Quadrature Based Approximations of Non-integer Order Integrator on Finite Integration Interval

Similar documents
A Comparison between Integer Order and Non-integer Order Controllers Applied to a Water Levelling System

Discretization of Fractional-Order Differentiators and Integrators

Approximating fractional derivatives through the generalized mean

SOME APPROXIMATIONS OF FRACTIONAL ORDER OPERATORS USED IN CONTROL THEORY AND APPLICATIONS. Abstract

The Time-Scaled Trapezoidal Integration Rule for Discrete Fractional Order Controllers

Using Continued Fraction Expansion to Discretize Fractional Order Derivatives

N OWADAYS, Application of Newton s Method to Analog and Digital Realization of Fractional-order Controllers

Research Article New Stability Tests for Discretized Fractional-Order Systems Using the Al-Alaoui and Tustin Operators

On the Designing of Fractional Order FIR Differentiator Using Radial Basis Function and Window

Optimal Controllers with Complex Order Derivatives

The Time-Scaled Trapezoidal Integration Rule for Discrete Fractional Order Controllers

The PLC Implementation of Fractional-Order Operator Using CFE Approximation

Numerical Integration (Quadrature) Another application for our interpolation tools!

Contents. I Basic Methods 13

Time-Domain Evaluation of Fractional Order Controllers Direct Discretization Methods

Tuning of fractional PI controllers for fractional order system models with and without time delays

A Legendre Computational Matrix Method for Solving High-Order Fractional Differential Equations

Robust PID and Fractional PI Controllers Tuning for General Plant Model

An Improved Design for the IIR Digital Fractional order Differential Filter using Indirect Discretization Approach

Introduction to Uncertainty Quantification in Computational Science Handout #3

Ch. 03 Numerical Quadrature. Andrea Mignone Physics Department, University of Torino AA

Rational Chebyshev pseudospectral method for long-short wave equations

Recurrence Relations and Fast Algorithms

Comparison of Clenshaw-Curtis and Gauss Quadrature

Stabilization of fractional positive continuous-time linear systems with delays in sectors of left half complex plane by state-feedbacks

MASSACHUSETTS INSTITUTE OF TECHNOLOGY Department of Mechanical Engineering Dynamics and Control II Fall 2007

The concept of fractional-order

Integration, differentiation, and root finding. Phys 420/580 Lecture 7

Quadrature for Uncertainty Analysis Stochastic Collocation. What does quadrature have to do with uncertainty?

Improving the Control Energy in Model Reference Adaptive Controllers Using Fractional Adaptive Laws

Section 6.6 Gaussian Quadrature

APPROXIMATIONS TO FRACTIONAL- ORDER LINEAR TIME INVARIANT SYSTEMS

Design and Tuning of Fractional-order PID Controllers for Time-delayed Processes

Lecture 9 Infinite Impulse Response Filters

12.0 Properties of orthogonal polynomials

EFFICIENT SPECTRAL COLLOCATION METHOD FOR SOLVING MULTI-TERM FRACTIONAL DIFFERENTIAL EQUATIONS BASED ON THE GENERALIZED LAGUERRE POLYNOMIALS

NUMERICAL COMPUTATION IN SCIENCE AND ENGINEERING

A FAST SOLVER FOR ELLIPTIC EQUATIONS WITH HARMONIC COEFFICIENT APPROXIMATIONS

In numerical analysis quadrature refers to the computation of definite integrals.

PART IV Spectral Methods

Synthesis of complete orthonormal fractional bases

MATHEMATICAL FORMULAS AND INTEGRALS

Numerical Methods for Engineers. and Scientists. Applications using MATLAB. An Introduction with. Vish- Subramaniam. Third Edition. Amos Gilat.

Index. for Ɣ(a, z), 39. convergent asymptotic representation, 46 converging factor, 40 exponentially improved, 39

MA2501 Numerical Methods Spring 2015

Pseudo-fractional ARMA modelling using a double Levinson recursion

TUNING-RULES FOR FRACTIONAL PID CONTROLLERS

FRACTIONAL DIFFERENTIAL EQUATIONS

An Accurate Fourier-Spectral Solver for Variable Coefficient Elliptic Equations

Backlash Vibration Suppression Control of Torsional System by Novel Fractional Order PID k Controller

SIMPLE CONDITIONS FOR PRACTICAL STABILITY OF POSITIVE FRACTIONAL DISCRETE TIME LINEAR SYSTEMS

MATHEMATICAL FORMULAS AND INTEGRALS

Lecture 3 - Design of Digital Filters

DOCTORAL THESIS Extended Abstract

The Perrin Conjugate and the Laguerre Orthogonal Polynomial

Digital Wideband Integrators with Matching Phase and Arbitrarily Accurate Magnitude Response (Extended Version)

Applied Numerical Analysis

machines ISSN

MODELLING OF RECIPROCAL TRANSDUCER SYSTEM ACCOUNTING FOR NONLINEAR CONSTITUTIVE RELATIONS

CHAPTER 7 FRACTIONAL ORDER SYSTEMS WITH FRACTIONAL ORDER CONTROLLERS

Recursive, Infinite Impulse Response (IIR) Digital Filters:

ISSN (Print) Research Article. *Corresponding author Nitin Rawal

256 Summary. D n f(x j ) = f j+n f j n 2n x. j n=1. α m n = 2( 1) n (m!) 2 (m n)!(m + n)!. PPW = 2π k x 2 N + 1. i=0?d i,j. N/2} N + 1-dim.

The Cooper Union Department of Electrical Engineering ECE111 Signal Processing & Systems Analysis Final May 4, 2012

Chapter 1 Divide and Conquer Polynomial Multiplication Algorithm Theory WS 2015/16 Fabian Kuhn

Instructions for Matlab Routines

MOMENT functions are used in several computer vision

Math 411 Preliminaries

Numerical Methods. King Saud University

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

Analysis of eddy currents in a gradient coil

Two special equations: Bessel s and Legendre s equations. p Fourier-Bessel and Fourier-Legendre series. p

Limits at Infinity. Horizontal Asymptotes. Definition (Limits at Infinity) Horizontal Asymptotes

Differentiation and Integration

Mathematics for Engineers. Numerical mathematics

SHAPE OF IMPULSE RESPONSE CHARACTERISTICS OF LINEAR-PHASE NONRECURSIVE 2D FIR FILTER FUNCTION *

Exam in TMA4215 December 7th 2012

An Analysis of Five Numerical Methods for Approximating Certain Hypergeometric Functions in Domains within their Radii of Convergence

Stability of interval positive continuous-time linear systems

Set-point Filter Design for a Two-degree-of-freedom Fractional Control System

Numerical Methods I: Numerical Integration/Quadrature

Solving Linear Time Varying Systems by Orthonormal Bernstein Polynomials

The Hardy-Littlewood Function: An Exercise in Slowly Convergent Series

Quadrature Prefilters for the Discrete Wavelet Transform. Bruce R. Johnson. James L. Kinsey. Abstract

ONE - DIMENSIONAL INTEGRATION

Multistage Methods I: Runge-Kutta Methods

Quarter-Sweep Gauss-Seidel Method for Solving First Order Linear Fredholm Integro-differential Equations

CS 450 Numerical Analysis. Chapter 8: Numerical Integration and Differentiation

Numerical Methods with MATLAB

8.3 Partial Fraction Decomposition

Notes on L (Optimal) Filters

Preface. 2 Linear Equations and Eigenvalue Problem 22

Numerical Analysis Preliminary Exam 10 am to 1 pm, August 20, 2018

Scientific Computing: Numerical Integration

Numerical Analysis. A Comprehensive Introduction. H. R. Schwarz University of Zürich Switzerland. with a contribution by

Fractional-Order Modelling in Modelica

INTEGRAL TRANSFORMS and THEIR APPLICATIONS

(Refer Slide Time: 01:28 03:51 min)

Fractional Order Oscillator with Independent Control of Phase and Frequency

A Delay-dependent Condition for the Exponential Stability of Switched Linear Systems with Time-varying Delay

Transcription:

Quadrature Based Approximations of Non-integer Order Integrator on Finite Integration Interval Jerzy Baranowski Abstract Implementation of non-integer order systems is the subject of an ongoing research. In this paper we consider the approximation of non-integer order integrator with the use of diffusive realization of pseudo differential operator. We propose a transformation of variables allowing easier approximation with use of quadratures. We then analyze the convergence and discuss the consequences of reduction in the integration interval. Keywords Diffusive realization Approximation Non-integer order integrator Quadratures 1 Introduction Non-integer order systems take an increasing part in science and engineering. Their applications include, among the others, modeling, control and signal processing. In most of the problems the infinite memory of non-integer order systems introduces problems with realization. Because of that realizations cannot come directly from the definition but have to take form of integer order realizations. In this paper, the problem of realization of non-integer order integrator, i.e., 1/s α is considered. A method based on diffusive realization of pseudo differential operators is investigated. Currently realizations on non-integer order systems are focused in four areas. First area is based on approximating the s α operator in the frequency domain. The most popular approaches are the Oustaloup s method [1, 2] and CFE (Continuous fraction expansion) method [3 5]. Both these methods are based on different premises Work realised in the scope of project titled Design and application of non-integer order subsystems in control systems. Project was financed by National Science Centre on the base of decision no. DEC-2013/09/D/ST7/03960. J. Baranowski (B) AGH University of Science and Technology, Al. Mickiewicza 30, 30-059 Cracow, Poland e-mail: jb@agh.edu.pl Springer International Publishing AG 2017 A. Babiarz et al. (eds.), Theory and Applications of Non-integer Order Systems, Lecture Notes in Electrical Engineering 407, DOI 10.1007/978-3-319-45474-0_2 11

12 J. Baranowski but both allow obtaining relatively easy approximations. CFE method is generally considered as worse of the two in the aspect of frequency response representation [6]. Ostaloup s method gives a very good representation of frequency response at the cost of high numerical sensitivity. This sensitivity can be lessened with use of time domain realizations [7, 8]. Second method of approximation relies on discrete realizations based on truncation of series representations. In particular, the methods include truncation of Grünwald Letnikov derivative definitions or power series expansions (PSE) in the z variable (discrete frequency) domain. These methods give good approximations only at high frequencies. Examples of the method can be found in [9]. Improvements on PSE methods with better band of correct approximation were investigated by Ferdi [10, 11]. Third method of approximation is based on approximation of non-integer order system impulse response with Laguerre functions. Early works in this area were unsuccessful in approximation of α order integrators [12, 13]. In [14] itwasshown that under certain assumptions method is convergent in L 1 and L 2 norms when approximating asymptotically stable non-integer transfer functions of relative degree equal or greater than one. Most applications of the methods are in filtering and parametric optimization [15 18]. Fourth type of approximation that recently rises in popularity is the method using diffusive realization of pseudo differential operators [19] for approximation. First works in the area used finite dimensional approximation with trapezoidal integration [20]. Later works used simple diagonal matrix time domain realization with modified Oustaloup nodes from frequency domain method [21, 22]. This method is especially useful in analysis of infinite dimensional systems using operator theory. It can be used, for example, to prove such properties as stability of closed loop system [23]. In this paper, diffusive realization is used for construction of approximation using quadrature methods. In particular, frequency domain representation of non-integer order integral will be used, i.e., 1/s α. For such system the input-output mapping will be approximated using two quadratures on finite integration intervals allowing creation of finite dimensional approximation in the form of sum of first order lags. The rest of the paper is organized as follows. Next section presents a theorem that is a basis of investigation along with its proof. Next, the approximation method is presented and a variable transformation allowing effective use of quadratures. Then two types of quadratures are analyzed. Operation of both quadratures is investigated with the use of H norm as an indicator of approximation quality. Then results are discussed and future works directions are presented. 2 Diffusive Realization of Non-integer Order Integrator As the basis for diffusive realization we consider the following theorem (see e.g. [21]) which determines equivalent form for integrator of order α.

Quadrature Based Approximations of Non-integer 13 Theorem 1 For α (0, 1) and s C we have: 1 s = sin απ 1 1 α 0 π x α s + x x.. (1) Proof Proof will be omitted, however it relies on computing inverse Laplace transform of 1/s α using contour integration using Bromwich contour (see Fig. 1). Remark 1 Grabowski in [23] proved the same result in a different way. The given proof is based on direct calculation of right-hand side of (1) using integration contour as in Fig. 2. This proof, however, is not constructive. The main advantage of the method used in Theorem 1 is that we do not need to integrate non-integer power of s. It allows to approximate the integral. 3 Approximation of Diffusive Realizations The basic approach to approximation of integrals relies on using infinite sum of form 0 sin απ π 1 x α 1 s + x x. n i=0 b i s + x i (2) Fig. 1 Bromwich contour

14 J. Baranowski Fig. 2 Contour from [23] where x i, are denote nodes of quadrature while b i are products of quadrature weights and integrand values. The choice of quadrature for approximation is not straightforward. In [20] the authors used trapezoidal rule. In [21], rectangle rule with nonuniform nodes was considered. Both approaches transform the integration interval from [0, ] to (0, ω max ] as x i are important only for high frequencies. It causes, however, errors in integrating, especially near ω max. Despite the fact that integral (2) is defined on infinite interval, multiple authors had considered limiting the interval only to the one including the interesting frequency band. In this paper we will focus on such bounded intervals, but using more advanced quadratures. It can be inferred, that because approximant (2) is a sum of first order low pass filters the quality of frequency response representation will depend on time constants. Because we are interested in good representation of frequencies for different orders of magnitude a following transformation is proposed. Let us set x = 10 θ Integrating (1) forx [10 a, 10 b ] reduces to b a sin απ π log(10)10 θ(1 α) 1 dθ. (3) s + 10θ This formula, while at first glance complicated allows balancing high and low frequencies. Following section contains analysis of quadratures for finite intervals.

Quadrature Based Approximations of Non-integer 15 4 Quadratures on Infinite Intervals In this section we will consider two types of quadratures on finite intervals, originating from interval [ 1, 1]. We will however present very elegant matlab code examples given originally by Trefethen [24]. Provided programs are fully functional and self explanatory. The first quadrature we consider is the Gauss quadrature, also known as Gauss Legendre quadrature. It is a quadrature, nodes of which are roots of Legendre polynomials, while formulas for weights one can find for example in [25]. Main advantage of the Gauss Legendre quadrature comes from the orthogonality of Legendre polynomials. In particular, let us consider a polynomial p(x) of order 2N + 1. Such polynomial can always be written as p(x) = q(x) l N+1 (x) + r(x), where q(x) is a polynomial of degree N + 1, l N+1 (x) is a Legendre polynomial of degree N + 1 and r(x) is a polynomial of degree N. When considering interpolation quadrature of degree N + 1 on nodes, we can easily see that 1 1 q(x) l N (x)dx = 0 because of orthogonality of Legendre polynomials and the integral of r(x) is exact because of uniqueness of interpolation polynomials. These facts give that Gauss Legendre quadratures on N + 1 nodes are exact for polynomials of order 2N + 1. Following code uses eigenvalue decomposition to obtain both weights and nodes. function I = gauss(f,n) % (n +1) - pt Gauss quadrature of f beta =.5./sqrt(1-(2*(1:n)).^(-2)); % 3-term recurrence coeffs T = diag(beta,1) + diag(beta,-1); % Jacobi matrix [V,D] = eig(t); % eigenvalue decomposition x = diag(d); [x,i] = sort(x); % nodes (= Legendre points) w = 2*V(1,i).^2; % weights I = w*feval(f,x); % the integral The second quadrature is the Clenshaw Curtis quadrature. It is an interpolation quadrature on Chebyshev nodes. In certain aspects it is also equivalent to discrete cosine transform, and can be computed using FFT. Again code given by Trefethen: function I = clenshaw_curtis(f,n) % (n+1)-pt C-C quadrature of f x = cos(pi*(0:n)'/n); % Chebyshev points fx = feval(f,x)/(2*n); % f evaluated at these points g = real(fft(fx([1:n+1 n:-1:2]))); % fast Fourier transform a = [g(1); g(2:n)+g(2*n:-1:n+2); g(n+1)]; % Chebyshev coeffs w = 0*a'; w(1:2:end) = 2./(1-(0:2:n).^2); % weight vector I = w*a; % the integral While theory gives us, that Gauss quadratures, thanks to orthogonality, are exact for polynomials of twice higher order than interpolation quadratures it is observed that in practice Clenshaw Curtis quadrature gives very similar precision [24]. Moreover it has much smaller computational complexity. It should be also noted, that while based on interpolation both of these quadratures are immune to Runge s phenomenon, as both Legendre and Chebyshev nodes have asymptotic distribution of 1/(1 x 2 ) for [ 1, 1] interval.

16 J. Baranowski 5 Approximation Analysis In order to analyze the approximation performance we conducted error analysis using H norm. We considered the difference e(s) = 1 n s b i α s + x i for frequency [10 5, 10 5 ] rad/s. The order of quadrature was increased in order to observe the change of norm of e. For initial analysis we have considered the integration interval of [ 5, 5] for (3) corresponding to [10 5, 10 5 ] rad/s for original integral. Both quadratures were tested simultaneously. In illustrated results α = 1/2 was considered. In the Fig. 3 we depicted this analysis. As it can be observed after initial convergence error stops decreasing and fixes on a certain value. This error is caused by the limitation of infinite interval into a finite one. Indeed one can observe in the Fig. 4 that increasing the interval to [ 7, 7] results in dropping of error by an order of magnitude. The fixed error is connected to the phase errors at low frequencies. As one can observe in the Fig. 5, 50th order approximate has a phase error, especially noticeable at low frequencies, where the gain of the integrator is the largest. Because of that any i=0 Fig. 3 Analysis of e norm with increasing order of quadrature for the integration interval of [ 5, 5]

Quadrature Based Approximations of Non-integer 17 Fig. 4 Analysis of e norm with increasing order of quadrature for the integration interval of [ 7, 7] Magnitude [db] 60 40 20 0-20 -40 Approximation order:50, Integration interval: [-7, 7] -60 10-5 10-4 10-3 10-2 10-1 10 0 10 1 10 2 10 3 10 4 10 5 Phase shift [deg] -42-43 -44-45 -46-47 Original Clenshaw Curtis Gauss-Laguerre -48 10-5 10-4 10-3 10-2 10-1 10 0 10 1 10 2 10 3 10 4 10 5 Frequency [rad/sec] Fig. 5 Analysis of frequency response of 50th order approximation with the integration interval of [ 7, 7]. Clenshaw Curtis and Gauss Legendre quadratures are overlapping

18 J. Baranowski Magnitude [db] Phase shift [deg] 60 40 20 0-20 -40 Approximation order:15, Integration interval: [-7, 7] -60 10-5 10-4 10-3 10-2 10-1 10 0 10 1 10 2 10 3 10 4 10 5-35 -40-45 -50-55 10-5 10-4 10-3 10-2 10-1 10 0 10 1 10 2 10 3 10 4 10 5 Frequency [rad/sec] Original Clenshaw Curtis Gauss-Laguerre Fig. 6 Analysis of frequency response of 15th order approximation with the integration interval of [ 7, 7] error in phase is multiplied by approximately 300. What is interesting the magnitude is almost ideally represented. As one can observe in the Fig. 4 there are values, where the error is smaller than the observed bias. For example such value is N = 15. In that case Gauss Laguerre quadrature gives much better result. It is however caused by not obtaining convergence yet. It can be seen in Fig. 6, that the phase fluctuates strongly, and for such N and this type of quadrature is very close to the ideal value for low frequencies. Such approximation, however, is obviously not acceptable for use. 6 Conclusion In this paper, we considered diffusive realization of integrator of order α and its approximation with use of quadratures on finite intervals. It can be observed, that initial convergence is relatively quick but bias caused by reduction of integration interval is fixed. This bias depends on integration interval, and as a-rule-of-a-thumb integration interval has to be greater than the band pass where the approximation is needed. The performance indicator taken as the error of H norm is dominated by lowfrequency terms, especially the phase errors. Nevertheless, the method should be further considered, especially in order to accelerate the convergence as there are

Quadrature Based Approximations of Non-integer 19 certain methods which show better performance for lower orders (see, e.g. Oustaloup method). It is probably due to the slow decay of function in diffusive realization (1). However, high order of approximation should not be feared, as system is directly realized with independent linear differential equations of order one which are very easily discretized. It should be also noted, that Oustaloup approximation, which has only geometric justification outperforms this approximation significantly for bounded intervals. This suggests, that this type of approximation has only the theoretical significance. Further works should be concentrated on using quadratures on finite intervals along with asymptotic approximation on the tail of the function. References 1. Oustaloup, A.: La Commande CRONE: Commande Robuste D ordre Non Entier. Hermes (1991) 2. Oustaloup, A., Levron, F., Mathieu, B., Nanot, F.M.: Frequency-band complex noninteger differentiator: characterization and synthesis. IEEE Trans. Circuits Syst. I, Fundam. Theory, 47(1), 25 39 (2000) 3. Vinagre, B.M., Chen, Y.Q., Petráš, I.: Two direct Tustin discretization methods for fractionalorder differentiator/integrator. J. Franklin Inst. 340(5), 349 362 (2003) 4. Vinagre, B., Podlubny, I., Hernandez, A., Feliu, V.: Some approximations of fractional order operators used in control theory and applications. Frac. Calc. Appl. Anal. 3(3), 231 248 (2000) 5. Chen, Y., Vinagre, B.M., Podlubny, I.: Continued fraction expansion approaches to discretizing fractional order derivatives-an expository review. Nonlinear Dyn. 38(1 4), 155 170 (2004) 6. Monje, C.A., Chen, Y., Vinagre, B.M., Xue, D., Feliu, V.: Fractional-order Systems and Controls. Fundamentals and Applications. Advances in Industrial Control. Springer, London (2010) 7. Baranowski, J., Bauer, W., Zagórowska, M., Dziwiński, T., Piątek, P.: Time-domain Oustaloup approximation. In: IEEE 20th International Conference on Methods and Models in Automation and Robotics (MMAR), pp. 116 120 (2015) 8. Baranowski, J., Bauer, W., Zagórowska, M.: Stability properties of discrete time-domain Oustaloup approximation. In Domek, S., Dworak, P. (eds.) Theoretical Developments and Applications of Non-Integer Order Systems, vol. 357 of Lecture Notes in Electrical Engineering, pp. 93 103. Springer International Publishing (2016) 9. Tseng, C.C.: Design of FIR and IIR fractional order Simpson digital integrators. Signal Process. 87(5), 1045 1057 (2007) 10. Leulmi, F., Ferdi, Y.: Improved digital rational approximation of the operator S α using secondorder s-to-z transform and signal modeling. Circuits Syst. Signal Process. 34(6), 1869 1891 (2015) 11. Ferdi, Y.: Computation of fractional order derivative and integral via power series expansion and signal modelling. Nonlinear Dyn. 46(1), 1 15 (2006) 12. Maione, G.: Laguerre approximation of fractional systems. Electron. Lett. 38(20), 1234 1236 (2002) 13. Aoun, M., Malti, R., Levron, F., Oustaloup, A.: Synthesis of fractional Laguerre basis for system approximation. Automatica 43(9), 1640 1648 (2007) 14. Bania, P., Baranowski, J.: Laguerre polynomial approximation of fractional order linear systems. In: Mitkowski, W., Kacprzyk, J., Baranowski, J. (eds.) Advances in the Theory and Applications of Non-integer Order Systems: 5th Conference on Non-integer Order Calculus and Its Applications, vol. 257 of the series Lecture Notes in Electrical Engineering, pp. 171 182. Springer (2013)

20 J. Baranowski 15. Zagórowska, M., Baranowski, J., Bania, P., Piątek, P., Bauer, W., Dziwiński, T.: Impulse response approximation method for fractional order lag. In: Latawiec, K.J., Łukaniszyn, M., Stanisławski, R. (eds.) Advances in Modelling and Control of Noninteger-order Systems - 6th Conference on Non-Integer Order Caculus and its Applications, vol. 320 of the series Lecture Notes in Electrical Engineering, pp. 113 122. Springer, Heidelberg (2014) 16. Zagórowska, M., Baranowski, J., Bania, P., Bauer, W., Dziwiński, T., Piątek, P.: Parametric optimization of PD controller using Laguerre approximation. In: IEEE 20th International Conference on Methods and Models in Automation and Robotics (MMAR), pp. 104 109, (2015) 17. Zagórowska, M.: Parametric optimization of non-integer order PD μ controller for delayed system. In: Domek, S., Dworak, P. (eds.) Theoretical Developments and Applications of Non- Integer Order Systems, vol. 357 of Lecture Notes in Electrical Engineering, pp. 259 270. Springer, Heidelberg (2016) 18. Baranowski, J., Bauer, W., Zagórowska, M., Piątek, P.: On digital realizations of non-integer order filters. Circuits Syst. Signal Process (2016) 19. Montseny, G.: Diffusive representation of pseudo-differential time-operators. ESAIM: Proc. 5, 159 175 (1998) 20. Heleschewitz, D., Matignon, D.: Diffusive realisations of fractional integrodifferential operators: Structural analysis under approximation. In: IFAC conference Systems Structure and Control, vol. 2, Nantes, France pp. 243 248 (1998) 21. Trigeassou, J., Maamri, N., Sabatier, J., Oustaloup, A.: State variables and transients of fractional order differential systems. Comput. Math. Appl. 64(10), 3117 3140 (2012) (Advances in FDE, III) 22. Liang, S., Peng, C., Liao, Z., Wang, Y.: State space approximation for general fractional order dynamic systems. Int. J. Syst. Sci. 45(10), 2203 2212 (2014) 23. Grabowski, P.: Stabilization of wave equation using standard/fractional derivative in boundary damping. In: Mitkowski, W., Kacprzyk, J., Baranowski, J. (eds.) Advances in the Theory and Applications of Non-integer Order Systems: 5th Conference on Non-integer Order Calculus and Its Applications, vol. 257 of the series Lecture Notes in Electrical Engineering, pp. 101 121. Springer International Publishing, Heidelberg (2013) 24. Trefethen, L.N.: Is Gauss quadrature better than Clenshaw-Curtis? SIAM Rev. 50(1), 67 87 (2008) 25. Abramowitz, M., Stegun, I.A.: Handbook of Mathematical Functions with Formulas, Graphs, and Mathematical Tables. Dover, New York (1964)

http://www.springer.com/978-3-319-45473-3