contributions of this work can be summarized as follows: To keep the unstable internal dynamics bounded, the method of output redefinition is introduced, thus the stability of the internal dynamics depends on that of the newly defined output. In the process of designing the control law, we only care about the external dynamics which includes the newly defined output, and disregard the internal dynamics, because it will get stable along with the stability of the external dynamics. To overcome the explosion of complexity problem in traditional backstepping design, the dynamic surface control DSC [9] method is firstly used to deal with the problem of tracking control for the nonlinear non-minimum phase systems in output-feedback form. 3 Benefiting from the bounded IID solved via SSC method, the paper realizes the casual output tracking for a class of nonlinear non-minimum phase systems. The paper is organized as follows. In Section II, the class of controlled nonlinear non-minimum phase system in output-feedback form is introduced, and the control purpose is formulated. In Section III, the methods of observer-based output redefinition and the solution of IID are presented. In Section IV, the output feedback DSC design procedure is provided. Section V gives the stability analysis. In Section VI, a simulation example is given to show the effectiveness of the proposed design method. Section VII draws the conclusions. II. PROBLEM FORMULATION In the paper, we consider a class of nonlinear non-minimum phase systems in the following output-feedback form: η = A η η + B η x, ẋ = x + ϕ y,. ẋ n = x n + ϕ n y, ẋ n = E η η + ϕ n y + βyu, y = x, where η R m is internal dynamics; x,..., x n is the external dynamics; u is the control input; y is the output signal; Nonlinear items ϕ i y only depend on the output y, i =,..., n. For all y R, the function βy 0. We assume the matrix A η is non-hurwitz, so when x = 0, the corresponding zero dynamics η = A η η is not asymptotically stable, thus η is the unstable internal dynamics, system can be called non-minimum phase. Since that the internal dynamics η is unstable and only output y is measured, the control object is to design an output feedback controller so that the system output y can track the desired trajectory signal y d t, while the unstable internal dynamics η can follow its causal and bounded IID. III. OBSERVE-BASED OUTPUT REDEFINITION A. Observer Design Since only the output signal y in is available for measurement, a set of observers must be constructed to provide estimates of the unmeasured state variables η, x,..., x n. To proceed, rewrite system as in which ẋ = Ax + ϕy + bβyu, y = cx, x = [ η m, x,..., x n ] T, ϕy = [ 0 m, ϕ y,..., ϕ n y ] T, A η B η 0 m n A = 0 n m 0 n I n, E η 0 0 n b = [ 0 m 0 n ] T, c = [ 0 m 0 n ], 3 and 0, I respectively stand for zero matrix and identity matrix.

98 IEEE/CAA JOURNAL OF AUTOMATICA SINICA, VOL. 3, NO., JANUARY 06 By choosing vector k such that A 0 = A kc is a Hurwitz matrix, the following full-order observer is proposed for the purpose of the tracking control: ˆx = Aˆx + ky ŷ + ϕy + bβyu, ŷ = cˆx, 4 where the parameter k = k T η, k,..., k n T, and k η = k η,..., k ηm T. Subtracting 4 from, the observer error can be derived as x = A 0 x, 5 where x = x ˆx. Since A 0 is Hurwitz, the observer error x can exponentially converge to zero. B. Output Redefinition From 4, the observer equation of the internal dynamics can be written as ˆη = A η ˆη + B η ˆx + k η x. 6 To tackle the unstable internal dynamics ˆη, the new output is defined as Thus 6 shows x = ˆx + M ˆη, 7 ˆη = A η0ˆη + B η x + k η x, 8 where choosing M to let A η0 = A η B η M be Hurwitz. Since the observer error x 0, the stability of internal dynamics ˆη in 8 depends on the newly defined output x. After output redefinition, system 4 can be separated into the following two parts: new external dynamics x = ˆx + ϕ y + k x + M ˆη, ˆx = ˆx 3 + ϕ y + k x,. ˆx n = ˆx n + ϕ n y + k n x, ˆx n = E η ˆη + ϕ n y + k n x + βyu, 9 and the internal dynamics 8. Hereafter, we only design control law for the external dynamics x, ˆx,..., ˆx n in 9, rather than the internal dynamics ˆη, because it will get stable with the stability of the external dynamics. Remark. Note that the method of output redefinition originated from [] is slightly revised in this paper, and the new output is based on the output observer system, rather than the original system. C. Solution of IID It is necessary to know the desired value of new output x before we design the output tracking control law. In Equation 7, the desired value of output observer ˆx is y d, but what is the desired value of ˆη? In 6, A η is non-hurwitz, so the internal dynamics does not have stable numerical solution. However, this does not mean that a bounded solution cannot be found for such an unstable system [5]. In fact, under suitable assumptions, via NSI approach [] or the SSC method [3], a bounded solution can be obtained. As to 6, setting ˆx = y d, x = 0, we can get a bounded solution of internal dynamics η d = A η η d + B η y d, 0 where η d is the so-called IID. For the IID equation 0, we turn to the causal SSC method to solve η d. For convenience, 0 can be rewritten as η d = A η η d + θ d y d, where θ d y d = B η y d. We assume θ d can be generated by a known exosystem Its characteristic polynomial is ẇ = Sw, θ d = Cw. P λ = detλi S = λ k + p k λ k + + p λ + p 0. 3 Thus the causal IID η d can be solved by the following matrix differential equation η k d + c k η k d + + c η d + c 0 η d = P k θ k d + + P θ d + P 0 θ d, 4 where the parameters c k,..., c, c 0 depend on the desired eigenvalues, the matrix P k,..., P, P 0 R n r n r can be computed by the formula in [3]. IV. CONTROL LAW DESIGN In this section, we design control law for the external dynamics x, ˆx,..., ˆx n via dynamic surface method. Step. Let the first error surface of new output be defined as S = x x d, 5 where x d = y d + Mη d. The time derivative of S is Ṡ = x x d = ˆx + ϕ y + k x + M ˆη x d. 6 Then a virtual control signal is selected as ˆx d = l S ϕ y k x M ˆη + x d, 7 and the error between ˆx and ˆx d is defined as x d = ˆx ˆx d. 8

SU: OUTPUT-FEEDBACK DYNAMIC SURFACE CONTROL FOR A CLASS OF NONLINEAR NON-MINIMUM PHASE SYSTEMS 99 So 6 becomes Ṡ = l S + x d. 9 To avoid the explosion of terms in the process of computing ˆx d, we let ˆx d pass through a low-pass filter τ x d + x d = ˆx d, x d 0 = ˆx d 0, 0 where τ is a time constant. Thus x d is the filtered signal of ˆx d. Step i i n. Define the i-th error surface as then the time derivative of S i satisfies S i = ˆx i x id, Ṡ i = ˆx i x id = ˆx i+ + ϕ i y + k i x x id. Choose a virtual control signal ˆx i+d as ˆx i+d = l i S i ϕ i y k i x + x id, 3 and define the error between ˆx i+ and ˆx i+d as Thus we can get x i+d = ˆx i+ ˆx i+d. 4 Ṡ i = l i S i + x i+d. 5 Letting x i+d be the filtered signal of ˆx i+d, that is, τ i+ x i+d + x i+d = ˆx i+d, x i+d 0 = ˆx i+d 0, 6 where τ i+ is a positive time constant. Step n. Finally, define the n-th error surface as thus Ṡ n = ˆx n x nd S n = ˆx n x nd, 7 = E η ˆη + ϕ n y + k n x + βyu x nd. 8 The actual control signal is thus chosen as u = l ns n E η ˆη ϕ n y k n x + x nd. 9 βy It could be readily checked that Ṡ n = l n S n. 30 Remark. According to the conventional backstepping method [4, 0], the actual control signal shows u = α n E η ˆη + x n d, 3 βy where αn α n = l n S n S n d n S i k i x y α n ˆx + ϕ y y n j= m j= α n ˆx j ˆx j+ + ϕ y + k j x α n ˆη j n ˆηj j= α n x j+ x j d, 3 d S n = ˆx n α n x n d, 33 α = l S d k + Mk η S ϕ y MA η ˆη MB η ˆx, 34 S = x x d, 35 and S i, α i i =,..., n have similar forms as S n, α n. Obviously, compared with the backstepping method, the proposed DSC scheme is simpler, and is easier to be realized in the practical systems with the increase of the dimension m of internal dynamics and system relative order n. V. STABILITY ANALYSIS In this section, we will give the stability analysis for the proposed output-feedback DSC scheme. While the control law design procedure is simple, the stability analysis is relatively complicated due to the derivation of low-pass filter. Firstly, define the filter error as z i = ˆx id x id, i =,..., n. 36 Taking 6 into consideration, the time derivative of x id shows Thus x id = τ i ˆx id x id = τ i z i. 37 ż i = ˆx id x id = τ i z i + ˆx id. 38 By considering and 36, x id in 4 can be rewritten as x id = ˆx i ˆx id = ˆx i x id + x id ˆx id = S i z i. 39 So the error surface in 9 and 5 can be described as Ṡ i = l i S i + S i+ z i+, i =,..., n. 40 On the other side, from 5-7, -3, we can obtain ż i + τ i z i = ˆx id B i S i, S i, z i, x, x i d, x i d, x i d, i =,..., n, 4

00 IEEE/CAA JOURNAL OF AUTOMATICA SINICA, VOL. 3, NO., JANUARY 06 where B i, i =,..., n, are continuous positive functions. Thus z i ż i τ i z i + B i z i B i zi + α i τ i α i, i =,..., n, 4 where α i are positive constants. Secondly, define the tracking error of internal dynamics as η = ˆη η d. 43 According to 8, 0 and 5, we can get η = A η0 η + B η S + k η x. 44 By considering the aforementioned equations, the resulting closed-loop system can be expressed as x = A 0 x, 45 η = A η0 η + B η S + k η x, 46 Ṡ i = l i S i + S i+ z i+, i =,..., n, 47 Ṡ n = l n S n, 48 B z i ż i i zi + α i τ i α i, i =,..., n. 49 Next, we present the following theorem. Theorem. Consider the Lyapunov function candidate as where V = n V i, 50 i=0 V 0 = η T P η η + x T P 0 x, 5 V i = S i + d i z i+, i n, 5 V n = S n, 53 and P η = Pη T, P 0 = P0 T respectively stand for the symmetric positive solutions of P η A η0 + A T η0p η = I, P 0 A 0 + A T 0 P 0 = I, A η0 is defined by 8. For the given compact set Ω, if V 0 = n V i 0 R 0, 54 i=0 then there exist γ, r, l i i =,..., n, d i i =,..., n, and τ i i =,..., n to let all signals of the closed-loop system be bounded, the tracking errors can converge to some residual sets that can be made arbitrarily small by properly choosing certain design parameters. Proof. The time derivative of V 0 is V 0 = η T A T η0p η η + η T P η A η0 η + S B T η P η η + η T P η B η S + x k T η P η η + η T P η k η x + x T A T 0 P 0 x + x T P 0 A 0 x = η Tˆη + γs + γ ηt P T η B η B T η P η η + γ x + γ ηt P T η k η k T η P η η x T x λ max P η λ maxpη T B η Bη T P γλ min P η λ maxpη T k η kη T P η η T P η η γλ min P η + γs λ max P 0 γ x T P 0 x. 55 λ min P 0 The derivative of V i satisfies V i = S i l i S i + S i+ z i+ + d i z i+ ż i+ l i Si + Si + Si+ + d i d i z i+ B + d i+ i zi+ + d i α i+ τ i+ α i+ l i S i + d i S i+ d i τ i+ B i+ α i+ Define the compact set as Ω = z i+ + d i α i+. 56 { S,..., S n, z,..., z n, x, η d, x d,..., x n d, x d,..., x n d, x d,..., x n d : n n Si + zi + x T x n + ηd T η d + i= + i= i= i= i= x i d n n x i d + } x i d R 0, 57 where R 0 is a positive constant. Since B i+ defined by 4 is continuous, it has a maximum value in Ω, i.e., M i+. So the derivative of V i shows V i l i S i + d i S i+ d i τ i+ M i+ α i+ z i+ + d i α i+. 58 Finally, in view of 30, the derivative of V n yields So V n = S n Ṡ n = l n S n. 59

SU: OUTPUT-FEEDBACK DYNAMIC SURFACE CONTROL FOR A CLASS OF NONLINEAR NON-MINIMUM PHASE SYSTEMS 0 V λ max P η λ maxpη T B η Bη T P γλ min P η λ maxpη T k 0 k0 T P η ˆη T P η ˆη γλ min P η λ max P 0 γ x T P 0 x λ min P 0 + n n d i α i+ i= i= n M i+ τ i= i+ α i+ + l n Then let the design parameters λ max P η = r + λ maxp T η B η B T η P γλ min P η c i d i γ d i z i+ S i S n. 60 λ maxpη T k η kη T P η, 6 γλ min P η λ max P 0 = r + γ λ min P 0, 6 l i = r + d i + + γ, 63 τ i+ = r + M i+ α i+ +, 64 l n = r +, 65 where r is a positive constant, it follows that where V rv + M, 66 M = n d i α i. 67 i= So when V = R 0, we have V rr 0 + M. That is, if r is chosen such that r > M R 0, 68 we have V < 0 on V = R 0, which implies that if V 0 R 0, then V t R 0 for all t 0, i.e., V R 0 is an invariant set. Moreover, solving 66 yields 0 V M r + V 0 M e rt. 69 r Hence lim T 0 V t M r. 70 That is, by properly choosing γ, l i i =,..., n, d i i =,..., n, τ i i =,..., n to make M sufficiently small, r sufficiently large, the tracking error S i i =,..., n can converge to any arbitrary small residual set. Since the observer error x exponentially converges to zero, and the desired output signal y d and the IID η d are all bounded, so all the signals of the closed-loop system are uniformly bounded. VI. SIMULATION RESULTS We consider the following nonlinear non-minimum phase system: ẋ = x + sin y, ẋ = x 3 + y, ẋ 3 = η + 00y + u, η = η + x, y = x, 7 where η is the unstable internal dynamics. The goal is to apply the proposed output-feedback DSC scheme to 7 so that the system output y and the internal dynamics η can respectively track their desired signals y d and η d. A. IID Solution via SSC Method At the beginning of the simulation, we firstly need to solve the IID η d of system 7 according to the SSC method described as -4. As to the internal dynamics equation η = η + x, by setting x = y d, its corresponding IID equation shows η d = η d + y d. 7 By selecting the desired trajectory signal y d = R cosωt, it can be generated by the following exosystem ẇ = Sw, 0 ω S = ω 0 whose characteristic polynomial is, 73 P λ = λi S = λ + ω. 74 According to the above equation, we can get k =, p = 0 and p 0 = ω. By setting the desired eigenvalues s, =, thus the parameters of the characteristic polynomial are c 0 =, c =. Via the SSC method in [3], we can get P = I + Q + Q I + ω Q I = 3 ω + ω, P 0 = c 0 Q P + Ip 0 Q = 3ω + ω, 75 where Q = A η =. By taking the parameters c 0, c, P and P 0 into the matrix differential equation 4, the IID η d can be solved from the following equation: η d + c η d + c 0 η d = P θd + P 0 θ d. 76

0 IEEE/CAA JOURNAL OF AUTOMATICA SINICA, VOL. 3, NO., JANUARY 06 B. IID Comparison Here through the IID equation 7, we give a detailed comparison of IID solved via the aforementioned methods: NSI, SSC, OR. Through OR method, the accurate and analytical bounded solution is η d = Rω cosωt + Rω sinωt + ω. 77 Detailed solution procedure of NSI and OR methods can be found in [7]. By selecting R =, ω =, the solved IID η d of the system 7 can be seen in Fig.. internal dynamics does not converge to zero at all. From this simulation case, we can see that it is unavoidable to solve the IID η d which plays an important pole in acquiring fine output tracking performance. Fig.. Desired trajectory y d and actual output y in Case. Fig.. Comparison of IID η d solved via three methods. From Fig., it can be seen that the IID solved via SSC method gradually converge to the accurate and analytical IID solved via OR method, while the IID solved NSI method would diverge from the IID solved via OR method at the end of simulation time, because the off-line pre-computing procedure of NSI method is conducted backward in time. Due to the limited practical use of NSI method, we turn to the SSC method to get the causal IID of 7. C. Simulation Results The system initial states are selected as x0 = 0.95, 0, 0, 0 T, ˆx0 =, 0, 0, 0 T, η d 0 = 0.8; The observer gain is selected as k = 3, 67, 75, 57 T to let the eigenvalues of A 0 be 3; The controller parameters l = l = l 3 = 5, the filter time constant τ = τ 3 = 0.0; The output transformation matrix is chosen as M = to place the eigenvalues of A η0 at. The desired output signal is selected as y d = R cosωt. To illustrate the effectiveness of the proposed control scheme, the simulations are done under the following two cases. Case. Without solving the IID η d, we directly set η d = 0, and simultaneously select R =, ω = 0.5. Under this circumstance, we mean to stabilize the internal dynamics to zero. However, from the simulation results Figs. and 3, it can be seen that the tracking performance is poor, and the Fig. 3. The IID η d and internal dynamics η in Case. Case. The IID η d is solved via the SSC method, and the amplitude R and the frequency ω of the desired output trajectory y d switch from to. and 0.5 to, respectively, at random time T = 5 + 5 rand. Such switches may occur in the case of obstacle avoidance. From the simulation results Figs. 4-6, it can be concluded that the output tracking performance is fine, and the internal dynamics can track its corresponding casual IID despite the fact that the amplitude and frequency of the desired signal y d change at any random time. Through the simulation cases, the following conclusions can be drawn: It is necessary to solve the IID which is fundamental to achieve desired tracking performance when dealing with non-

SU: OUTPUT-FEEDBACK DYNAMIC SURFACE CONTROL FOR A CLASS OF NONLINEAR NON-MINIMUM PHASE SYSTEMS 03 Based on output redefinition, the proposed outputfeedback DSC controller for nonlinear non-minimum phase systems is effective. VII. CONCLUSION Fig. 4. Desired trajectory y d and actual output y in Case. The paper has proposed an output-feedback tracking control scheme for a class of nonlinear non-minimum phase systems via DSC method. After output redefinition, we directly design control law for the external dynamics rather than the internal dynamics, because the internal dynamics will get stable with the stability of the external dynamics. The proposed outputfeedback DSC controller not only drives the system output signal to track the desired trajectory, but also makes the unstable internal dynamics to follow its corresponding bounded IID. The stability analysis has proved that the tracking errors can converge to zero and the closed-loop system is semi-globally stable. The effectiveness of the proposed output-feedback DSC control scheme has been illustrated by the simulation results. REFERENCES [] Isidori A. Nonlinear Control of Systems. London: Spring-Verlag, 995. [] Marino R, Tomei P. Nonlinear Control Design: Geometric, Adaptive and Robust. New Jersey: Prentice-Hall, 995. [3] Shkolnikov I A, Shtessel Y B. Tracking in a class of nonminimum-phase systems with nonlinear internal dynamics via sliding mode control using method of system center. Automatica, 00, 385: 837 84 [4] Kanellakopoulos I, Kokotovic P V, Morse A S. Adaptive output-feedback control of a class of nonlinear systems. In: Proceedings of the 30th Conference on Decision and Control. Brighton, England: IEEE, 99. 08 087 Fig. 5. The IID η d and internal dynamics η in Case. [5] Yang B J, Calise A J. Adaptive stabilization for a class of non-affine non-minimum phase systems using neural networks. In: Proceedings of the 006 American Control Conference. Minneapolis, USA: IEEE, 006. 9 96 [6] Karagiannis D, Jiang Z P, Ortega R, Astolfi A. Output-feedback stabilization of a class of uncertain non-minimum-phase nonlinear systems. Automatica, 005, 49: 609 65 [7] Xie B, Yao B. Robust output feedback stabilization of a class of nonminimum phase nonlinear systems. In: Proceedings of the 006 American Control Conference. Minneapolis, USA: IEEE, 006: 498 4986 [8] Hoseini S M, Farrokhi M, Koshkouei A J. Adaptive neural network output feedback stabilization of nonlinear non-minimum phase systems. International Journal of Adaptive Control and Signal Processing, 00, 4: 65 8 [9] Isidori A. A tool for semi-global stabilization of uncertain nonminimum-phase nonlinear systems via output feedback. IEEE Transactions on Automatic Control, 000, 450: 87 87 [0] Nazrulla S, Khalil H K. Robust stabilization of non-minimum phase nonlinear systems using extended high-gain observers. IEEE Transactions on Automatic Control, 0, 564: 80 83 Fig. 6. The control signal u in Case. minimum phase systems. [] Gopalswamy S, Hedrick J K. Tracking nonlinear non-minimum phase systems using sliding control. International Journal of Control, 993, 575: 4 58 [] Devasia S, Chen D G, Paden B. Nonlinear inversion-based output tracking. IEEE Transactions on Automatic Control, 996, 47: 930 94

04 IEEE/CAA JOURNAL OF AUTOMATICA SINICA, VOL. 3, NO., JANUARY 06 [3] Huang C S, Yuan K. Output tracking of a non-linear non-minimum phase PVTOL aircraft based on non-linear state feedback control. International Journal of Control, 00, 756: 466 473 [4] Zhu B, Wang X H, Cai K Y. Approximate trajectory tracking of input-disturbed PVTOL aircraft with delayed attitude measurements. International Journal of Robust and Nonlinear Control, 00, 04: 60 6 [5] Al-Hiddabi S A, McClamroch N H. Tracking and maneuver regulation control for nonlinear nonminimum phase systems: application to flight control. IEEE Transactions on Control Systems Technology, 00, 06: 780 79 [6] Su S W, Lin Y. Robust output tracking control of a class of non-minimum phase systems and application to VTOL aircraft. International Journal of Control, 0, 84: 858 87 [7] Isidori A, Byrnes C I. Output regulation of nonlinear systems. IEEE Transactions on Automatic Control, 990, 35: 3 40 05, 4: 9 in Chinese [9] Swaroop D, Hedrick J K, Yip P P, Gerdes J C. Dynamic surface control for a class of nonlinear systems. IEEE Transactions on Automatic Control, 000, 450: 893 899 [0] Kristic M, Kanellakopoulos I, Kokotovic P V. Nonlinear design of adaptive controllers for linear systems. IEEE Transactions on Automatic Control, 994, 394: 738 75 Shanwei Su received the Ph. D. degree in control theory and control engineering from Beihang University, China in 0. He is a lecturer at the Research Institute of Unmanned Aerial Vehicle, Beihang University. His research interests include nonminimum phase systems and flight control. [8] Su Shan-Wei, Zhu Bo, Xiang Jin-Wu, Lin Yan. A survey on the control of nonlinear non-minimum phase systems. Acta Automatica Sinica,