Adaptive neural network control for nonstrict-feedback uncertain nonlinear systems with input delay and asymmetric time-varying state constraints

https://doi.org/10.1016/j.jfranklin.2021.07.020Get rights and content

Abstract

This paper is devoted to adaptive neural network control issue for a class of nonstrict-feedback uncertain systems with input delay and asymmetric time-varying state constraints. State-related external disturbances are involved into the system, and the upper bounds of disturbances are assumed as functions of state variables instead of constants. Additionally, during the approximations of unknown functions by neural networks, the online computation burdens are declined sharply, since the norms of neural network weight vectors are only estimated. In the process of dealing with input delay, an auxiliary function is applied such that the conditions for time delay are more general than the ones in existing literature. A novel adaptive neural network controller is designed by constructing the asymmetric barrier Lyapunov function, which guarantees that the output of system has a good tracking performance and the state variables never violate the asymmetric time-varying constraints. Finally, numerical simulations are presented to verify the proposed adaptive control scheme.

Introduction

Over the past several decades, adaptive control has been successfully applied in many fields, such as bioengineering, machinery, navigation, communication. In recent years, adaptive control for nonlinear system has been paid much attention because of the development of adaptive control theory and designing method [1], [2], [3], [4], [5]. The authors in [2], [3], [4] studied adaptive backstepping control problem for a class of strict-feedback nonlinear system based on the ability of universal approximation of neural networks or fuzzy logic system. However, there are many practical systems that are described by the nonlinear systems with nonstrict-feedback form, and it is difficult to design the adaptive control scheme for these systems by applying the traditional backstepping method. To overcome this difficulty, some researchers improved the traditional backstepping method by using the boundedness of basis function of neural networks or fuzzy logic system, and the adaptive control schemes were designed for single-input single-output systems [6], [7], [8], [9], [10], [11], multi-input multi-output systems [12], [13] and interconnected large scale systems[14], [15] with nonstrict-feedback form.

For many real world applications, time delays are usually inevitable due to the finite transferring speed of signal, data or materials. Recently, the system with time delay has been an active research area because it may result in the instability or poor performances [16], [17], [18], [19], [20], [21]. Input delays, as a kind of time delays, usually arise from both physical nature of the plant and communication latencies [22]. In order to study the effect of input delay, a great number of control methods have been reported for linear systems [23], [24]. Actually, the most of practice systems are nonlinear dynamical systems and have inherent model uncertainties including external disturbance and model error. Therefore, the control schemes for uncertainty nonlinear system with input delay are more useful and realistic. As a result, adaptive control for nonlinear system with input delay has been paid much attention, such as [25], [26], [27], [28], [29], [30] and the references therein. In order to eliminate the effect of input delay, an auxiliary system is constructed in [25] and the Pade approximation method is adopted in [26], [27]. However, the input delay in these works was assumed to be constant, and small input delay was further required in the Pade approximation method. Therefore, the authors in [28], [29] introduced an auxiliary signal into the error systems to break through these limitations, while there are some rigid conditions for time-varying delay function. Besides, motivated by the auxiliary system in [25], Ma et al. [30] proposed a similar auxiliary system for the system with time-varying input delay to design the adaptive control scheme, but the Lyapunov function to deal with input delay is complicated, and the time-varying input delay and its derivative also need to satisfy some restrictive assumptions. Thus, how to construct a suitable Lyapunov function and design a suitable controller to relax these limitations is one of great challenges for the nonstrict-feedback nonlinear systems with time-varying input delay.

On the other hand, the state constraints are involved in many actual systems, such as the nonholonomic wheeled systems [31] and uncertain n-link robot [32]. In recent years, some excellent results concerning constrained systems were derived by a growing number of researchers [3], [28], [33], [34], [35], [36], [37]. The systems with symmetric and time-invariant state constraints (i.e., the state variables xi satisfy |xi|ki with positive constant ki) were considered in [3], [20] to design adaptive control schemes. As an extension, nonlinear systems with asymmetric and time-varying state constraints were presented in [34]. Meanwhile, the adaptive control problem for nonlinear systems with time-varying output constrained (as particular case of state constraints) was also investigated in [14], [28], [36]. However, all of systems in [3], [20], [28], [34], [36] are the ones with strict-feedback form. For time-varying state constrained system with nonstrict-feedback form, there is only a few results about adaptive control problem [38]. It is worth noting that input delay is overlooked in most of the above works. Although Deng et al. [28] studied adaptive control of uncertain nonlinear systems with input delay, the authors only considered the output constraints rather than full state constraints. In fact, it is difficult to design the adaptive control scheme for nonlinear system with both time-varying full state constraints and input delay, which is another challenge of this paper.

Motivated by the discussions above, in this paper, we study the adaptive control for a class of nonstrict-feedback uncertain systems with input delay and time-varying state constraints. Our main contributions lie into three aspects:

  • (i)

    The adaptive neural networks control scheme is designed for the nonlinear system with both input delay and time-varying state constraints in this paper, which is difficult because these two factors act on the last equation of our system at the same time. Compared with [28], [29], [30], some rigid conditions for time-varying delay function are released. Due to the using of radial basis function neural networks, some restrictive conditions for feedback gains in [28] are removed in this paper.

  • (ii)

    Different from [6], [12], [39], the time-varying disturbances with whole state-related upper bounds are involved in our system. A major limitation of the method in [11], [12] is the requirement of monotonic increase of upper bound functions for unknown nonlinear functions, but this limitation is deleted in this paper.

  • (iii)

    Compared with many existing works with time-varying state constraints, where the state constraints are regarded as known conditions during the stability analysis and it is unreasonable, we provide a more reasonable inference process, and obtain the conclusion that the state variables never violate the time-varying constraints under the designed control scheme.

The rest of this paper is organized as follows: Section 2 is devoted to formulating our problem and giving some useful assumptions and lemmas. The design procedure of adaptive controller and main theorem for adaptive scheme are presented in Section 3. The simulation results are obtained in Section 4. In Section 5, a brief discussion and summary concerning the main results are provided.

Notation: R represents the set of real numbers, N+ denotes the set of positive integers and Rn is Euclidean space with dimension nN+. For x=(x1,,xn)TRn, let x¯i=(x1,x2,,xi)T with 1in, and · denotes Eucludian norm of real vector. The notation of function is abbreviated sometimes, for example f(x) is denoted by f(·) or f. For function f(x), its i-th derivative is denoted by f(i)(x) for iN+.

Section snippets

Problem formulation and preliminaries

Consider the following nonstrict-feedback uncertain systems with input delay and asymmetric time-varying state constraints{x˙i=fi(x)+gi(x)xi+1+Di(x,t),x˙n=fn(x)+gn(x)u(tτ(t))+Dn(x,t),y=x1,i=1,,n1,where x=(x1,x2,,xn)TRn and yR are state variable and system output, respectively. fi(x) and gi(x)(1in) are the unknown smooth nonlinear functions, which satisfy fi(0)=0(i=1,,n) and [fi(x)+gi(x)xi+1+Di(x,t)]/xi+10(i=1,,n1). Di(x,t)(1in) are external disturbances. u(tτ(t))R denotes the

Controller design and stability analysis

This section is devoted to designing the controller for system (1), which is mainly based on the backstepping method with asymmetric barrier Lyapunov functions.

In order to measure the performance of output y(t), i.e., the distance between reference signal yd(t) and output, we first define the error systems as follows{z1(t)=x1(t)yd(t),zi(t)=xi(t)αi1,i=2,3,,n1,zn(t)=xn(t)αn1+Gzu(t),where αi(i=1,2,,n1) are virtual control functions that will be designed later, yd(t) denotes the reference

Numerical simulation

In this section, we illustrate the effectiveness of our designed control scheme by numerical simulations. A block diagram of the proposed control scheme is presented in Fig. 1.

Example 1

Consider the following nonlinear system{x˙1=0.11x1e0.6x1+x2sin(0.2x2)+(2+0.3sin(1.5x12))x2+0.1x1x2sin(0.5t),x˙2=x1x22+(3+0.2cosx1)uτ+0.1x2cos(0.5t),y=x1The state variables x1 and x2 are constrained by R̲i(t)xi(t)R¯i(t), where R̲1(t)=0.6+0.3sin(0.45t), R¯1(t)=2.2+0.5sint,R̲2(t)=0.8+0.1cost and R¯2(t)=0.3+0.2cost. The

Conclusion

In this paper, the issue of neural networks adaptive control has been studied for a class of nonstrict-feedback uncertain systems with input delay and time-varying state constraints. In order to solve this issue, the asymmetric time-varying barrier Lyapunov function has been introduced in the process of control design. By applying a critical result of neural networks (see Lemma 3), an adaptive control scheme has been obtained and its stability has also been proven. Our method relaxed the

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

This work was jointly supported by the National Natural Science Foundation of China under Grant No. 61673111, and the National Key Research and Development Program of China under Grant No. 2018AAA0100202.

References (46)

  • A. Molavi et al.

    Adaptive fuzzy control of a class of nonaffine nonlinear system with input saturation based on passivity theorem

    ISA Trans.

    (2017)
  • H. Wang et al.

    Adaptive neural tracking control for a class of nonlinear systems with dynamic uncertainties

    IEEE Trans. Cybern.

    (2017)
  • Z. Chen et al.

    Adaptive neural control of uncertain MIMO nonlinear systems with state and input constraints

    IEEE Trans. Neural Netw. Learn. Syst.

    (2017)
  • S. Tong et al.

    Observer-based adaptive fuzzy tracking control for strict-feedback nonlinear systems with unknown control gain functions

    IEEE Trans. Cybern.

    (2020)
  • H.A. Yousef et al.

    L1Adaptive fuzzy controller for a class of nonlinear systems with unknown backlash-like hysteresis

    Int. J. Syst. Sci.

    (2017)
  • B. Chen et al.

    Observer-based adaptive neural network control for nonlinear systems in nonstrict-feedback form

    IEEE Trans. Neural Netw. Learn. Syst.

    (2016)
  • S. Tong et al.

    Adaptive fuzzy tracking control design for SISO uncertain nonstrict feedback nonlinear systems

    IEEE Trans. Fuzzy Syst.

    (2016)
  • S. Tong et al.

    Adaptive fuzzy output feedback control for switched nonstrict-feedback nonlinear systems with input nonlinearities

    IEEE Trans. Fuzzy Syst.

    (2016)
  • B. Cao et al.

    Adaptive neural networks control for Markov jump nonlinear systems with nonstrict-feedback form and uncertain transition rate

    2020 39th Chinese Control Conference (CCC)

    (2020)
  • Y. Sun et al.

    Finite-time adaptive control for a class of nonlinear systems with nonstrict feedback structure

    IEEE Trans. Cybern.

    (2018)
  • H. Wang et al.

    Neural-based adaptive output-feedback control for a class of nonstrict-feedback stochastic nonlinear systems

    IEEE Trans. Cybern.

    (2015)
  • B. Chen et al.

    Adaptive fuzzy tracking control for a class of MIMO nonlinear systems in nonstrict-feedback form

    IEEE Trans. Cybern.

    (2015)
  • Y. Li et al.

    Adaptive fuzzy output constrained control design for multi-input multioutput stochastic nonstrict-feedback nonlinear systems

    IEEE Trans. Cybern.

    (2017)
  • Cited by (11)

    View all citing articles on Scopus
    View full text