Research article Special Issues

A pseudo-spectral approach for optimal control problems of variable-order fractional integro-differential equations

  • Nonlinear optimal control problems governed by variable-order fractional integro-differential equations constitute an important subgroup of optimal control problems. This group of problems is often difficult or impossible to solve analytically because of the variable-order fractional derivatives and fractional integrals. In this article, we utilized the expansion of Lagrange polynomials in terms of Chebyshev polynomials and the power series of Chebyshev polynomials to find an approximate solution with high accuracy. Subsequently, by employing collocation points, the problem was transformed into a nonlinear programming problem. In addition, variable-order fractional derivatives in the Caputo sense were represented by a new operational matrix, and an operational matrix represented fractional integrals. As a result, the mentioned integro-differential optimal control problem becomes a nonlinear programming problem that can be easily solved with the repetitive optimization method. In the end, the proposed method is illustrated by numerical examples that demonstrate its efficiency and accuracy.

    Citation: Zahra Pirouzeh, Mohammad Hadi Noori Skandari, Kamele Nassiri Pirbazari, Stanford Shateyi. A pseudo-spectral approach for optimal control problems of variable-order fractional integro-differential equations[J]. AIMS Mathematics, 2024, 9(9): 23692-23710. doi: 10.3934/math.20241151

    Related Papers:

    [1] Obaid Algahtani, M. A. Abdelkawy, António M. Lopes . A pseudo-spectral scheme for variable order fractional stochastic Volterra integro-differential equations. AIMS Mathematics, 2022, 7(8): 15453-15470. doi: 10.3934/math.2022846
    [2] Rohul Amin, Kamal Shah, Hijaz Ahmad, Abdul Hamid Ganie, Abdel-Haleem Abdel-Aty, Thongchai Botmart . Haar wavelet method for solution of variable order linear fractional integro-differential equations. AIMS Mathematics, 2022, 7(4): 5431-5443. doi: 10.3934/math.2022301
    [3] Yumei Chen, Jiajie Zhang, Chao Pan . Numerical approximation of a variable-order time fractional advection-reaction-diffusion model via shifted Gegenbauer polynomials. AIMS Mathematics, 2022, 7(8): 15612-15632. doi: 10.3934/math.2022855
    [4] A. H. Tedjani, A. Z. Amin, Abdel-Haleem Abdel-Aty, M. A. Abdelkawy, Mona Mahmoud . Legendre spectral collocation method for solving nonlinear fractional Fredholm integro-differential equations with convergence analysis. AIMS Mathematics, 2024, 9(4): 7973-8000. doi: 10.3934/math.2024388
    [5] Younes Talaei, Sanda Micula, Hasan Hosseinzadeh, Samad Noeiaghdam . A novel algorithm to solve nonlinear fractional quadratic integral equations. AIMS Mathematics, 2022, 7(7): 13237-13257. doi: 10.3934/math.2022730
    [6] Umair Ali, Sanaullah Mastoi, Wan Ainun Mior Othman, Mostafa M. A Khater, Muhammad Sohail . Computation of traveling wave solution for nonlinear variable-order fractional model of modified equal width equation. AIMS Mathematics, 2021, 6(9): 10055-10069. doi: 10.3934/math.2021584
    [7] Ismail Gad Ameen, Dumitru Baleanu, Hussien Shafei Hussien . Efficient method for solving nonlinear weakly singular kernel fractional integro-differential equations. AIMS Mathematics, 2024, 9(6): 15819-15836. doi: 10.3934/math.2024764
    [8] Chuanli Wang, Biyun Chen . An $ hp $-version spectral collocation method for fractional Volterra integro-differential equations with weakly singular kernels. AIMS Mathematics, 2023, 8(8): 19816-19841. doi: 10.3934/math.20231010
    [9] Xiaojun Zhou, Yue Dai . A spectral collocation method for the coupled system of nonlinear fractional differential equations. AIMS Mathematics, 2022, 7(4): 5670-5689. doi: 10.3934/math.2022314
    [10] Sumbal Ahsan, Rashid Nawaz, Muhammad Akbar, Saleem Abdullah, Kottakkaran Sooppy Nisar, Velusamy Vijayakumar . Numerical solution of system of fuzzy fractional order Volterra integro-differential equation using optimal homotopy asymptotic method. AIMS Mathematics, 2022, 7(7): 13169-13191. doi: 10.3934/math.2022726
  • Nonlinear optimal control problems governed by variable-order fractional integro-differential equations constitute an important subgroup of optimal control problems. This group of problems is often difficult or impossible to solve analytically because of the variable-order fractional derivatives and fractional integrals. In this article, we utilized the expansion of Lagrange polynomials in terms of Chebyshev polynomials and the power series of Chebyshev polynomials to find an approximate solution with high accuracy. Subsequently, by employing collocation points, the problem was transformed into a nonlinear programming problem. In addition, variable-order fractional derivatives in the Caputo sense were represented by a new operational matrix, and an operational matrix represented fractional integrals. As a result, the mentioned integro-differential optimal control problem becomes a nonlinear programming problem that can be easily solved with the repetitive optimization method. In the end, the proposed method is illustrated by numerical examples that demonstrate its efficiency and accuracy.



    In recent decades, fractional calculus has been employed as a powerful tool to describe many engineering and physical phenomena [1,3,4,6,16]. A new concept that has been introduced recently is a variable-order fractional operator, which is an extension of classical fractional calculus that can vary in terms of time, place, or any other variable. These types of operators are of considerable importance due to their memory property, allowing many scientific problems to be modeled using differential equations based on these operators [5]. Therefore, it is important to find the approximate solution to variable-order fractional differential equations (VOFDEs). In the variable-order fractional optimal control problems (VOFOCPs), the VOFDEs are considered as the dynamic system of problems, which can be defined according to different definitions of fractional derivatives, with the the Riemann-Liouville and Caputo derivatives being the most important ones. In [7], the fractional-order Bessel wavelets were used by Dehestani et al. to solve optimal control problems under variable-order fractional dynamical systems. They proposed a collocation method and used pseudo-operational matrices of variable-order fractional derivatives, and dual operational matrices to solve the problem. In [10], Heydari and Avazzadeh used the Legendre wavelets to solve VOFOCPs. Then, using the operational matrix of the Riemann-Liouville fractional integration and the Legendre wavelet properties, they turned the performance index into a nonlinear algebraic equation and the dynamic system into a system of algebraic equations. Heydari [8] solved a class of VOFOCPs by introducing cardinal Chebyshev polynomials and obtaining their operational matrix corresponding to the Atangana-Baleanu-Caputo derivative.

    In recent years, numerical techniques have been developed to propose approximate solutions for variable-order fractional integro-differential equations (VOFIDEs). As shown in [9], Heydari presented a method for computing nonlinear fractional quadratic integral equation solutions based on Chebyshev cardinal wavelets and a new operational matrix of variable-order fractional integration derived for the mentioned basis functions. Substituting the mentioned expansion into the intended problem results in a system of nonlinear algebraic equations. In [12], by applying piecewise integral quadratic spline interpolation to the estimation of fractional integral operators with variable order, new discretization techniques were proposed. In [13], using the second kind of Chebyshev polynomials, differential operational matrices, and integral operational matrices were derived. After adopting the collocation points, the original equation can be transformed into an algebraic system by combining two types of operational matrices. However, no studies have been published in the area of solving the optimal control (OC) problems governed by VOFIDEs. In this article, we attempt to find an approximate solution to such problems for the first time. We first solve these problems by using the collocation spectral method and a different definition of Lagrange polynomial, which is based on Chebyshev polynomials. By substituting the power series of the Chebyshev-modified polynomials, we then transform the dynamic system into a system of algebraic equations. By accurately calculating the fractional integral and derivative of the power series at collocation points, we can calculate the derivative matrix with precision. Also, we approximate the integral objective function using the Gauss quadrature rules.

    The structure of the paper is as follows. In Section 2, some necessary preliminaries are given. A statement of the problem is presented in Section 3. In Section 4, a Chebyshev pseudo-spectral approach is given to solve the OC problem of VOFIDEs. Also, we present a convergence analysis for the method. In Section 5, some numerical examples are given to show the efficiency of the method. Finally, the conclusions and suggestions are presented in Section 6.

    In this section, we present some basic definitions and mathematical preliminaries related to the fixed-order and variable-order fractional derivatives (see [2]).

    Definition 2.1. Let χ(.) be defined on the interval [0,Tf]. The left and right Riemann-Liouville fractional integrals of fixed-order μ>0 are denoted by 0Iμτχ(τ) and τIμTfχ(τ), respectively, and defined by

    0Iμτχ(τ)=1Γ(μ)τ0(τη)μ1χ(η)dη,0<τTf, (2.1)
    τIμTfχ(τ)=1Γ(μ)Tfτ(ητ)μ1χ(η)dη,0τ<Tf. (2.2)

    Definition 2.2. Consider χ(.) a function defined on interval [0,Tf]. The left and right Riemann-Liouville fractional derivatives of fixed-order 0<μ<1 are denoted by 0Dμτχ(τ) and τDμTfχ(τ), respectively, and defined by

    0Dμτχ(τ)=1Γ(1μ)ddττ0(τη)μχ(η)dητ>0, (2.3)
    τDμTfχ(τ)=(1)Γ(1μ)ddτTfτ(ητ)μχ(η)dη,τ<Tf. (2.4)

    Definition 2.3. Let us suppose that function χ(.) is defined on the finite interval [0,Tf]. The left and right Caputo fractional derivatives of χ(.) of fixed-order 0<μ<1 are denoted by C0Dμτχ(τ) and CτDμTfχ(τ), respectively, and defined by

    C0Dμτχ(τ)=1Γ(1μ)τ0(τη)μχ(η)dη,0τ<Tf, (2.5)
    CTfDμTfχ(τ)=(1)Γ(1μ)Tfτ(ητ)μχ(η)dη,0<τTf. (2.6)

    We now present the basic concepts of variable-order fractional calculus and take the fractional order in the derivative and integral as a continuous function on (0,Tf). First, we introduce the generalization of a fixed-order fractional integral called the variable-order Riemann-Liouville integral.

    Definition 2.4. Assuming that the continuously differentiable function χ is defined on (0,Tf). The left and right Riemann-Liouville fractional integrals of order μ(τ) are defined as follows:

    0Iμ(τ)τχ(τ)=τ01Γ(μ(τ))(τη)μ(τ)1χ(η)dη,τ>0, (2.7)

    and

    τIμ(τ)Tfχ(τ)=Tfτ1Γ(μ(τ))(ητ)μ(τ)1χ(η)dη,τ<Tf. (2.8)

    Definition 2.5. Consider χ:[0,Tf]R as a continuously differentiable function and suppose μ:[0,Tf][0,1] is a given function.

    (1) The type I left and right Caputo variable-order fractional derivatives (VOFDs) of χ(τ) of order μ(.), respectively, are defined by

    C0Dμ(τ)τχ(τ)=1Γ(1μ(τ))ddττ0(τη)μ(τ)[χ(η)χ(0)]dη, (2.9)
    CTfDμ(τ)τχ(τ)=1Γ(1μ(τ))ddτTfτ(ητ)μ(τ)[χ(η)χ(Tf)]dη. (2.10)

    (2) The type II left and right Caputo VOFDs of χ(τ) of order μ(.), respectively, are given by

    CTfDμ(τ)τχ(τ)=ddτ(1Γ(1μ(τ))τ0(τη)μ(τ)[χ(η)χ(0)]dη), (2.11)
    CTfDμ(τ)τχ(τ)=ddτ(1Γ(1μ(τ))Tfτ(ητ)μ(τ)[χ(η)χ(Tf)]dη). (2.12)

    (3) The type III left and right Caputo VOFDs of χ(τ) of order μ(τ), respectively, are defined by

    C0Dμ(τ)τχ(τ)=1Γ(1μ(τ))τ0(τη)μ(τ)χ(η)dη, (2.13)
    CτDμ(τ)Tfχ(τ)=1Γ(1μ(τ))Tfτ(ητ)μ(τ)χ(η)dη. (2.14)

    Lemma 2.1. Let χ(τ)=(τa)β for τ[a,b] where β>0. Then,

    CaDμ(τ)τχ(τ)={Γ(β+1)Γ(βμ(τ)+1)(τa)βμ(τ),β1,0,β<1. (2.15)
    aIτμ(τ)χ(τ)=Γ(β+1)Γ(β+μ(τ)+1)(τa)β+μ(τ). (2.16)

    In this paper, we will focus on the type Ⅲ Caputo VOFDs.

    We consider a special case of the Jacobi polynomials-Chebyshev polynomials (of the first kind), ˜Tn(t), which are proportional to Jacobi polynomials J12,12n and are orthogonal with respect to the weight function ω(t)=(1t2)12.

    The three-term recurrence relation for the Chebyshev polynomial reads

    ˜Tn+1(t)=2τ˜Tn(t)˜Tn1(t),n1,
    ˜T0(t)=1,˜T1(t)=t,1t1.

    For practical use of Chebyshev polynomials on the interval of interest [0,Tf], it is necessary to change the defining domain by means of the following substitution:

    τ=(Tf2)(t+1),0τTf,1t1.

    So, the shifted Chebyshev polynomials STn(τ) on [0,Tf] are obtained as follows:

    STn(τ)=˜Tn(2Tfτ1),0τTf,n=1,2,...

    The orthogonality condition for these shifted polynomials is

    Tf0STn(τ)STm(τ)1(2τTf1)2dτ={πTf4, if n=m=1,2,,πTf2, if n=m=0,0, if nm. (2.17)

    Shifted Chebyshev polynomials can be analytically written as follows:

    STn(τ)=np=0bp,nτp,n=0,1,2,...,bp,n=(1)npn(n+p1)!22p(np)!(2p)!Tfp. (2.18)

    In this paper, we consider the following OC problem of VOFIDE:

    MinimizeL(χ,υ)=Tf0f(τ,χ(τ),υ(τ)), (3.1)
    subject to {C0Dμ1(τ)τχ(τ)+0Iμ2(τ)τχ(τ)=g(τ,χ(τ),υ(τ)),χ(0)=χ0, (3.2)

    where χ0R is given, f:R×Rn×RmR, and g:R×Rn×RmR are continuous functions, χ(τ) and υ(τ) are the state and control variables, respectively, μ1,μ2:[0,Tf][0,1] are two given continuous functions, C0Dμ1(τ)τ is the type Ⅲ Caputo VOFD operator, and 0Iμ2(τ)τ is the left Riemann-Liouville fractional integral.

    In this section, we try to get an approximate solution to the optimal control problem (3.1)-(3.2). In subsection 4.1, we discuss the implementation of the method and present a new method for calculating the derivative and integral matrices. Also, we present a convergence analysis for the suggested method.

    For interpolating in the CPS method, the following Lagrange polynomials are utilized:

    hj(τ)=Ni=0ijττiτjτi=2NμjNn=0STn(τj)μnSTn(τ),j=0,1,2,,N,0τTf, (4.1)

    where

    μj={2, j=0,N,1, 1jN1,

    τj=Tf2(tj+1) and tj=cos(πjN),j=0,1,2,...,N, are roots of (1t2)˜TN(t).

    The Lagrange polynomials have the useful property of the delta Kronecker, i.e.,

    hj(τk)=δjk={1, j=k,0, jk. (4.2)

    Now, we approximate the variables of problem (3.1)-(3.2) in terms of the Lagrange functions as follows:

    χ(τ)χN(τ)=Nj=0ˉχjhj(τ). (4.3)

    Also, we have

    C0Dμ1(τ)τχ(τ)Nj=0ˉχjC0Dμ1(τ)τhj(τ),0Iμ2(τ)τχ(τ)Nj=0ˉχj0Iμ2(τ)τhj(τ). (4.4)

    Moreover, we approximate the control variable as

    υ(τ)υN(τ)=Nj=0ˉυjhj(τ), (4.5)

    where ˉχ=(ˉχ1,ˉχ2,...,ˉχN) and ˉυ=(ˉυ1,ˉυ2,...,ˉυN) are unknown coefficients. By applying the interpolation property of the Lagrange polynomial, we get

    χ(τk)ˉχk,υ(τk)ˉυk. (4.6)

    Now, by relations (4.1) and (4.3), we have

    χ(τ)χN(τ)=Nj=0ˉχj2NμjNn=0STn(τj)μnSTn(τ).

    Also, with the help of the Chebyshev polynomial analytical form shown in relation (2.18) in the previous section, we will have

    χ(τ)χN(τ)=Nj=0ˉχj2NμjNn=0STn(τj)μnnp=0(1)npn(n+p1)!22p(np)!(2p)!Tpfτp. (4.7)

    By applying Lemma 2.1 on the above relation, we reach the following relation, which is the result of the fractional derivative and integral effect on the power function:

    C0Dμ1(τ)τχ(τ)C0Dμ1(τ)τχN(τ)=Nj=0ˉχj2NμjNn=0STn(τj)μn(np=0(1)npn(n+p1)!22p(np)!(2p)!TpfΓ(p+1)Γ(p+1μ1(τ))τpμ1(τ)), (4.8)
    0Iμ2(τ)τχ(τ)0Iμ2(τ)τχN(τ)=Nj=0ˉχj2NμjNn=0STn(τj)μn(np=0(1)npn(n+p1)!22p(np)!(2p)!TpfΓ(p+1)Γ(p+1+μ2(τ))τp+μ2(τ)). (4.9)

    Lemma 4.1. Suppose q:[1,1]R is a continuous function. The following integral approximation is referred to as the Chebyshev-Gauss-Lobatto (CGL) quadrature rule:

    11q(τ)dτNj=0ˉωjq(τj), (4.10)

    where τj=cos(NjNπ),j=0,1,...,N are roots of (1τ2)ddτ˜TN(τ) and ˜TN(τ)=cos(Ncos1(τ)) is the Chebyshev polynomial of order N and ˉωj=1τ2jπ˜cjN,j=0,1,...,N are the quadrature weights of the numerical approximation (4.10), where ˜c0=˜cN=2,˜cj=1 for j=1,...,N1.

    Now, using relations (4.6), (4.8) and (4.9) and the above lemma, we reach the following discrete system, which can be solved by optimization methods:

    Minimize LN(ˉχ,ˉυ)=Nj=0ωjf(τj,ˉχj,ˉυj), (4.11)
    subject to {Nj=0ˉχj(D{μ1}kj+I{μ2}kj)=g(τk,ˉχk,ˉυk),k=1,2,...,N,ˉχ0=χ0, (4.12)

    where

    D{μ1}kj=2NμjNn=0STn(τj)μnnp=0(1)npn(n+p1)!22p(np)!(2p)!TpfΓ(p+1)Γ(p+1μ1(τk))τkpμ1(τk), (4.13)
    I{μ2}kj=2NμjNn=0STn(τj)μnnp=0(1)npn(n+p1)!22p(np)!(2p)!TpfΓ(p+1)Γ(p+1+μ2(τk))τkp+μ2(τk). (4.14)

    Here, ˉχ=(ˉχ0,ˉχ1,...,ˉχN) and ˉυ=(ˉυ0,ˉυ1,...,ˉυN) are unknown coefficients and μj=2 for j=0,N and μj=1 for j=1,2,...,N1. Also, ωj=Tf2~ωj. Note that there is a new technique to calculate operation matrices of variable-order fractional derivatives and integrals. First, we rewrite the Lagrange polynomial as follows:

    hj(τ)=Ni=0ij ττiτjτi=Np=0ηpj(ττ0)p,j=0,1,2,...,N,0τT,

    where ηpj are unknown coefficients that are determined as below. Using the delta Kronecker property (4.2), we get

    hj(τk)=Np=0ηpj(τkτ0)p=δkj,j=0,1,...,N,k=0,1,2,...,N.

    In matrix form, the above relation can be expressed as follows:

    Aηj=δj,j=0,1,2,...,N,

    where

    A=[10001(τ1τ0)(τ1τ0)N1τNτ0(τNτ0)N],ηj=[η0jη1jηNj],δj=[δj0δj1δjN].

    According to (4.2) and the invertibility of matrix A

    ηj=(A1)(j),ηpj=(A1)(p+1)(j+1),

    finally,

    hj(τ)=Np=0(A1)(p+1)(j+1)(ττ0)p. (4.15)

    Therefore, the components of the operational matrix of derivative can be obtained as follows:

    C0Dμ1(τ)τhj(τ)=Np=0(A1)(p+1)(j+1)C0Dμ1(τ)τ(ττ0)p=Np=0(A1)(p+1)(j+1)Γ(p+1)Γ(p+1μ1(τ))(ττ0)pμ1(τ), (4.16)

    and

    0Iμ2(τ)τhj(τ)=Np=0(A1)(p+1)(j+1)Γ(p+1)Γ(p+1+μ2(τ))(ττ0)p+μ2(τ). (4.17)

    According to (4.16) and (4.17), we get

    D{μ1}kj=Np=0(A1)(p+1)(j+1)Γ(p+1)Γ(p+1μ1(τk))(τkτ0)pμ1(τk), (4.18)

    and

    I{μ2}kj=Np=0(A1)(p+1)(j+1)Γ(p+1)Γ(p+1+μ2(τk))(τkτ0)p+μ2(τk). (4.19)

    Here, we show the convergence of the method by applying an assumption.

    We assume that the OC problem (3.1)-(3.2) has a Lagrange interpolating polynomial based on the shifted Chebyshev-Gauss-Lobatto (SCGL) points, which uniformly converges to it.

    Theorem 4.1. Assume that {(ˉχj,ˉvj)}Nj=0 is an optimal solution of (4.11)-(4.12) and define ˉXN(τ)=Nj=0ˉχjhj(τ) and ˉUN(τ)=Nj=0ˉυjhj(τ) on [0,Tf]. Also, assume {(ˉXN(.),ˉUN(.))}N=N0 uniformly converges to (ˉχ(.),ˉυ(.)) such that ˉχ(.) and ˉυ(.) are continuously differentiable and C0Dμ1(τ)τˉχ(.) and 0Iμ2(τ)τˉχ(.) are in C((0,Tf]). Then, (ˉX(.),ˉU(.)) is an optimal solution for the main OC problem of VOFIDE (3.1)-(3.2).

    Proof. We first show that (ˉX(.),ˉU(.)) is a feasible solution for the problem (3.1)-(3.2). Suppose that τ(0,Tf] is given. Since shifted CGL points {τk}Nk=0 with N is dense on [0,Tf], there exists a subsequence {τkj}j=0 such that limjkj= and limjτkj=τ. By continuity of functions g(.,.,.), C0Dμ1(τ)τˉχ(.) and 0Iμ2(τ)τˉχ(.), we get

    C0Dμ1(τ)τˉX(τ)+0Iμ2(τ)τˉX(τ)g(τ,ˉX(τ),ˉU(τ))=limNlimj(C0Dμ1(τkj)τˉXN(τkj)+0Iμ2(τ)τkjˉXN(τ)g(τkj,ˉXN(τkj),ˉUN(τkj)))=0.

    Also, for the initial condition,

    ˉX(0)X0=limN(ˉXN(0)X0)=0.

    Now, we want to show that (ˉX(.),ˉU(.)) is an optimal solution for the problem (3.1)-(3.2). By objective function (4.11), we define

    LN(ˉχ,ˉυ)=Nk=0ωkf(τk,ˉχk,ˉυk). (4.20)

    Also, by objective functional (3.1) and replacing continuous function q(.) in relation (4.10) with f(.,ˉX(.),ˉU(.)), we gain

    L(ˉX(.),ˉU(.))=Tf0f(τ,ˉX(τ),ˉU(τ))dτ=limNNk=0ωkf(τk,ˉX(τk),ˉU(τk)). (4.21)

    Moreover, since Nk=0ωk=Tf and (ˉXN(.),ˉUN(.)) is uniformly convergent to (ˉX(.),ˉU(.)), we get

    limNNk=0ωk(f(τk,ˉX(τk),ˉU(τk))f(τk,Nj=0ˉχjhj(τk),Nj=0ˉυjhj(τk)))L1limNNk=0ωk(ˉX(τk)Nj=0ˉχjhj(τk)+ˉU(τk)Nj=0ˉυjhj(τk))=L1limNNk=0ωk(ˉX(τk)ˉXN(τk)+ˉU(τk)ˉUN(τk))L1TflimN(ˉX(.)ˉXN(.)+ˉU(.)ˉUN(.))=0, (4.22)

    where L1>0 is the Lipschitz constant of continuously differentiable function f(.,.,.). Thus, by (4.20)–(4.22), we gain

    L(ˉX(.),ˉU(.))=Tf0f(τ,ˉX(τ),ˉU(τ))dτ=limN(Nk=0ωkf(τk,Nj=0ˉχjhj(τk),Nj=0ˉυjhj(τk))+Nk=0ωk[f(τk,ˉX(τk),ˉU(τk))f(τk,Nj=0ˉχjhj(τk),Nj=0ˉυjhj(τk))])=limNNk=0ωkf(τk,Nj=0ˉχjhj(τk),Nj=0ˉυjhj(τk))=limNLN(ˉχ,ˉυ).

    Hence,

    L(ˉX(.),ˉU(.))=limNLN(ˉχ,ˉυ). (4.23)

    On the other hand, for any optimal solution (X(.),U(.)) of problem (3.1)-(3.2), there exists a corresponding sequence {(χj,υj)}i=0 such that

    limNX(.)Nj=0χjhj(.)=limNU(.)Nj=0υjhj(.)=0.

    Since (X(.),U(.)) satisfies the constraint (3.2), sequence {(χj,υj)}Nj=0 with N satisfies constraint (4.12). Similar to the relation (4.23) and the process of achieving it, we can conclude

    L(X(.),U(.))=limNLN(χ,υ), (4.24)

    where χ=(χ0,χ1,...,χN) and υ=(υ0,υ1,...,υN). By relations (4.23) and (4.24), and optimality of pairs (ˉχ,ˉν) and (X(.),U(.)), we achieve

    L(X(.),U(.))L(ˉX(.),ˉU(τ))=limNLN(ˉχ,ˉν)limNLN(χ,υ)=L(X(.),U(.)), (4.25)

    which tends to L(X(.),U(.))=L(ˉX(.),ˉU(.)). Thus, (ˉX(.),ˉU(.)) is an optimal solution for the OC problem (3.1)-(3.2).

    In this section, some examples are shown to depict the efficiency and practicability of the devised approximation method. MATLAB has been used for all examples. The absolute errors are computed as

    Eχ(τ)=|χ(τ)χN(τ)|, (5.1)

    and

    Eυ(τ)=|υ(τ)υN(τ)|, (5.2)

    where pairs (χ,υ) and (χN,υN) are the exact and approximate solutions, respectively. We should also point out that the CPU time for program running in solving the discussed problems in all examples, requires less than 3 seconds for N = 5.

    Example 5.1. Consider the OC problem

    MinimizeL=10[(χ(τ)τ3)2+(υ(τ)τ1)2]dτ

    under the variable-order fractional dynamical system

    C0Dτμ1(τ)χ(τ)+0Iτμ2(τ)χ(τ)=Γ(4)χ(τ)[τμ1(τ)Γ(4μ1(τ))+τμ2(τ)Γ(4+μ2(τ))]+υ(τ)τ1,

    and the initial condition is given by χ(0)=0, so that the optimal solutions are

    χ(τ)=τ3,υ(τ)=τ+1,L=0.

    Now, we solve the above problem for the following orders:

    {μ11(τ)=10.4eτ,μ12(τ)=10.5eτ,{μ21(τ)=0.950.35sin(πτ),μ22(τ)=0.950.25sin(πτ),
    {μ31(τ)=0.75+0.2sin(10τ),μ32(τ)=0.75+0.2sin(50τ),{μ41(τ)=0.25+0.2τ2,μ42(τ)=0.25+0.5τ2.

    In Figure 1, we present the results obtained using the proposed technique as well as the exact solution for υ(τ) and χ(τ) when N=5. Using the presented technique, Figure 2 shows the absolute errors for υ(τ) and χ(τ) at various μi1(τ) and μi2(τ),i=1,2,3,4, and N=5. The approximate values of the performance index L with different μi1(τ) and μi2(τ),i=1,2,3,4, and N=5 are reported in Table 1. As the figures show, numerical solutions agree well with exact solutions when compared with the numerical results.

    Figure 1.  The result received for χ(τ) and υ(τ) with μ1(τ)=10.4exp(τ), μ2(τ)=10.5exp(τ) when N=5 in Example 5.1.
    Figure 2.  The absolute errors of χ(τ),υ(τ) for various orders μi1(τ) and μi2(τ), i=1,2,3,4 with N=5 in Example 5.1.
    Table 1.  The approximate values of L for μi1(τ), μi2(τ), i=1,2,3,4, where N=5 in Example 5.1.
    μ11(τ), μ12(τ) μ21(τ), μ22(τ) μ31(τ), μ32(τ) μ41(τ), μ42(τ)
    L 2.618331e-13 2.943066e-14 6.639050e-13 1.257254e-13

     | Show Table
    DownLoad: CSV

    Example 5.2. Consider the OC problem

    MinimizeL=10[(χ(τ)τ2)2+(υ(τ)+τ42τ2μ1(τ)Γ(3μ1(τ)))2]dτ,

    subject to the dynamical system

    C0Dτμ1(τ)χ(τ)+0Iτμ2(τ)χ(τ)=τ2χ(τ)+υ(τ)+2τ2+μ2(τ)Γ(3+μ2(τ)),χ(0)=0.

    The optimal value for the performance index is L=0, which is obtained by

    χ(τ)=τ2,υ(τ)=τ4+2τ2μ1(τ)Γ(3μ1(τ)).

    This problem is solved by the proposed method with N=5 for the following variable orders:

    {μ11(τ)=0.25+0.2sin(2πτ),μ12(τ)=0.25+0.2sin(πτ),{μ21(τ)=0.15+0.15|τ1|sin(τ),μ22(τ)=0.55+0.15|τ1|sin(τ),
    {μ31(τ)=10.67eτ,μ32(τ)=10.47eτ,{μ41(τ)=0.45τ,μ42(τ)=0.35τ.

    Figure 3 illustrates the behavior of the numerical solutions with N=5 for state and control variables χ(τ),υ(τ) for the above-mentioned μ11(τ),μ12(τ). The absolute errors obtained by the suggested method in the state variable and control variable are reported in Figure 4. Table 2 contains the performance index with N=5 for the variant (μi1(τ),μi2(τ)),i=1,2,3,4. In all these, it is evident that the proposed method yields numerical solutions that are highly accurate for all cases of orders of derivatives and integrals.

    Figure 3.  The result received for χ(τ) and υ(τ) with μ1(τ)=0.25+0.2sin(2πτ), μ2(τ)=0.25+0.2sin(πτ) when N=5 in Example 5.2.
    Figure 4.  The absolute errors of obtained χ(τ) and υ(τ) for various orders μi1(τ) and μi2(τ), i=1,2,3,4 with N=5 in Example 5.2.
    Table 2.  The approximate values of L for μi1(τ), μi2(τ), i=1,2,3,4, where N=5 for Example 5.2.
    μ11(τ), μ12(τ) μ21(τ), μ22(τ) μ31(τ), μ32(τ) μ41(τ), μ42(τ)
    L 6.467053e-15 4.948704e-13 1.587644e-13 4.199681e-15

     | Show Table
    DownLoad: CSV

    Example 5.3. Another OC problem of VOFIDE is as follows:

    MinimizeL=10[(χ(τ)τ2)2+(υ(τ)2τsin(τ))2]dτ,

    under the constraint

    C0Dτμ1(τ)χ(τ)+0Iτμ2(τ)χ(τ)=υ(τ)+2χ(τ)[τμ1(τ)Γ(3μ1(τ))+τμ2(τ)Γ(3+μ2(τ))]2τsin(τ),χ(0)=0. (5.3)

    The solutions χ(τ)=τ2 and υ(τ)=2τ+sin(τ) minimize the performance index L and L=0. Using the method described in this paper, we are able to solve this problem for different variable orders (μi1(τ),μi2(τ)),i=1,2,3,4. The fractional orders used are as follows:

    {μ11(τ)=0.15+0.25sin(τ),μ12(τ)=0.15+0.35sin(τ),{μ21(τ)=10.7e2τ,μ22(τ)=10.8e2τ,{μ31(τ)=0.2+0.7τ5,μ32(τ)=0.2+0.7τ8,{μ41(τ)=0.5+cos2(τ)eτ240,μ42(τ)=0.7+cos2(τ)eτ220. (5.4)

    Considering Table 3 and Figures 5 and 6, it is evident that the proposed method achieves numerical solution with high accuracy.

    Table 3.  The approximate values of L for μi1(τ), μi2(τ), i=1,2,3,4 where N=5 for Example 5.3.
    μ11(τ), μ12(τ) μ21(τ), μ22(τ) μ31(τ), μ32(τ) μ41(τ), μ42(τ)
    L 1.323914e-13 7.560300e-14 8.074596e-13 1.127018e-13

     | Show Table
    DownLoad: CSV
    Figure 5.  The result received for χ(τ) and υ(τ) with μ1(τ)=0.15+0.25sin(τ), μ2(τ)=0.15+0.35sin(τ) when N=5 in Example 5.3.
    Figure 6.  The absolute errors of χ(τ),υ(τ) for various μi1(τ), μi2(τ), i=1,2,3,4 with N=5 in Example 5.3.

    Example 5.4. In this example, we consider an optimal control problem of the energy for a fractional RLC series electrical circuit. We solve this problem by using our method. This problem can be formulated as the following fractional form (where R=1(Ω),L=1(H) and C=1(F)):

    MinimizeL=T0υ2(t)dt, (5.5)
    subjecttoC0Dμ1(t)tχ(t)+0Iμ2(t)tχ(t)=χ(t)+υ(t), (5.6)
    χ(0)=χ0,χ(T)=χT, (5.7)

    where state χ and control υ are the current and voltage in the RLC circuit, respectively. Note that if μ1(t)=μ2(t)=1, then Eq (5.6) can be written as the following equivalent form:

    ˙χ(t)+t0χ(τ)dτ=χ(t)+υ(t), (5.8)

    which is the Kirchhoff's voltage law. Note that the integro-differential equation (5.8) has been discussed and analyzed by many researchers, for example, see relation (8.3) in [15]. Also, in relation (29) in [15], the fractional form of (5.8), i.e., integro-differential equation (5.6), was introduced and studied. Some other fractional form of a RLC electrical circuit can be seen in relation (1.3) in [14] and relation (9) in [11]. In fact, fractional derivatives and integrals play an important role in the modeling of electrical circuits that contain super resistants, super capacitors, and super inductors [17]. Moreover, fractional models provide a more efficient description and representation of real electrical systems. However, the goal of solving the minimum energy problem (5.5)-(5.7), is to move an electrical initial current χ(0)=χ0 by a voltage in minimum energy to a desired final state χ(T)=χT. We assume χ0=1 (ampere), χT=0.25 (ampere) and T=1 (second) and solve problem (5.5)-(5.7) for different μ1(t) and μ2(t). The obtained optimal solutions are shown in Figures 7 and 8 for N=5. Also, the obtained minimum energy L for different cases is given in Tables 4 and 5.

    Figure 7.  The obtained optimal solutions for fixed derivative orders μ1 and μ2 in Example 5.4.
    Figure 8.  The obtained optimal solutions for variable derivative orders μ1(t) and μ2(t) in Example 5.4.
    Table 4.  The obtained approximate energy L for Example 5.4 where derivative orders are fixed.
    N=5 μ1=μ2=0.94 μ1=μ2=0.96 μ1=μ2=0.98 μ1=μ2=1
    L 0.094177 0.078091 0.060905 0.043829

     | Show Table
    DownLoad: CSV
    Table 5.  The obtained approximate energy L for Example 5.4 where derivative orders are variable.
    μ1=μ2
    N=5 0.95+0.01et 0.95+0.03et 0.95+0.05et 1
    L 0.080426 0.068278 0.055853 0.043829

     | Show Table
    DownLoad: CSV

    It can be seen that when μ1(t) and μ2(t) tend to 1, the obtained optimal state (current) and optimal control (voltage) go to the corresponding optimal solution with μ1(t)=μ2(t)=1. This issue, regarding the obtained optimal value for the performance index L, can also be seen in the tables, which confirms the correctness of proposed method to solve this practical problem. Further, we can conclude that in fractional RLC model (0<μ1,μ2<1), more total energy (that is, L) is needed to bring the current in the circuit from an initial state to the desired state in a certain time compared with the real RLC model (μ1=μ2=1).

    In this paper, we investigated an approach in order to solve a nonlinear optimal control problem involving variable-order fractional integro-differential equations as the dynamic system. Pseudo-spectral collocation is the basis of this method. At first, by using the expansion of Lagrange polynomials in terms of Chebyshev polynomials and the power series of them, the problem was converted into an nonlinear programming problem, which was easier to solve. Then, variable-order fractional derivatives in the Caputo sense were represented by a new operational matrix, and fractional integrals were represented by an operational matrix. With the suggested method, the optimal control problem of the variable-order fractional integro-differential equation could easily be solved. Using the numerical results, we could see that the approximate and exact solutions are in good agreement and the method is efficient and accurate as well.

    Zahra Pirouzeh, Mohammad Hadi Noori Skandari, Kamele Nassiri Pirbazari and Stanford Shateyi: Conceptualization, Methodology, Writing-review & editing, Software, Validation. All authors of this article have been contributed equally. All authors have read and approved the final version of the manuscript for publication.

    The authors declare they have not used Artificial Intelligence (AI) tools in the creation of this article.

    The authors declare no conflicts of interest.



    [1] M. A. Abdelkawy, R. T. Alqahtani, Shifted Jacobi collocation method for solving multi-dimensional fractional Stokes' first problem for a heated generalized second grade fluid, Adv. Differ. Equ., 2016 (2016), 1–17. https://doi.org/10.1186/s13662-016-0845-z doi: 10.1186/s13662-016-0845-z
    [2] R. Almeida, D. Tavares, D. F. M. Torres, The variable-order fractional calculus of variations, Springer, 2019.
    [3] A. Ansari, A. R. Sheikhani, H. S. Najafi, Solution to system of partial fractional differential equations using the fractional exponential operators, Math. Methods Appl. Sci., 35 (2012), 119–123. https://doi.org/10.1002/mma.1545 doi: 10.1002/mma.1545
    [4] D. Baleanu, O. G. Mustafa, R. P. Agarwal, An existence result for a superlinear fractional differential equation, Appl. Math. Lett., 23 (2010), 1129–1132. https://doi.org/10.1016/j.aml.2010.04.049 doi: 10.1016/j.aml.2010.04.049
    [5] A. H. Bhrawy, M. A. Zaky, Numerical algorithm for the variable-order Caputo fractional functional differential equation, Nonlinear Dyn., 85 (2016), 1815–1823. https://doi.org/10.1007/s11071-016-2797-y doi: 10.1007/s11071-016-2797-y
    [6] A. H. Bhrawy, J. F. Alzaidy, M. A. Abdelkawy, A. Biswas, Jacobi spectral collocation approximation for multi-dimensional time-fractional Schrödinger equations, Nonlinear Dyn., 84 (2016), 1553–1567. https://doi.org/10.1007/s11071-015-2588-x doi: 10.1007/s11071-015-2588-x
    [7] H. Dehestani, Y. Ordokhani, M. Razzaghi, Fractional-order Bessel wavelet functions for solving variable order fractional optimal control problems with estimation error, Int. J. Syst. Sci., 51 (2020), 1032–1052. https://doi.org/10.1080/00207721.2020.1746980 doi: 10.1080/00207721.2020.1746980
    [8] M. H. Heydari, Chebyshev cardinal functions for a new class of nonlinear optimal control problems generated by Atangana-Baleanu-Caputo variable-order fractional derivative, Chaos Solitons Fract., 130 (2020), 109401. https://doi.org/10.1016/j.chaos.2019.109401 doi: 10.1016/j.chaos.2019.109401
    [9] M. H. Heydari, Chebyshev cardinal wavelets for nonlinear variable-order fractional quadratic integral equations, Appl. Numer. Math., 144 (2019), 190–203. https://doi.org/10.1016/j.apnum.2019.04.019 doi: 10.1016/j.apnum.2019.04.019
    [10] M. H. Heydari, Z. Avazzadeh, A new wavelet method for variable-order fractional optimal control problems, Asian J. Control, 20 (2018), 1804–1817. https://doi.org/10.1002/asjc.1687 doi: 10.1002/asjc.1687
    [11] M. M. Khader, J. F. Gómez-Aguilar, M. Adel, Numerical study for the fractional RL, RC, and RLC electrical circuits using Legendre pseudo-spectral method, Int. J. Circuit Theory Appl., 49 (2021), 3266–3285. https://doi.org/10.1002/cta.3103 doi: 10.1002/cta.3103
    [12] F. K. Keshi, B. P. Moghaddam, A. Aghili, A numerical approach for solving a class of variable-order fractional functional integral equations, Comput. Appl. Math., 37 (2018), 4821–4834. https://doi.org/10.1007/s40314-018-0604-8 doi: 10.1007/s40314-018-0604-8
    [13] J. P. Liu, X. Li, L. M. Wu, An operational matrix technique for solving variable order fractional differential-integral equation based on the second kind of Chebyshev polynomials, Adv. Math. Phys., 2016 (2016), 6345978. https://doi.org/10.1155/2016/6345978 doi: 10.1155/2016/6345978
    [14] N. Magesh, A. Saravanan, Generalized differential transform method for solving RLC electric circuit of non-integer order, Nonlinear Eng., 7 (2018), 127–135. https://doi.org/10.1515/nleng-2017-0070 doi: 10.1515/nleng-2017-0070
    [15] S. V. Puscasu, S. M. Bibic, M. Rebenciuc, A. Toma, D. Ş. Nicolescu, Aspects of fractional calculus in RLC circuits, 2018 International Symposium on Fundamentals of Electrical Engineering (ISFEE), 2018, 1–5. https://doi.org/10.1109/ISFEE.2018.8742421
    [16] P. J. Torvik, R. L. Bagley, On the appearance of the fractional derivative in the behavior of real materials, J. Appl. Mech., 51 (1984), 294–298. https://doi.org/10.1115/1.3167615 doi: 10.1115/1.3167615
    [17] W. K. Zahra, M. M. Hikal, T. A. Bahnasy, Solutions of fractional order electrical circuits via Laplace transform and nonstandard finite difference method, J. Egyptian Math. Soc., 25 (2017), 252–261. https://doi.org/10.1016/j.joems.2017.01.007 doi: 10.1016/j.joems.2017.01.007
  • Reader Comments
  • © 2024 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Metrics

Article views(954) PDF downloads(50) Cited by(0)

Figures and Tables

Figures(8)  /  Tables(5)

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog