
This article studied the locally exponential stability (LES) of time-delay systems subject to delayed impulses. Some Lyapunov-Razumikhin (L-R) theorems were presented, in which the information about the delays within the impulses was fully incorporated and then integrated into the stability analysis of the concerned systems. Our results highlight a critical finding: the delays in impulses can have dual effects on the stability of the systems, i.e., they may either destabilize the systems or contribute to the stability of the systems. Moreover, the effects of the nonlinear rate in discrete dynamics were fully considered, where a new relationship between the discrete dynamics, the continuous dynamics, and the initial region was established. As applications, several sufficient conditions that formulated in terms of linear matrix inequalities (LMIs) were obtained to ensure the stability of certain time-delay systems with highly nonlinear delayed impulses. To illustrate the applicability and effectiveness of the proposed results, two numerical examples were provided.
Citation: Hongwei Zheng, Yujuan Tian. Exponential stability of time-delay systems with highly nonlinear impulses involving delays[J]. Mathematical Modelling and Control, 2025, 5(1): 103-120. doi: 10.3934/mmc.2025008
[1] | Wei Zhao, Lei Liu, Yan-Jun Liu . Adaptive neural network control for nonlinear state constrained systems with unknown dead-zones input. AIMS Mathematics, 2020, 5(5): 4065-4084. doi: 10.3934/math.2020261 |
[2] | Mengmeng Jiang, Qiqi Ni . Adaptive state-feedback control for low-order stochastic nonlinear systems with an output constraint and SiISS inverse dynamics. AIMS Mathematics, 2025, 10(5): 11208-11233. doi: 10.3934/math.2025508 |
[3] | Yihang Kong, Xinghui Zhang, Yaxin Huang, Ancai Zhang, Jianlong Qiu . Prescribed-time adaptive stabilization of high-order stochastic nonlinear systems with unmodeled dynamics and time-varying powers. AIMS Mathematics, 2024, 9(10): 28447-28471. doi: 10.3934/math.20241380 |
[4] | Bin Hang, Weiwei Deng . Finite-time adaptive prescribed performance DSC for pure feedback nonlinear systems with input quantization and unmodeled dynamics. AIMS Mathematics, 2024, 9(3): 6803-6831. doi: 10.3934/math.2024332 |
[5] | Kunting Yu, Yongming Li . Adaptive fuzzy control for nonlinear systems with sampled data and time-varying input delay. AIMS Mathematics, 2020, 5(3): 2307-2325. doi: 10.3934/math.2020153 |
[6] | Lu Zhi, Jinxia Wu . Adaptive constraint control for nonlinear multi-agent systems with undirected graphs. AIMS Mathematics, 2021, 6(11): 12051-12064. doi: 10.3934/math.2021698 |
[7] | Lichao Feng, Mengyuan Dai, Nan Ji, Yingli Zhang, Liping Du . Prescribed-time stabilization of nonlinear systems with uncertainties/disturbances by improved time-varying feedback control. AIMS Mathematics, 2024, 9(9): 23859-23877. doi: 10.3934/math.20241159 |
[8] | Mohamed Kharrat, Moez Krichen, Loay Alkhalifa, Karim Gasmi . Neural networks-based adaptive command filter control for nonlinear systems with unknown backlash-like hysteresis and its application to single link robot manipulator. AIMS Mathematics, 2024, 9(1): 959-973. doi: 10.3934/math.2024048 |
[9] | Le You, Chuandong Li, Xiaoyu Zhang, Zhilong He . Edge event-triggered control and state-constraint impulsive consensus for nonlinear multi-agent systems. AIMS Mathematics, 2020, 5(5): 4151-4167. doi: 10.3934/math.2020266 |
[10] | Zaihua Xu, Jian Li . Adaptive prescribed performance control for wave equations with dynamic boundary and multiple parametric uncertainties. AIMS Mathematics, 2024, 9(2): 3019-3034. doi: 10.3934/math.2024148 |
This article studied the locally exponential stability (LES) of time-delay systems subject to delayed impulses. Some Lyapunov-Razumikhin (L-R) theorems were presented, in which the information about the delays within the impulses was fully incorporated and then integrated into the stability analysis of the concerned systems. Our results highlight a critical finding: the delays in impulses can have dual effects on the stability of the systems, i.e., they may either destabilize the systems or contribute to the stability of the systems. Moreover, the effects of the nonlinear rate in discrete dynamics were fully considered, where a new relationship between the discrete dynamics, the continuous dynamics, and the initial region was established. As applications, several sufficient conditions that formulated in terms of linear matrix inequalities (LMIs) were obtained to ensure the stability of certain time-delay systems with highly nonlinear delayed impulses. To illustrate the applicability and effectiveness of the proposed results, two numerical examples were provided.
The control problems of nonlinear systems have received a great deal of attention, and a considerable amount of literatures have been published [1,2,3,4,5]. A two-layer neural networks (NNs) based robust control for nonlinear induction motors is proposed in [1]. In [2] the unknown nonlinear items in the dynamic model of robotic manipulators are identified by NNs. Dynamic properties and optimal stabilization issues of fractional-order (FO) self-sustained electromechanical seismograph system is investigated in [3]. A neural adaptive control scheme is raised for a class of uncertain multi-input/multi-output nonlinear systems [4] and only one learning parameter is updated in the parameter identification. And coupled FO chaotic electromechanical devices are studied in [5]. Specifically, an adaptive dynamic programming policy is proposed to address the zero-sum differential game issue in the optimal neural feedback controller. Obviously, approximation-based adaptive backstepping control has been widely utilized in designing controllers for various nonlinear systems [6,7,8]. However, traditional approximation-based adaptive backstepping technique faces two crucial issues that hinder its application. The first issue is so-called "EOC" arising from the derivations of virtual control input [7]. The second one is a large computational burden caused by high precision approximation requirements [8].
For the backstepping technique, the design of the control law relies on intermediate state variables as virtual control signals. The controller of each subsystem requires the virtual control signal and its derivative. The lower-order derivatives of the virtual control signals are likely simple in theory, but the higher-order derivatives in higher-order systems are quite complex, which is called the "EOC" problem. For handling the "EOC" problem, a first-order filter was used to calculate the virtual control signal derivatives in each recursive step [9]. This technique is the so-called dynamic surface control (DSC). The utilization of the first-order filter overcomes the EOC problem, but its own characteristics lead to the derivative errors of virtual control signal, which will certainly affect the tracking performance of the system. Based on this, a modified scheme of the DSC method named command filter based control (CFC) method was proposed in [10]. On the one hand, the EOC problem is avoided by substituting a first-order filter with a second-order one to obtain the derivatives of the virtual control signal. On the other hand, the filter errors are compensated by a constructed error compensation mechanism for obtaining better tracking performance of the system. Moreover, constructing an effective error compensation mechanism needs to be further studied.
For the approximation-based control, the NNs or fuzzy logic systems (FLSs) are utilized to approximate unknown functions and external disturbances for ensuring better tracking performance [11,12], which is also a kind of learning control [13]. In [11] a NNs-based approximator is utilized to solve the unmodeled dynamics of the system. FLSs are employed to identify unknown functions existing in systems [12]. Iterative learning control schemes are designed to suppress the vibrations in bending and twisting of the flexible micro aerial vehicle [13]. The approximation accuracy improves with the increase of the number of neural network nodes or fuzzy rules in general, but it also significantly increases the number of estimated parameters. Therefore, the burdens of computation required for online learning will become very heavy. For decreasing the computational burden of approximation-based control, a tuning-function is inserted in the controller of strict-feedback systems [14,15], in which the number of parameter to be updated is the same as the number of unknown parameters. Recently, a kind of one-parameter estimation approach is proposed in [16,17], which needs only one parameter to be updated online and can significantly reduce the computational burden. Nevertheless, the mentioned control schemes do not take issues of input constraints, state constraints and prescribed performance into account. Therefore, for applying to a broader range of control problems with security, reliability and performance consideration in reality, further research is needed.
In the real world, many physical constraints are generated with security and reliability consideration, such as the output of MEMS resonator [18], the state constraint of aircraft engine [19] and the input constraint of magnetic-field electromechanical transducer [20]. Obviously, severe security matters, performance degradation, and other troubles can be caused without considering these constraints. For the issue of input constraint, a considerable amount of literatures have been published about it [21,22,23]. A non-smooth and piecewise input constraint model is described in [21]. Furthermore, the model in [22] is a smooth but piecewise function. The input constraint nonlinearity is tackled by asymmetric smooth input constraint model in [23]. For the issues of output constraint and state constraint, Barrier Lyapunov Function (BLF) is seen as an effective tool, and a significant number of typical works have been published including symmetric BLF [24,25] and asymmetric BLF [26,27]. However, the aforementioned BLF-based controllers have the following three drawbacks: 1) Discontinuous actions exist when constructing asymmetric BLF deals with asymmetric constraints; 2) Output/state constraints are achieved indirectly through error constraint, which leads to a more conservative initial output and state; 3) It is not allowed to handle both constrained and unconstrained systems without changing the controller structure. Although the integral BLF (IBLF)-based approach is possible to tackle output/state constraints directly [28], it can only overcome the disadvantages 1) and 2). By constructing a novel state transformation nonlinear function in [29,30], all those shortcomings can be overcome simultaneously. However, in practical applications, ensuring system security and reliability is the foundation, and achieving high performance control of the system is surely the ultimate goal. To be the best of our knowledge, no relevant results have been reported which can overcome the all above drawbacks and ensure safety reliability and high performance of systems simultaneously.
In this paper, with consideration of security, reliability and high performance, a LAFN-PPC of strict-feedback systems considering asymmetric full-state and input constraints is raised. In the controller design procedure, the constrained system is transformed into an unconstrained system using one-to-one nonlinear state transformation technique. One-to-one nonlinear error transformation technique is used to guarantee the prescribed performance. Furthermore, the unknown control direction and the input constraint nonlinearity are resolved by Nussbaum gain technique simultaneously. By introducing command filter and an error compensation mechanism, the constructed scheme can not only overcomes the so-called "EOC" problem, but also reduces filter errors. Moreover, the maximum values of the norm of optimal weight vector in FWNN is constructed as a virtual parameter, and the only one virtual parameter is estimated instead of the optimal weight vectors (OWVs). Regardless of the order of the system and the dimension of the system parameters, only one parameter is required to be updated online, which significantly reduces the computational burdens. The major contributions comparing with the existing ones are listed as:
1) In order to ensure the controlled systems with higher security, faster response speed and lower tracking error simultaneously, we combine a simple state transformation function with an error transformation function. All states and tracking error are always in symmetric or asymmetric prescribed bounds. Compared with the BLF-based methods [24,25,26,27], the LAFN-PPC can overcome all the three drawbacks, because we utilize the state transformation function instead of BLF, by which the constrained system is converted to an unconstrained system. In contrast to state transformation based methods [29,30], the tracking error is always remained within the prescribed performance bound by using an error transformation function.
2) By using command filtering to get the virtual control signal derivatives, the "EOC" problem of traditional backstepping method is overcome, the filter error caused by command filter is compensated by the carefully constructed error compensation mechanism. Compared with [1,2,3,4,5], the method we take only requires the reference signal and its first derivative, which greatly reduces the amount of calculation and meets many practical engineering requirements.
3) To significantly improve the computational efficiency of FWNN-based approximator and to replace estimating the OWVs in each step of backstepping, we construct the maximum value of the norm of OWVs in the FWNN as a virtual parameter. Only one virtual parameter needs to be estimated in the FWNN-based approximator, with this one-parameter estimation-based approach, the number of parameters updated online is independent of the order of the system and the dimension of OWVs, and the computational burden is significantly reduced, while the computational efficiency is significantly improved.
The considered strict-feedback systems with input constraint nonlinearity are given as
{˙xi=fi(¯xi,ℓi)+xi+1,˙xn=fn(¯xn,ℓn)+gu,y=x1i=1,…,n−1u=C(v) | (2.1) |
where ¯xi=[x1,…,xi]T∈Ri, i=1,…,n and v∈R are the states and the system input. y∈R denotes the system output. x=[x1,…,xn]T∈Rn are the whole states of the system, fi(ˉxi,ℓi), i=1,…,n are unknown smooth functions, Specifically, fi(ˉxi,ℓi) denote system uncertainties and external disturbances, ℓi are the unknown constant parameters inseparable from fi(ˉxi,ℓi). g denotes the unknown control gain. Here, u is the actual control signal which is subjected to the input constraint nonlinearity C(v):R→R. And the input constraint nonlinearity will be given later. In this paper, xi(t) are constrained in the open sets (−κ_i,¯κi), i.e., xi(t)∈Ωi.
As is known that input saturation of actuator is a common problem. The input saturation nonlinearity can seriously affect the safety and performance of the system. How to cope with the saturation nonlinearity has become an urgent and challenging research issue. In this paper, we take the input constraint nonlinearity into account. Mostly, the input constraint nonlinearity [21] can be expressed as
u=C(v)={u−,vi<u−,vi,u−≤vi≤u+,u+,vi>u+, | (2.2) |
where u+ and u− are the upper/lower bounds of u(t).
To simplify the design of the control, we can define Δ(v)=u−cvv, where cv is a positive constant. We can rewrite the input constraint model as:
u=cvv+Δ(v). | (2.3) |
Based on the above strict-feedback systems, the control object of this paper is to design a LAFN-PPC for system (2.1) to realize the following purposes:
(a) All signals of the system are in the sense of uniformly ultimate boundedness.
(b) The input signal and full state variables can be strictly restricted in asymmetric upper and lower bounds.
(c) The output signal can track the reference signal very well. And the output tracking error can be strictly restricted in upper and lower bounds.
The FWNN [20] has strong power in function approximation, which consists of a series of fuzzy IF-THEN rules as:
If Z1 is Mj1 and … Zn is Mjn, then ˆfs is ωj, j=1,…Ns, where Mji, i=1,…,ns, j=1,…,Ns, is jth member function for ith input, ns and Ns represent the number of inputs and rulers (fuzzy logical system), respectively.
The FWNN shown in Figure 1 consists of five lawyers, including an input layer, a fuzzification layer, a membership layer, a rule layer, and an output layer. The firing degrees of rulers are defined as
⌢ξi=∑nj=1(1+(Zi−cji)2ωji)e−(Zi−cji)2ωji, |
ξ⌣i=∑nj=1(1−(Zi−cji)2ωji)e−(Zi−cji)2ωji, |
where i=1,…,N, j=1,…,n, n and N denote the number of inputs and rulers (neural network system). cji and ωji represent the center and width of member function.
The firing degrees are defined as
ξi=ˉξi+ξ−iN∑i=1(ˉξi+ξ−i),i=1,…,N. |
The FWNN can be described as
f(θ,Z)=θTξ(Z)+ε(Z) | (2.4) |
where f(θ,Z) is a continuous function which is bounded in closed compact set ℧→Rn, Z=[Z1,Z2,⋯,ZN]∈℧⊂Rn is the input vector, n is the input dimension of neural network. θ=[θ1,θ2,⋯θN]T∈RN denotes the weight vector, and N>1 is the node number of neuron. and ξ(Z)=[ξ1(Z),ξ2(Z),⋯,ξN(Z)]T∈RN indicates the basic function vector. ε is the estimation error. And there is a positive constant εMi which satisfies |εi|⩽εMi, i=1,…,n.
Lemma 1 [20]. Continuous function f(Z) is defined on a compact set ℧. And for ε(Z)>0, there is a FWNN satisfying
supx∈Ω|f(Z)−ˆf(Z,θ)|≤ε(Z). |
The optimal parameter ⌢θ is equal to argminθ∈℧θ[supZ∈℧|f(Z)−ˆf(Z,θ)|], where ℧θ is a compact set, and ˜θ=⌢θ−θ, where ⌢θ represents the estimation of θ.
For resolving physical constraints generated with security and reliability consideration, the following state transformation function [29,30] is introduced to achieve asymmetric constraints, symmetric constraints and no constraints on states simultaneously in a unified form:
si(t)=κ_iˉκixi(t)(κ_i+xi(t))(ˉκi−xi(t)) | (2.5) |
where κ_i and ˉκi are positive constants, the initial state xi(0)∈Ωi, i=1,2,…,n. It is obvious that if the states xi(t)→−κ_i or xi(t)→ˉκi, the transformed states si(t)→±∞. Therefore, the state transformation function can constrain xi(t) within the open sets (−κ_i,ˉκi), i.e., xi(t)∈Ωi.
Remark 1. If asymmetric constraints need to be addressed, i.e., let κ_i≠ˉκi, i=1,2,…,n. Else symmetric constraints need to be tackled, i.e., let κi=κ_i=ˉκi. Then (2.5) becomes
si(t)=κ2ixi(t)κ2i−x2i(t). | (2.6) |
If no constraints need to be handled, let κi→+∞. It is clear that si(t)→xi(t), i.e.,
lim | (2.7) |
The above state transformation function can solve the control problems with asymmetric, symmetric and no state constraints in a unified form. Furthermore, it can handle above three kinds of control problems without changing adaptive laws. Figure 2 shows that the state transformation function can constrain {x_i}\left(t \right) within the open set \left({ - {{\underline{\kappa }}_{i}}, {{\bar \kappa }_i}} \right) (For the symmetric one: {{\underline{\kappa }}_{i}} = {\bar \kappa _i} = 4; For the asymmetric one: {{\underline{\kappa }}_{i}} = 4, {\bar \kappa _i} = 6).
The state transformation function can be rewritten as
{x_i}\left( t \right) = {\eta _i}{s_i}\left( t \right) | (2.8) |
where
{\eta _i} = \frac{{\left( {{{\underline{\kappa }}_{i}} + {x_i}\left( t \right)} \right)\left( {{{\bar \kappa }_i} - {x_i}\left( t \right)} \right)}}{{{{\underline{\kappa }}_{i}}\;{{\bar \kappa }_i}}} . |
Remark 2. {\eta _i} are also bounded in the sets {{{\Omega}} _{\eta i}}, i.e., 0 < {\eta _i} \leqslant {\bar \eta _i}, and {\bar \eta _i} = \frac{{{{\left({{{\underline{\kappa }}_{i}} + {{\bar \kappa }_i}} \right)}^2}}}{{4{{\underline{\kappa }}_{i}}{{\bar \kappa }_i}}}, i = 1, 2, \ldots, n.
Proof. See Appendix A. Taking the time derivative of the state transformation function (2.5):
{\dot{s}}_{i}\left(t\right) = {\rho }_{i}{\dot{x}}_{i}\left(t\right) | (2.9) |
where
{\rho _i} = \frac{{{{\underline{\kappa }}_{i}}\;{{\bar \kappa }_i}\left( {{{\underline{\kappa }}_{i}}\;{{\bar \kappa }_i} + x_i^2\left( t \right)} \right)}}{{{{\left( {{{\underline{\kappa }}_{i}} + {x_i}\left( t \right)} \right)}^2}{{\left( {{{\bar \kappa }_i} - {x_i}\left( t \right)} \right)}^2}}} . |
The constrained system is converted to an unconstrained system as
\left\{\begin{array}{cc}\begin{array}{l}{\dot{s}}_{i} = {\rho }_{i}{f}_{i}\left({\overline x }_{i}, {\ell}_{i}\right)+{\rho }_{i}{x}_{i+1}, \\ {\dot{s}}_{n} = {\rho }_{n}{f}_{n}\left({\overline x }_{n}, {\ell}_{n}\right)+{\rho }_{n}gu, \\ y = {\eta }_{1}{s}_{1}\end{array}& \begin{array}{l}i = 1, \dots , n-1\\ u = C\left(v\right)\\ \end{array}\end{array}\right. . | (2.10) |
For achieving the prescribed performance and guaranteeing the transformed output tracking error to converge within the prescribed performance bounds. Firstly, we define transformed output tracking error, virtual control errors and the command filters [10] as:
\left\{ \begin{array}{l} {z_1} = {s_1} - {\alpha _{1c}} \hfill \\ {e_i} = {s_i} - {\alpha _{ic}} \hfill \\ i = 2, 3, \ldots , n,\hfill \\ \end{array} \right. | (2.11) |
\left\{\begin{array}{l}{\dot{\alpha }}_{jc} = \omega {\alpha }_{jc, j}\\ {\dot{\alpha }}_{jc, j} = -2\varpi \omega {\alpha }_{jc, j}-\omega \left({\alpha }_{jc}-{\alpha }_{j-1}\right), \end{array}\right. | (2.12) |
where {z_1} is the transformed output tracking error, {e_i} are intermediate tracking errors, {\alpha _{ic}} are the outputs of the command filters, and the virtual control{\alpha _i} are the inputs of the command filters. \omega and \varpi \in \left({0, 1} \right) are positive design parameters. The initial values satisfy {\alpha _{jc}}\left(0 \right) = {\alpha _{j - 1}}\left(0 \right) and {\alpha _{jc, j}}\left(0 \right) = 0 for j = 2, 3, \ldots, n. {\alpha _{1c}} and its time derivative are calculated as:
\left\{\begin{array}{l}{\alpha }_{1c} = \frac{{\underline \kappa }_{1}\;{\overline \kappa }_{1}\;{y}_{r}}{\left({\underline \kappa }_{1}\;+\;{y}_{r}\right)\left({\overline \kappa }_{1}\;-\;{y}_{r}\right)}\\ {\dot{\alpha }}_{1c} = \frac{{\underline \kappa }_{1}{\overline \kappa }_{1}\left({\underline \kappa }_{1}{\overline \kappa }_{1}\;+\;{y}_{r}^{2}\right)}{{\left({\underline \kappa }_{1}\;+\;{y}_{r}\right)}^{2}{\left({\overline \kappa }_{1}\;-\;{y}_{r}\right)}^{2}}\;{\dot{y}}_{r} = {\rho }_{1r}{\dot{y}}_{r}\end{array}\right. | (2.13) |
where {y_r} is the reference signal.
To ensure the transformed output tracking error {z_1} strictly converges in the prescribed performance region during the whole time, we define
- \underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{k} \mu \left( t \right) < {z_1}\left( t \right) < \bar k\mu \left( t \right) , \forall t \geqslant 0, | (2.14) |
where the design parameters \underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{k} and \bar k are positive constants, and
\mu \left( t \right) = \left( {{\mu _0} - {\mu _\infty }} \right){e^{ - \hbar t}} + {\mu _\infty } |
is the prescribed performance function, \mu \left(0 \right) = {\mu _0}, 0 < {\mu _\infty } < {\mu _0}. The parameter \hbar is also a positive constant.
Remark 3. It is obvious that the transformed output tracking error is the output tracking error of unconstrained system and is restricted in the prescribed domain, i.e. {z_1}\left(t \right) \in \left({ - \underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{k} \mu \left(t \right), \bar k\mu \left(t \right)} \right). However, our control objective is that the output tracking error of original constrained system {e_r} = {x_1} - {y_r} is restricted in a prescribed domain. The output tracking error is bounded in a set
{e_r} \in \left( {{{ - \underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{k} \mu \left( t \right){{\left( {{{\underline{\kappa }}_{1}} + {{\bar \kappa }_1}} \right)}^2}} / {{{\underline{\kappa }}_{1}}{{\bar \kappa }_1}}}, {{\bar k\mu \left( t \right){{\left( {{{\underline{\kappa }}_{1}} + {{\bar \kappa }_1}} \right)}^2}} / {{{\underline{\kappa }}_{1}}{{\bar \kappa }_1}}}} \right). |
Proof. See Appendix B. The error transformation can be defined as:
{z}_{1}\left(t\right) = \mu \left(t\right)\Upsilon \left({\mathbb{Z}}_{1}\left(t\right)\right), \forall t\ge 0 | (2.15) |
where {\mathbb{Z}}_{1}\left(t\right) is the transformed error and \Upsilon \left({\mathbb{Z}}_{1}\right) is defined as:
\Upsilon \left({\mathbb{Z}}_{1}\right) = \frac{{\overline \kappa} {e}^{{\mathbb{Z}}_{1}}\;-\;{\underline \kappa} {e}^{-{\mathbb{Z}}_{1}}}{{e}^{{\mathbb{Z}}_{1}}\;+\;{e}^{-{\mathbb{Z}}_{1}}} . | (2.16) |
Remark 4. It is obvious that \Upsilon \left(t \right) is strictly constrained in symmetric or asymmetric domain \left({ - \underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{k}, \bar k} \right) showed in Figure 3 (For the symmetric one: \underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{k} = \bar k = 0.5. For the asymmetric one: \underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{k} = 0.5, \bar k = 0.7). It means that the error transformation function can deal with the issues considering symmetric or asymmetric exponential constraint simultaneously.
According to (2.15) and (2.16), one can obtain:
{\mathbb{Z}}_{1}\left(t\right) = {\Upsilon }^{-1}\left(\frac{{z}_{1}\left(t\right)}{\mu \left(t\right)}\right) = \frac{1}{2}\mathit{ln}\frac{\Upsilon +{\underline{ \kappa } }}{{\overline {\kappa} }-\Upsilon } | (2.17) |
and
{\dot{\mathbb{Z}}}_{1}\left(t\right) = \rho \left({\dot{z}}_{1}\left(t\right)-\frac{\dot{\mu }\left(t\right){z}_{1}\left(t\right)}{\mu \left(t\right)}\right) | (2.18) |
where
\rho = \frac{\left[\left(\frac{1}{\left(\Upsilon \left({\mathbb{Z}}_{1}\right)+{\underline \kappa }\right)}\right)+\left(\frac{1}{\left({\overline \kappa }-\Upsilon \left({\mathbb{Z}}_{1}\right)\right)}\right)\right]}{2}\mu \left(t\right) . |
We finally define the transformed tracking error as
{e}_{1} = {\mathbb{Z}}_{1}\left(t\right) . | (2.19) |
In order to reduce the errors caused by the command filters, we introduce compensation signals {\varsigma _i}, and the compensation errors can be defined as {v_i} = {e_i} - {\varsigma _i}, i = 1, \ldots, n.
FWNN is used to approximate the system uncertainties and external disturbances {f}_{i} :
{f}_{i} = {\theta }_{i}^{T}{\xi }_{i}\left(\mathbb{Z}\right)+{ \varepsilon }_{i}\left(\mathbb{Z}\right) . | (3.1) |
For facilitating adaptive law design, we define
\left\{ \begin{array}{l} {{\bar \theta }_i} = \max \left\{ {{{\left\| {{\theta _i}} \right\|}^2}, \varepsilon _{Mi}^2} \right\} \hfill \\ \theta = \max \left\{ {{{\bar \theta }_1}, \ldots , {{\bar \theta }_n}} \right\} \hfill \\ \end{array} \right. , i = 1, \ldots , n, | (3.2) |
where {\bar \theta _i} are unknown virtual parameters with {\bar \theta _i} < \theta . {\theta _i} is the ideal constant weight of ith NN and {\varepsilon _{Mi}} is the upper bound of approximation error {\varepsilon _i}.
Assumption 1. The reference trajectory {y_r} is continuous, and satisfies
{\left[{y}_{r}, {\dot{y}}_{r}, {\ddot{y}}_{r}\right]}^{T}\in \mathit{\Xi }_{r} , |
{\Xi _r} is a known compact set
\left\{{\left[{y}_{r}, {\dot{y}}_{r}, {\ddot{y}}_{r}\right]}^{T}:{y}_{d}^{2}+{\dot{y}}_{d}^{2}+{\ddot{y}}_{d}^{2}\le {B}_{r}\right\}\subset {R}^{3} . |
{B_r} is a known positive constant. Furthermore, - {\underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{\kappa } _1} < {y_r} < {\bar \kappa _1} holds.
Assumption 2. There are some constants \bar \Delta , \underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{g} and \bar g which satisfied \Delta \left(v \right) \leqslant \bar \Delta and \underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{g} \leqslant \left| g \right| \leqslant \bar g.
For solving the unknown gain and input constraint of systems simultaneously, Nussbaum function is introduced [10]. Obviously, a Nussbaum-type function N\left(\chi \right) needs to hold following the properties:
\left\{ \begin{array}{l} \mathop {\lim }\limits_{s \to + \infty } \sup \frac{1}{s}\int_0^s {N\left( \chi \right)d\chi } = + \infty \hfill \\ \mathop {\lim }\limits_{s \to + \infty } \inf \frac{1}{s}\int_0^s {N\left( \chi \right)d\chi } = - \infty. \hfill \\ \end{array} \right. | (3.3) |
In general, \ln \left({\chi + 1} \right)\cos \sqrt {\ln \left({\chi + 1} \right)} , {\chi ^2}\cos \left(\chi \right) and {e^{{\chi ^2}}}\cos \left({{{\pi \chi }/ 2}} \right) are commonly used Nussbaum-type function, and N\left(\chi \right) = {\chi ^2}\cos \left(\chi \right) is employed in this paper.
Lemma 2 [10]. Let V\left(\cdot \right) and \chi \left(\cdot \right) be smooth functions defined on \left[{0, {t_f}} \right) with V\left(t \right) \geqslant 0. \forall t \in \left[{0, {t_f}} \right), and N\left(\cdot \right) is a smooth Nussbaum-type function. If the following inequality holds:
V\left(t\right)\le {c}_{0}+{\int }_{0}^{t}\left({g}_{f}N\left(\chi \right)+1\right)\dot{\chi }d\tau | (3.4) |
where {g_f} is a non-zero parameter and {c_0} is an appropriate constant, then V\left(t \right) , \chi \left(t\right) and \int_0^t {\left({gN\left(\chi \right) + 1} \right)\dot \chi d\tau } must be bounded on \left[{0, {t_f}} \right).
Step 1. Taking the time derivative of {e_1} based on (2.17)–(2.19), it has:
{\dot{e}}_{1} = \rho \left({\rho }_{1}{x}_{2}+{\rho }_{1}{f}_{1}-{\rho }_{1r}{\dot{y}}_{r}-\frac{\dot{\mu }\left(t\right){z}_{1}\left(t\right)}{\mu \left(t\right)}\right) . | (3.5) |
The error compensation signal is constructed as
{\dot{\varsigma }}_{1} = -{k}_{1}{\varsigma }_{1}+\rho {\rho }_{1}{\eta }_{2}\left({\varsigma }_{2}+{\alpha }_{2c}-{\alpha }_{1}\right) . | (3.6) |
Choosing a Lyapunov function and taking its time derivative, one can obtain
\left\{\begin{array}{l}{V}_{1} = \frac{1}{2}{v}_{1}^{2}+\frac{1}{2\gamma }{\widetilde{\theta }}^{2}\\ {\dot{V}}_{1} = {v}_{1}\left(\rho \left({\rho }_{1}{\eta }_{2}\left({e}_{2}+{\alpha }_{2c}\right)+{\rho }_{1}{f}_{1}-{\rho }_{1r}{\dot{y}}_{r}-\frac{\dot{\mu }\left(t\right){z}_{1}\left(t\right)}{\mu \left(t\right)}\right)-{\dot{\varsigma }}_{1}\right)+\frac{1}{\gamma }\widetilde{\theta }\dot{{\overset\frown{\theta }}}\end{array}\right. | (3.7) |
where the compensation error {v_1} = {e_1} - {\varsigma _1}.
Using a FWNN to approximate the unknown item {f_1}, one can obtain:
{\dot{V}}_{1} = {v}_{1}\left(\rho \left({\rho }_{1}{\eta }_{2}\left({e}_{2}+{\alpha }_{2c}\right)+{\rho }_{1}\left({\theta }_{1}^{T}{\xi }_{1}+{ \varepsilon }_{1}\right)-{\rho }_{1r}{\dot{y}}_{r}-\frac{\dot{\mu }\left(t\right){z}_{1}\left(t\right)}{\mu \left(t\right)}\right)-{\dot{\varsigma }}_{1}\right)+\frac{1}{\gamma }\widetilde{\theta }\dot{{\overset\frown{\theta }}} . | (3.8) |
By the Young's inequality, it has
\left\{ \begin{array}{l} \rho {\rho _1}{v_1}\theta _1^T{\xi _1} \leqslant {\rho ^2}\rho _1^2v_1^2{\left\| {{\theta _1}} \right\|^2}{\left\| {{\xi _1}} \right\|^2} + \frac{1}{4} \hfill \\ \rho {\rho _1}{v_1}{\varepsilon _1} \leqslant {\rho ^2}\rho _1^2v_1^2\varepsilon _{M1}^2 + \frac{1}{4} .\hfill \\ \end{array} \right. | (3.9) |
Therefore, we have
\rho {\rho _1}{v_1}\left( {\theta _1^T{\xi _1} + {\varepsilon _1}} \right) \leqslant {\bar \theta _1}v_1^2{\zeta _1} + \frac{1}{2} \leqslant \theta v_1^2{\zeta _1} + \frac{1}{2} , | (3.10) |
with
{\bar \theta _1} = \max \left\{ {{{\left\| {{\theta _1}} \right\|}^2}, \varepsilon _{M1}^2} \right\}, | (3.11) |
{\zeta _1} = {\rho ^2}\rho _1^2{\left\| {{\xi _1}} \right\|^2} + {\rho ^2}\rho _1^2 > 0 . | (3.12) |
Hence, it has
{\dot{V}}_{1}\le {v}_{1}\left(\rho \left({\rho }_{1}{\eta }_{2}\left({e}_{2}+{\alpha }_{2c}\right)-{\rho }_{1r}{\dot{y}}_{r}-\frac{\dot{\mu }\left(t\right){z}_{1}\left(t\right)}{\mu \left(t\right)}\right)-{\dot{\varsigma }}_{1}\right)+\theta {v}_{1}^{2}{\zeta }_{1}+\frac{1}{2}+\frac{1}{\gamma }\widetilde{\theta }\dot{{\overset\frown{\theta }}} . | (3.13) |
By substituting compensation signal {\dot \varsigma _1} from (3.6) into (3.13), it has
{\dot{V}}_{1}\le {v}_{1}\left(\rho {\rho }_{1}{\eta }_{2}{v}_{2}-\rho {\rho }_{1r}{\dot{y}}_{r}-\rho \frac{\dot{\mu }\left(t\right){z}_{1}\left(t\right)}{\mu \left(t\right)}+{k}_{1}{\varsigma }_{1}+\rho {\rho }_{1}{\eta }_{2}{\alpha }_{1}\right)+\theta {v}_{1}^{2}{\zeta }_{1}+\frac{1}{2}+\frac{1}{\gamma }\widetilde{\theta }\dot{{\overset\frown{\theta }}} . | (3.14) |
Then, the virtual control law {\alpha _1} is designed as the following
{\alpha }_{1} = \left(-{k}_{1}{e}_{1}-{v}_{1}{\overset\frown{\theta }}{\zeta }_{1}+\rho {\rho }_{1r}{\dot{y}}_{r}+\rho \frac{\dot{\mu }\left(t\right){z}_{1}\left(t\right)}{\mu \left(t\right)}\right){\rho }^{-1}{\rho }_{1}^{-1}{\eta }_{2}^{-1}. | (3.15) |
Substituting the virtual control {\alpha _1} into (3.14) results in
{\dot{V}}_{1}\le -{k}_{1}{v}_{1}^{2}+{v}_{1}\left(\rho {\rho }_{1}{\eta }_{2}{v}_{2}-{v}_{1}{\overset\frown{\theta }}{\zeta }_{1}\right)+\theta {v}_{1}^{2}{\zeta }_{1}+\frac{1}{2}+\frac{1}{\gamma }\widetilde{\theta }\dot{\overset\frown{\theta }}. | (3.16) |
The adaptive law is given as
\dot{{\overset\frown{\theta }}} = \gamma {\sum }_{k = 1}^{n}{v}_{k}^{2}{\zeta }_{k}-\sigma \overset\frown{\theta }. | (3.17) |
By integrating (3.16) and (3.17), we have
{\dot{V}}_{1}\le -{k}_{1}{v}_{1}^{2}+{v}_{1}\rho {\rho }_{1}{\eta }_{2}{v}_{2}+\frac{1}{2}+\widetilde{\theta }{\sum }_{k = 2}^{n}{v}_{k}^{2}{\zeta }_{k}-\frac{\sigma }{\gamma }\widetilde{\theta }\overset\frown{\theta }. | (3.18) |
Note that
-\frac{\sigma }{\gamma }\widetilde{\theta }{\overset\frown{\theta }}\le -\frac{\sigma }{2\gamma }{\widetilde{\theta }}^{2}+\frac{\sigma }{2\gamma }{\theta }^{2}. | (3.19) |
One has
{\dot{V}}_{1}\le -{k}_{1}{v}_{1}^{2}-\frac{\sigma }{2\gamma }{\widetilde{\theta }}^{2}+{v}_{1}\rho {\rho }_{1}{\eta }_{2}{v}_{2}+\widetilde{\theta }{\sum }_{k = 2}^{n}{v}_{k}^{2}{\zeta }_{k}+\frac{\sigma }{2\gamma }{\theta }^{2}+\frac{1}{2}. | (3.20) |
Step 2. Taking the time derivative of {e_2}, it has:
{\dot{e}}_{2} = {\rho }_{2}{\dot{x}}_{2}-{\dot{\alpha }}_{2c} = {\rho }_{2}{x}_{3}+{\rho }_{2}{f}_{2}-{\dot{\alpha }}_{2c}. | (3.21) |
The error compensation signal is constructed as
{\dot{\varsigma }}_{2} = -{k}_{2}{\varsigma }_{2}-\rho {\rho }_{1}{\eta }_{2}{\varsigma }_{1}+{\rho }_{2}{\eta }_{3}\left({\varsigma }_{3}+{\alpha }_{3c}-{\alpha }_{2}\right). | (3.22) |
Choosing a Lyapunov function and taking its time derivative, one can obtain
\left\{\begin{array}{l}{V}_{2} = {V}_{1}+\frac{1}{2}{v}_{2}^{2}\\ {\dot{V}}_{2} = {\dot{V}}_{1}+{v}_{2}\left({\rho }_{2}{\eta }_{3}\left({e}_{3}+{\alpha }_{3c}\right)+{\rho }_{2}{f}_{2}-{\dot{\alpha }}_{2c}-{\dot{\varsigma }}_{2}\right)\end{array}\right. | (3.23) |
where the compensation error {v_2} = {e_2} - {\varsigma _2}.
By integrating compensation signal {\dot{\varsigma }}_{2} and (3.23), it has
{\dot{V}}_{2}\le -{k}_{1}{v}_{1}^{2}-\frac{\sigma }{2\gamma }{\widetilde{\theta }}^{2}+{v}_{2}\left(\begin{array}{l}{\rho }_{2}{\eta }_{3}{v}_{3}+{\rho }_{2}{f}_{2}+\rho {\rho }_{1}{\eta }_{2}{e}_{1}\\ -{\dot{\alpha }}_{2c}+{k}_{2}{\varsigma }_{2}+{\rho }_{2}{\eta }_{3}{\alpha }_{2}\end{array}\right)+\widetilde{\theta }{\sum }_{k = 2}^{n}{v}_{k}^{2}{\zeta }_{k}+\frac{\sigma }{2\gamma }{\theta }^{2}+\frac{1}{2}. | (3.24) |
Using a FWNN to approximate the unknown item {f_2}, one can obtain:
{\dot{V}}_{2}\le -{k}_{1}{v}_{1}^{2}-\frac{\sigma }{2\gamma }{\widetilde{\theta }}^{2}+\widetilde{\theta }\sum \limits_{k = 2}^{n}{v}_{k}^{2}{\zeta }_{k}+\frac{\sigma }{2\gamma }{\theta }^{2}+\frac{1}{2}\\ \;\;\;\;+{v}_{2}\left(\begin{array}{l}{\rho }_{2}{\eta }_{3}{v}_{3}+{\rho }_{2}\left({\theta }_{2}^{T}{\xi }_{2}+{ \varepsilon }_{2}\right)\\ +\rho {\rho }_{1}{\eta }_{2}{e}_{1}-{\dot{\alpha }}_{2c}+{k}_{2}{\varsigma }_{2}+{\rho }_{2}{\eta }_{3}{\alpha }_{2}\end{array}\right) . | (3.25) |
By the Young's inequality, it has
\left\{ \begin{array}{l} {\rho _2}{v_2}\theta _2^T{\xi _2} \leqslant \rho _2^2v_2^2{\left\| {{\theta _2}} \right\|^2}{\left\| {{\xi _2}} \right\|^2} + \frac{1}{4} \hfill \\ {\rho _2}{v_2}{\varepsilon _2} \leqslant \rho _2^2v_2^2\varepsilon _{M2}^2 + \frac{1}{4} \hfill \\ \end{array} \right. . | (3.26) |
Therefore, we have
{\rho _2}{v_2}\left( {\theta _2^T{\xi _2} + {\varepsilon _2}} \right) \leqslant {\bar \theta _2}v_2^2{\zeta _2} + \frac{1}{2} \leqslant \theta v_2^2{\zeta _2} + \frac{1}{2} , | (3.27) |
with
{\bar \theta _2} = \max \left\{ {{{\left\| {{\theta _2}} \right\|}^2}, \varepsilon _{M2}^2} \right\}, | (3.28) |
{\zeta _2} = \rho _2^2{\left\| {{\xi _2}} \right\|^2} + \rho _2^2 > 0 . | (3.29) |
Hence, it has
{\dot{V}}_{2}\le -{k}_{1}{v}_{1}^{2}-\frac{\sigma }{2\gamma }{\widetilde{\theta }}^{2}+\widetilde{\theta }\sum \limits_{k = 2}^{n}{v}_{k}^{2}{\zeta }_{k}+\frac{\sigma }{2\gamma }{\theta }^{2}+\frac{1}{2} \\ \;\;\;\;+{v}_{2}\left(\begin{array}{l}{\rho }_{2}{\eta }_{3}{v}_{3}+\rho {\rho }_{1}{\eta }_{2}{e}_{1}\\ -{\dot{\alpha }}_{2c}+{k}_{2}{\varsigma }_{2}+{\rho }_{2}{\eta }_{3}{\alpha }_{2}\end{array}\right)+\theta {v}_{2}^{2}{\zeta }_{2}+\frac{1}{2} . | (3.30) |
Then, the virtual control law {\alpha _2} is designed as the following
{\alpha }_{2} = \left(-{k}_{2}{e}_{2}-{v}_{2}{\overset\frown{\theta }}{\zeta }_{2}-\rho {\rho }_{1}{\eta }_{2}{e}_{1}+{\dot{\alpha }}_{2c}\right){\rho }_{2}^{-1}{\eta }_{3}^{-1}. | (3.31) |
Substituting the virtual control {\alpha _2} into (3.30) results in
{\dot{V}}_{2}\le -{k}_{1}{v}_{1}^{2}-{k}_{2}{v}_{2}^{2}-\frac{\sigma }{2\gamma }{\widetilde{\theta }}^{2}+{v}_{2}{\rho }_{2}{\eta }_{3}{v}_{3}+\widetilde{\theta }{\sum }_{k = 3}^{n}{v}_{k}^{2}{\zeta }_{k}+\frac{\sigma }{2\gamma }{\theta }^{2}+\frac{2}{2}. | (3.32) |
Step i. Taking the time derivative of {e_i}, it has:
{\dot{e}}_{i} = {\rho }_{i}{\dot{x}}_{i}-{\dot{\alpha }}_{ic} = {\rho }_{i}{x}_{i+1}-{\dot{\alpha }}_{ic}. | (3.33) |
The error compensation signal is constructed as
{\dot{\varsigma }}_{i} = -{k}_{i}{\varsigma }_{i}-{\rho }_{i-1}{\eta }_{i}{\varsigma }_{i-1}+{\rho }_{i}{\eta }_{i+1}\left({\varsigma }_{i+1}+{\alpha }_{i+1c}-{\alpha }_{i}\right). | (3.34) |
Choosing a Lyapunov function and taking its time derivative, one can obtain
\left\{\begin{array}{l}{V}_{i} = {V}_{i-1}+\frac{1}{2}{v}_{i}^{2}\\ {\dot{V}}_{i} = {\dot{V}}_{i-1}+{v}_{i}\left({\rho }_{i}{\eta }_{i+1}\left({e}_{i+1}+{\alpha }_{i+1c}\right)+{\rho }_{i}{f}_{i}-{\dot{\alpha }}_{ic}-{\dot{\varsigma }}_{i}\right)\end{array}\right. | (3.35) |
where the compensation error {v_i} = {e_i} - {\varsigma _i}.
By integrating compensation signal {\dot{\varsigma }}_{i} and (3.35), it has
{\dot{V}}_{i}\le -\sum\limits _{k = 1}^{i-1}{k}_{k}{v}_{k}^{2}-\frac{\sigma }{2\gamma }{\widetilde{\theta }}^{2}+\widetilde{\theta }\sum \limits_{k = i}^{n}{v}_{k}^{2}{\zeta }_{k}+\frac{\sigma }{2\gamma }{\theta }^{2}+\frac{i-1}{2}\\ \;\;\;\;+{v}_{i}\left(\begin{array}{l}{\rho }_{i}{\eta }_{i+1}{v}_{i+1}+{\rho }_{i}{f}_{i}-{\dot{\alpha }}_{ic}\\ +{k}_{i}{\varsigma }_{i}+{\rho }_{i-1}{\eta }_{i}{e}_{i-1}+{\rho }_{i}{\eta }_{i+1}{\alpha }_{i}\end{array}\right). | (3.36) |
Using a FWNN to approximate the unknown item {f_i}, one can obtain:
{\dot{V}}_{i}\le -\sum \limits_{k = 1}^{i-1}{k}_{k}{v}_{k}^{2}-\frac{\sigma }{2\gamma }{\widetilde{\theta }}^{2}+\widetilde{\theta }\sum \limits_{k = i}^{n}{v}_{k}^{2}{\zeta }_{k}+\frac{\sigma }{2\gamma }{\theta }^{2}+\frac{i-1}{2} \\ \;\;\;\;+{v}_{i}\left(\begin{array}{l}{\rho }_{i}{\eta }_{i+1}{v}_{i+1}+{\rho }_{i}\left({\theta }_{i}^{T}{\xi }_{i}+{ \varepsilon }_{i}\right)\\ -{\dot{\alpha }}_{ic}+{k}_{i}{\varsigma }_{i}+{\rho }_{i-1}{\eta }_{i}{e}_{i-1}\\ +{\rho }_{i}{\eta }_{i+1}{\alpha }_{i}\end{array}\right). | (3.37) |
By the Young's inequality, it has
\left\{ \begin{array}{l} {\rho _i}{v_i}\theta _i^T{\xi _i} \leqslant \rho _i^2v_i^2{\left\| {{\theta _i}} \right\|^2}{\left\| {{\xi _i}} \right\|^2} + \frac{1}{4} \hfill \\ {\rho _i}{v_i}{\varepsilon _i} \leqslant \rho _i^2v_i^2\varepsilon _{Mi}^2 + \frac{1}{4} \hfill \\ \end{array} \right. . | (3.38) |
Therefore, we have
{\rho _i}{v_i}\left( {\theta _i^T{\xi _i} + {\varepsilon _i}} \right) \leqslant {\bar \theta _i}v_i^2{\zeta _i} + \frac{1}{2} \leqslant \theta v_i^2{\zeta _i} + \frac{1}{2} , | (3.39) |
with
{\bar \theta _i} = \max \left\{ {{{\left\| {{\theta _i}} \right\|}^2}, \varepsilon _{Mi}^2} \right\}, | (3.40) |
{\zeta _i} = \rho _i^2{\left\| {{\xi _i}} \right\|^2} + \rho _i^2 > 0 . | (3.41) |
Hence, it has
{\dot V_i} \leqslant - \sum\limits_{k = 1}^{i - 1} {{k_k}v_k^2} - \frac{\sigma }{{2\gamma }}{\tilde \theta ^2} + \tilde \theta \sum\limits_{k = i}^n {v_k^2{\zeta _k}} + \frac{\sigma }{{2\gamma }}{\theta ^2} + \frac{{i - 1}}{2} + {v_i}\left( \begin{gathered} {\rho _i}{\eta _{i + 1}}{v_{i + 1}} - {{\dot \alpha }_{ic}} \hfill \\ + {k_i}{\varsigma _i} + {\rho _{i - 1}}{\eta _i}{e_{i - 1}} \hfill \\ + {\rho _i}{\eta _{i + 1}}{\alpha _i} \hfill \\ \end{gathered} \right) + \theta v_i^2{\zeta _i} + \frac{1}{2} . | (3.42) |
Then, the virtual control law {\alpha _i} is designed as the following
{\alpha }_{i} = \left(-{k}_{i}{e}_{i}-{v}_{i}{\overset\frown{\theta }}{\zeta }_{i}-{\rho }_{i-1}{\eta }_{i}{e}_{i-1}+{\dot{\alpha }}_{ic}\right){\rho }_{i}^{-1}{\eta }_{i+1}^{-1}. | (3.43) |
Substituting the virtual control {\alpha _i} into (3.42) results in
{\dot{V}}_{i}\le -{\sum }_{k = 1}^{i}{k}_{k}{v}_{k}^{2}-\frac{\sigma }{2\gamma }{\widetilde{\theta }}^{2}+\widetilde{\theta }{\sum }_{k = i+1}^{n}{v}_{k}^{2}{\zeta }_{k}+\frac{\sigma }{2\gamma }{\theta }^{2}+\frac{i}{2}+{v}_{i}{\rho }_{i}{\eta }_{i+1}{v}_{i+1}. | (3.44) |
Step n. Taking the time derivative of {e_n}, it has:
{\dot{e}}_{n} = {\rho }_{n}{\dot{x}}_{n}-{\dot{\alpha }}_{nc} = {\rho }_{n}gu+{\rho }_{n}{f}_{n}-{\dot{\alpha }}_{nc}. | (3.45) |
The error compensation signal is constructed as
{\dot{\varsigma }}_{n} = -{k}_{n}{\varsigma }_{n}-{\rho }_{n-1}{\eta }_{n}{\varsigma }_{n-1}. | (3.46) |
Choosing a Lyapunov function and taking its time derivative, one can obtain
\left\{\begin{array}{l}{V}_{n} = {V}_{n-1}+\frac{1}{2}{v}_{n}^{2}\\ {\dot{V}}_{n} = {\dot{V}}_{n-1}+{v}_{n}\left({\rho }_{n}gu+{\rho }_{n}{f}_{n}-{\dot{\alpha }}_{nc}-{\dot{\varsigma }}_{n}\right)\end{array}\right. | (3.47) |
where the compensation error {v_n} = {e_n} - {\varsigma _n}.
By integrating compensation signal {\dot{\varsigma }}_{n} and (3.47), it has
{\dot{V}}_{n} = {\dot{V}}_{n-1}+{v}_{n}\left({\rho }_{n}gu+{\rho }_{n}{f}_{n}-{\dot{\alpha }}_{nc}+{k}_{n}{\varsigma }_{n}+{\rho }_{n-1}{\eta }_{n}{\varsigma }_{n-1}\right). | (3.48) |
Using a FWNN to approximate the unknown item {f_n}, one can obtain:
{\dot{V}}_{n} = {\dot{V}}_{n-1}+{v}_{n}\left({\rho }_{n}gu+{\rho }_{n}\left({\theta }_{n}^{T}{\xi }_{n}+{ \varepsilon }_{n}\right)-{\dot{\alpha }}_{nc}+{k}_{n}{\varsigma }_{n}+{\rho }_{n-1}{\eta }_{n}{\varsigma }_{n-1}\right). | (3.49) |
By the Young's inequality, it has
\left\{ \begin{array}{l} {\rho _n}{v_n}\theta _n^T{\xi _n} \leqslant \rho _n^2v_n^2{\left\| {{\theta _n}} \right\|^2}{\left\| {{\xi _n}} \right\|^2} + \frac{1}{4} \hfill \\ {\rho _n}{v_n}{\varepsilon _n} \leqslant \rho _n^2v_n^2\varepsilon _{Mn}^2 + \frac{1}{4} \hfill \\ \end{array} \right. . | (3.50) |
Therefore, we have
{\rho _n}{v_n}\left( {\theta _n^T{\xi _n} + {\varepsilon _n}} \right) \leqslant {\bar \theta _n}v_n^2{\zeta _n} + \frac{1}{2} \leqslant \theta v_n^2{\zeta _n} + \frac{1}{2} | (3.51) |
with
{\bar \theta _n} = \max \left\{ {{{\left\| {{\theta _n}} \right\|}^2}, \varepsilon _{Mn}^2} \right\}, | (3.52) |
{\zeta _n} = \rho _n^2{\left\| {{\xi _n}} \right\|^2} + \rho _n^2 > 0 . | (3.53) |
Hence, it has
{\dot{V}}_{n}\le {\dot{V}}_{n-1}+{v}_{n}\left({\rho }_{n}g{c}_{v}v+\frac{1}{2}{\rho }_{n}^{2}{v}_{n}-{\dot{\alpha }}_{nc}+{k}_{n}{\varsigma }_{n}+{\rho }_{n-1}{\eta }_{n}{\varsigma }_{n-1}\right)+\theta {v}_{n}^{2}{\zeta }_{n}+\frac{1}{2}+\frac{1}{2}{{\overline {g} }}^{2}{{\overline {\Delta} }}^{2}. | (3.54) |
Then, the control input v is designed as the following
v = N\left(\chi \right)\psi {\rho }_{n}^{-1}, | (3.55) |
\psi = {k}_{n}{e}_{n}+{v}_{n}{\overset\frown{\theta }}{\zeta }_{n}+{\rho }_{n-1}{\eta }_{n}{e}_{n-1}-{\dot{\alpha }}_{nc}+\frac{1}{2}{\rho }_{n}^{2}{v}_{n}, | (3.56) |
\dot{\chi } = {v}_{n}\psi . | (3.57) |
Substituting the virtual control v into (3.54) results in
{\dot{V}}_{n}\le -{\sum }_{k = 1}^{n}{k}_{k}{v}_{k}^{2}+g{c}_{v}N\left(\chi \right)\dot{\chi }+\dot{\chi }-\frac{\sigma }{2\gamma }{\widetilde{\theta }}^{2}+\frac{\sigma }{2\gamma }{\theta }^{2}+\frac{1}{2}{{\overline {g} }}^{2}{{\overline {\Delta} }}^{2}+\frac{n}{2}. | (3.58) |
Up to now, the whole construction of LAFN-PPC is completed.
For any given positive constant p, consider a closed set
\left\{\begin{array}{l}{\Theta }_{1} = \left\{\left({v}_{1}, {\overset\frown{\theta }}\right):{v}_{1}^{2}+\frac{1}{\gamma }{\widetilde{\theta }}^{2}\le 2p\right\}\\ {\Theta }_{2} = \left\{\left({v}_{1}, {v}_{2}, {\overset\frown{\theta }}\right):{\sum }_{k = 1}^{2}{v}_{k}^{2}+\frac{1}{\gamma }{\widetilde{\theta }}^{2}\le 2p\right\}\\ {\Theta }_{i} = \left\{\left({v}_{1}, {v}_{2}, \dots , {v}_{i}, {\overset\frown{\theta }}\right):{\sum }_{k = 1}^{i}{v}_{k}^{2}+\frac{1}{\gamma }{\widetilde{\theta }}^{2}\le 2p\right\}\\ {\Theta }_{n} = \left\{\left({v}_{1}, {v}_{2}, \dots , {v}_{n}, {\overset\frown{\theta }}\right):{\sum }_{k = 1}^{n}{v}_{k}^{2}+\frac{1}{\gamma }{\widetilde{\theta }}^{2}\le 2p\right\}\end{array}\right. . | (4.1) |
Theorem 1. For the strict-feedback systems (2.1) with full-state and input constraints under Assumptions 1 and 2, the controllers (3.15), (3.31), (3.43) and (3.55)–(3.57), adaptive law (3.17) and compensation signal (3.6), (3.22), (3.34) and (3.46) are constructed. If initial conditions satisfy {\Theta _i}, i = 1, 2, \ldots, n, {x_i}(0) \in \left({ - {{\underline{\kappa }}_{i}}, {{\bar \kappa }_i}} \right) , i = 1, 2, \ldots, n, and {y_r}(0) \in \left({ - {{\underline{\kappa }}_{1}}, {{\bar \kappa }_1}} \right), then the proposed control scheme ensures the achievement of objectives (a)–(c).
Proof. The Lyapunov function choosing as
V = \frac{1}{2}{\sum }_{k = 1}^{n}{v}_{k}^{2}+\frac{1}{2\gamma }{\widetilde{\theta }}^{2}. | (4.2) |
From (3.58), we have
\dot{V}\le -{\sum }_{k = 1}^{n}{k}_{k}{v}_{k}^{2}+g{c}_{v}N\left(\chi \right)\dot{\chi }+\dot{\chi }-\frac{\sigma }{2\gamma }{\widetilde{\theta }}^{2}+\frac{\sigma }{2\gamma }{\theta }^{2}+\frac{1}{2}{{\overline {g} }}^{2}{{\overline {\Delta} }}^{2}+\frac{n}{2} . | (4.3) |
To facilitate analysis, the above inequality can be written as
\dot{V}\le -{\rm A}V+{\rm B}+\left(g{c}_{v}N\left(\chi \right)+1\right)\dot{\chi , } | (4.4) |
where
{\rm A} \leqslant \min \left\{ {2{k_1}, \ldots , 2{k_n}, \sigma } \right\}, |
{\rm B} = \frac{\sigma }{{2\gamma }}{\theta ^2} + \frac{1}{2}{\bar g^2}{\bar \Delta ^2} + \frac{n}{2}. |
By computing the integration of the above differentiation inequality at the interval \left[{0, t} \right), one can obtain
0\le V\le V\left(0\right){e}^{-{\rm A}t}+\frac{{\rm B}}{{\rm A}}\left(1-{e}^{-{\rm A}t}\right)+{e}^{-{\rm A}t}{\int }_{0}^{t}\left(g{c}_{v}N\left(\chi \right)+1\right)\dot{\chi }{e}^{{\rm A}\tau }d\tau . | (4.5) |
From Lemma 2, we can get that V, \chi and {\int }_{0}^{t}\left(g{c}_{v}N\left(\chi \right)+1\right)\dot{\chi }d\tau are bounded at the interval \left[{0, t} \right). And the result holds even if t \to \infty . Furthermore, let \mathbb{C} be the upper bound of {e}^{-{\rm A}t}{\int }_{0}^{t}\left(g{c}_{v}N\left(\chi \right)+1\right)\dot{\chi }{e}^{{\rm A}\tau }d\tau . The following inequality will holds
V\le \left(V\left(0\right)-\frac{{\rm B}}{{\rm A}}\right){e}^{-{\rm A}t}+\frac{{\rm B}}{{\rm A}}+\mathbb{C} . | (4.6) |
We denote {F_\infty } as the set of all bounded functions. According to the above analysis, we can get V \leqslant p. Therefore, we can obtain that {v_1}, {v_2}, \cdots , {v_n}, \overset{\lower0.5em\hbox{$\smash{\scriptscriptstyle\frown}$}}{\theta } , {\varsigma _1}, {\varsigma _2}, \cdots , {\varsigma _n} are bounded. Furthermore, {e_i} are also bounded due to {v_i} = {e_i} - {\varsigma _i}. As {e}_{1} = {\mathbb{Z}}_{1}\left(t\right) and {e_1} \in {F_\infty } it has {\mathbb{Z}}_{1}\left(t\right)\in {F}_{\infty } . It implies {z_1}\left(t \right), \rho \in {F_\infty } and {s_1} = {z_1} + {\alpha _{1c}} \in {F_\infty }. Noting (2.9), we obtain {\rho _1} \in {F_\infty }. From (2.18), it yields \rho, {\mu }_{1}\left(t\right), {\dot{\mu }}_{1}\left(t\right)\in {F}_{\infty } . With the help of (3.15) and (2.12), we have {\alpha _1} \in {F_\infty } and {\alpha }_{2c}, {\dot{\alpha }}_{2c}\in {F}_{\infty } . Since {s_2} = {v_2} + {\alpha _{2c}} + {\varsigma _2}, we get {s_2} \in {F_\infty }. Noting (2.9), we obtain {\rho _2} \in {F_\infty }. From (3.31) and (2.12), we get {\alpha _2} \in {F_\infty } and {\alpha }_{3c}, {\dot{\alpha }}_{3c}\in {F}_{\infty } . Similarly, we can easily get
{s}_{3}, {\rho }_{3}, {\alpha }_{4c}, {\dot{\alpha }}_{4c}, \cdots , {s}_{n-1}, {\rho }_{n-1}, {\alpha }_{nc}, {\dot{\alpha }}_{nc}, {s}_{n}, {\rho }_{n}, v\in {F}_{\infty } . |
From (2.8), it yields {x}_{1}, {x}_{2}, \cdots {, x}_{n}\in {F}_{\infty } . This finishes the proof.
Remark 5. In the proposed low-cost adaptive neural prescribed performance control (LAFN-PPC) scheme, we can obtain a satisfactory performance by reasonably adjusting {k_i}, \gamma , \sigma , {\bar \kappa _i}, {{\underline{\kappa }}_{i}} , \bar k, \underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{k} , {\mu _0}, {\mu _\infty } and \hbar , where i = 1, 2, \ldots, n. The larger {k_i} and smaller\gamma , \sigma can improve the convergence speed and tracking accuracy of the controller. But too large {k_i} and small \gamma , \sigma values can result in large control input, which may be far beyond physical limitations of actuator. The adaptive parameters \gamma and \sigma play the part of regulators between controller and control output. To avoid the large control input, the values of the parameters {k_i}, \gamma , \sigma , i = 1, 2, \ldots, n are limited to a certain interval. Meanwhile, {\bar \kappa _i} and {{\underline{\kappa }}_{i}} are used to constrain the states {x_i}\left(t \right) within the open sets \left({ - {{\underline{\kappa }}_{i}}, {{\bar \kappa }_i}} \right), which can be selected according to the actual application requirements. From this it can be concluded that the time varying parameter
{\rho _i} = {{\left( {{{\underline{\kappa }}_{i}}{{\bar \kappa }_i}\left( {{{\underline{\kappa }}_{i}}{{\bar \kappa }_i} + x_i^2\left( t \right)} \right)} \right)} / {\left( {{{\left( {{{\underline{\kappa }}_{i}} + {x_i}\left( t \right)} \right)}^2}{{\left( {{{\bar \kappa }_i} - {x_i}\left( t \right)} \right)}^2}} \right)}} . |
The parameters \bar k, \underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{k} , {\mu _0}, {\mu _\infty } and \hbar are designed to constrain the transformed output tracking error within the open set \left({ - \underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{k} \mu \left(t \right), \bar k\mu \left(t \right)} \right), where
\mu \left( t \right) = \left( {{\mu _0} - {\mu _\infty }} \right){e^{ - \hbar t}} + {\mu _\infty }. |
{\mu _0}, \hbar and {\mu _\infty } determine the initial error, the error convergence rate and the steady-state error of the transformed output tracking error bound, which can be selected according to the performance requirements. \bar k and \underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{k} are always within \left({0, 1} \right] and can deal with the issues considering symmetric or asymmetric transformed output tracking error constraint.
In order to prove the effectiveness and feasibility of our control scheme, this section provides comparison simulation cases. Meanwhile, the control schemes based on works in [29,30] are compared with our suggested control scheme.
A rigid manipulator system [30] is given as
\left\{\begin{array}{l}{J}_{L}\ddot{\theta }+M{g}_{v}L\mathit{sin}\theta +{T}_{E} = u\\ u = H\left(v\right)\end{array}\right. | (5.1) |
where \theta , \dot{\theta } and \ddot{\theta } are the position, velocity and acceleration of the link, respectively. M, {g_v} and L denote the link mass, gravity constant and the distance from the joint to the mass center of the link, respectively. {T_E} indicates the unknown external load. u is the actual control torque which is subjected to the unknown input constraint nonlinearity H\left(v \right), and v is system control input signal.
To facilitate the controller design, we transfer the system (5.1) with new variables. Let {x_1} = \theta , {x}_{2} = \dot{\theta } . Then the dynamic model of system (5.1) can be expressed as follows:
\left\{\begin{array}{l}{\dot{x}}_{1} = {x}_{2}\\ {\dot{x}}_{2} = {J}_{L}^{-1}\left(u-MgL\mathit{sin}{x}_{1}-{T}_{E}\right)\\ u = H\left(v\right)\\ y = {x}_{1}\end{array}\right. . | (5.2) |
To simplify the expression, the dynamic model is rewritten as
\left\{\begin{array}{l}{\dot{x}}_{1} = {x}_{2}\\ {\dot{x}}_{2} = {G}_{L}u+{f}_{L}\\ u = H\left(v\right) = {c}_{v}v+\Delta \left(v\right)\\ y = {x}_{1}\end{array}\right. | (5.3) |
where
{f_L} = \frac{{ - MgL\sin {x_1} - {T_E}}}{{{J_L}}}, {G_L} = J_L^{ - 1}. |
In the simulation, we set {f_1}\left({{x_1}} \right) = 0 , {f_2}\left({{x_1}, {x_2}} \right) = {f_L} and g = {G_L}. M = 0.35Kg, {g}_{v} = 9.8 m/s2, L = 1.47m and {T_E} = 0.1\sin t. The tracking signal of following cases are set as {y_r} = 0.5\sin t.
Case 1. The initial system states and output are x\left(0 \right) = \left[{0.1, 0} \right]. And \overset{\lower0.5em\hbox{$\smash{\scriptscriptstyle\frown}$}}{\theta } (0) = {\alpha _{jc}}(0) = {\alpha _{jc, j}}(0) = 0, {\varsigma _1}(0) = {\varsigma _2}(0) = 0. The proposed controller, adaptive laws and error compensation schemes are constructed as
\left\{\begin{array}{l}{\alpha }_{1} = \left(-{k}_{1}{e}_{1}+\rho {\rho }_{1r}{\dot{y}}_{r}+\rho \frac{\dot{\mu }\left(t\right){z}_{1}\left(t\right)}{\mu \left(t\right)}\right){\rho }^{-1}{\rho }_{1}^{-1}{\eta }_{2}^{-1}\\ v = N\left(\chi \right)\psi {\rho }_{2}^{-1}\\ \psi = {k}_{2}{e}_{2}+{v}_{2}{\overset\frown{\theta }}\zeta +\rho {\rho }_{1}{\eta }_{2}{e}_{1}-{\dot{\alpha }}_{nc}+\frac{1}{2}{\rho }_{2}^{2}{v}_{2}\\ \dot{\chi } = {v}_{2}\psi \end{array}\right. , | (5.4) |
\left\{\begin{array}{l}{\dot{\varsigma }}_{1} = -{k}_{1}{\varsigma }_{1}+\rho {\rho }_{1}{\eta }_{2}\left({\varsigma }_{2}+{\alpha }_{2c}-{\alpha }_{1}\right)\\ {\dot{\varsigma }}_{2} = -{k}_{2}{\varsigma }_{2}-\rho {\rho }_{1}{\eta }_{2}{\varsigma }_{1}\end{array}\right. , | (5.5) |
\dot{{\overset\frown{\theta }}} = \gamma {v}_{2}^{2}{\zeta }_{2}-\sigma \overset\frown{\theta }, | (5.6) |
where the parameters are chosen as {k_1} = 50, {k_2} = 15, \gamma = 10, \sigma = 0.7, \omega = 300, \varpi = 0.95. The parameters of FWNN are chosen as: The nodes of neural network is 7, the center c_1^j is distributed in the field of \left[{- 3, 3} \right], and its width \omega _1^j = 1. The parameters of full-state constraints, input constraint and prescribed performance constraint are set as: {\bar \kappa _1} = 1.5, {\underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{\kappa } _1} = 1.5, {\underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{\kappa } _2} = 3, {\bar \kappa _2} = 4, {u^ + } = 30, {u^ - } = - 30, {\mu _0} = 0.2, {\mu _\infty } = 0.02, \hbar = 1, \underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{k} = 1, \bar k = 1.
Case 2. Base on the case 1, we further study the output tracking errors when give different prescribed performance. {e_{r1}} for {\mu _\infty } = 0.02, {e_{r2}} for {\mu _\infty } = 0.016, {e_{r3}} for {\mu _\infty } = 0.012, {e_{r4}} for {\mu _\infty } = 0.008, {e_{r5}} for {\mu _\infty } = 0.004, {e_{r6}} for {\mu _\infty } = 0.002.
Case 3. Based on the case 2, we set {\mu _\infty } = 0.004 and {y_d} = {y_r}, two comparative simulation from [29,30] are carried out to further show the advantage of our control scheme, which is still based on the rigid manipulator system (5.3). The controller, adaptive laws and first-order filter of work in [29] are given as
\left\{\begin{array}{l}{\alpha }_{1} = -{c}_{1}\frac{{{\underline{F } }}_{1}{{\overline {F} }}_{1}\left({{\underline{F } }}_{1}{{\overline {F} }}_{1}-{x}_{1}^{2}\right)}{{\left({{\underline{F } }}_{1}+{x}_{1}\right)}^{2}{\left({{\overline {F} }}_{1}-{x}_{1}\right)}^{2}}{z}_{1}+\frac{{{\underline{F } }}_{1}{{\overline {F} }}_{1}\left({{\underline{F } }}_{1}{{\overline {F} }}_{1}-{y}_{d}^{2}\right)}{{\left({{\underline{F } }}_{1}+{y}_{d}\right)}^{2}{\left({{\overline {F} }}_{1}-{y}_{d}\right)}^{2}}{\dot{y}}_{d}\\ u = -\left({c}_{2}+\widehat{\theta }{\mathit{\Phi}} \right)\frac{{{\underline{F } }}_{2}{{\overline {F} }}_{2}\left({{\underline{F } }}_{2}{{\overline {F} }}_{2}-{x}_{2}^{2}\right)}{{\left({{\underline{F } }}_{2}+{x}_{2}\right)}^{2}{\left({{\overline {F} }}_{2}-{x}_{2}\right)}^{2}}{z}_{2}\\ {\mathit{\Phi}} = {‖\frac{{{\underline{F } }}_{1}{{\overline {F} }}_{1}\left({{\underline{F } }}_{1}{{\overline {F} }}_{1}-{x}_{1}^{2}\right)}{{\left({{\underline{F } }}_{1}+{x}_{1}\right)}^{2}{\left({{\overline {F} }}_{1}-{x}_{1}\right)}^{2}}‖}^{2}\left(1+{\phi }^{2}\right)+{‖{\dot{\alpha }}_{2f}‖}^{2}\end{array}\right. , | (5.7) |
\dot{\widehat{\theta }} = \gamma {‖{z}_{2}‖}^{2}{\mathit{\Phi}} -\sigma \widehat{\theta , } | (5.8) |
\varepsilon {\dot{\alpha }}_{2f}+{\alpha }_{2f} = \frac{{{\underline{F } }}_{2}{{\overline {F} }}_{2}}{\left({{\underline{F } }}_{2}+{x}_{2}\right)\left({{\overline {F} }}_{2}-{x}_{2}\right)}{\alpha }_{1}, | (5.9) |
where {z_1} = {x_1} - {y_d}, {z_2} = {x_2} - {\alpha _{2f}}, \theta \left(0 \right) = 0, {\alpha _{2f}}\left(0 \right) = 0, {c_1} = 40, {c_2} = 80, {\underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{F} _1} = 1.5, {\bar F_1} = 1.5, {\underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{F} _2} = 3, {\bar F_2} = 4, \gamma = 15, \sigma = 0.5, \varphi = 5 and \varepsilon = 0.03. Detailed controller design and parameter meaning is found in [29].
The controller, adaptive laws and first-order filter of works in [30]. are given as
\left\{\begin{array}{l}{\alpha }_{1} = -{\left(\frac{{{\underline{ \kappa } }}_{1}{{\overline {\kappa} }}_{1}\left({{\underline{ \kappa } }}_{1}{{\overline {\kappa} }}_{1}+{x}_{1}^{2}\right)}{{\left({{\underline{ \kappa } }}_{1}+{x}_{1}\right)}^{2}{\left({{\overline {\kappa} }}_{1}-{x}_{1}\right)}^{2}}\right)}^{-1}\left({k}_{1}{z}_{1}-\frac{{{\underline{ \kappa } }}_{1}{{\overline {\kappa} }}_{1}\left({{\underline{ \kappa } }}_{1}{{\overline {\kappa} }}_{1}+{y}_{d}^{2}\right)}{{\left({{\underline{ \kappa } }}_{1}+{y}_{d}\right)}^{2}{\left({{\overline {\kappa} }}_{1}-{y}_{d}\right)}^{2}}{\dot{y}}_{d}\right)\\ \mu = N\left({\zeta }_{2}\right){v}_{2}\\ {v}_{2} = {\left(\frac{{{\underline{ \kappa } }}_{2}{{\overline {\kappa} }}_{2}\left({{\underline{ \kappa } }}_{2}{{\overline {\kappa} }}_{2}+{x}_{2}^{2}\right)}{{\left({{\underline{ \kappa } }}_{2}+{x}_{2}\right)}^{2}{\left({{\overline {\kappa} }}_{2}-{x}_{2}\right)}^{2}}\right)}^{-1}\left({k}_{2}{z}_{2}+{z}_{2}{\widehat{\theta }}^{T}\varphi \left(Z\right)-{\dot{\psi }}_{2}\right)\\ {\dot{\zeta }}_{2} = \frac{{{\underline{ \kappa } }}_{2}{{\overline {\kappa} }}_{2}\left({{\underline{ \kappa } }}_{2}{{\overline {\kappa} }}_{2}+{x}_{2}^{2}\right)}{{\left({{\underline{ \kappa } }}_{2}+{x}_{2}\right)}^{2}{\left({{\overline {\kappa} }}_{2}-{x}_{2}\right)}^{2}}{z}_{2}{v}_{2}\end{array}\right. , | (5.10) |
\dot{\widehat{\theta }} = \Gamma \left(\varphi \left(Z\right){z}_{2}^{2}-\sigma \widehat{\theta }\right), | (5.11) |
\delta {\dot{\psi }}_{2}+{\psi }_{2} = \frac{{{\underline{ \kappa } }}_{2}{{\overline {\kappa} }}_{2}}{\left({{\underline{ \kappa } }}_{2}+{x}_{2}\right)\left({{\overline {\kappa} }}_{2}-{x}_{2}\right)}{\alpha }_{1}, | (5.12) |
where {z_1} = {x_1} - {y_d}, {z_2} = {x_2} - {\psi _2}, {\zeta _2}\left(0 \right) = 0, {\psi _2}\left(0 \right) = 0, {k_1} = 50, {k_2} = 30, {\underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{k} _1} = 1.5, {\bar k_1} = 1.5, {\underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{k} _2} = 3, {\bar k_2} = 4, \Gamma = 15 and \sigma = 0.5.\phi \left(Z \right) is the activation function of neural network, Z is the input of neural network. The parameters of RBF are chosen as: The nodes of neural network is 7, the center {\mu _1} is distributed in the field of \left[{- 0.3, 0.3} \right], and its width {\sigma _1} = 0.1. Detailed controller design and parameter meaning is found in [30].
Figures 4–7 reflect the main results of our control scheme. Figure 4 shows that output y of rigid manipulator system can track desired trajectory well without violating the output constraint. Figure 5 reveals the output tracking error {e_r} is in the region boundary all the times. The actual controls uis showed in Figure 6. And the state {x_2} is illustrated in Figure 7. From Figure 8, the proposed control scheme tracks the desired signal well with different prescribed performance, and the boundedness of prescribed performance is not violated.
Figures 9–11 present the comparative results of tracking trajectories, output tracking errors and input signals. We can easily find that the results of LAFN-PPC is better than controllers in [29,30]. Hence, for strict-feedback systems with purpose of high-precision tracking performance, full-state constraints and input constraint, we can conclude that our suggested control scheme has a potential to control them.
The LAFN-PPC is newly constructed for strict-feedback systems with prescribed output performance, full-state constraints and input constraint. The newly constructed command filter based adaptive control scheme with an error compensation mechanism can not only overcome the so-called "EOC" problem, but also reduce filter errors. By introducing a one-to-one nonlinear state transformation function, the full-state constraints are resolved. The prescribed performance can be guaranteed by using the one-to-one nonlinear error transformation function. The unknown control direction and the input constraint nonlinearity are resolved by introducing the Nussbaum function simultaneously. Moreover, the large computational cost is solved by introducing a virtual parameter of adaptive laws. Only one parameter needs to be updated online. Future works can focus on command-filter based optimal control of nonstrict-feedback systems considering performance constraint and address how to minimize resources consumption while ensuring performance.
Proof. According to the state transformation function (2.8), we have
{\eta _i} = \frac{{ - x_i^2 \;+ \;\left( {{{\bar \kappa }_i} - {{\underline{\kappa }}_{i}}} \right){x_i} + {{\underline{\kappa }}_{i}}\;{{\bar \kappa }_i}}}{{{{\underline{\kappa }}_{i}}\;{{\bar \kappa }_i}}}. | (A.1) |
{x_i} are always constrained in the open sets \left({ - {{\underline{\kappa }}_{i}}, {{\bar \kappa }_i}} \right), i.e., {x_i}\left(t \right) \in {{{\Omega}} _i}. Therefore, we can easily obtain that {\eta _i} > 0 and {\eta _i}take the maximal value at the points x_i^0 = {{\left({{{\bar \kappa }_i} - {{\underline{\kappa }}_{i}}} \right)} /2} according to the properties of the quadratic function. It is obvious that x_i^0 is in the interval \left({ - {{\underline{\kappa }}_{i}}, {{\bar \kappa }_i}} \right). Thus, the maximal value of {\eta _i} is
{\bar \eta _i} = \frac{{{{\left( {{{\bar \kappa }_i} + {{\underline{\kappa }}_{i}}} \right)}^2}}}{{4{{\underline{\kappa }}_{i}}{{\bar \kappa }_i}}}, i = 1, 2, \cdots , n. | (A.2) |
This finishes the proof.
Proof. According to (2.5), (2.11), (2.13) and (2.14), we have
\begin{gathered} {s_1} - {\alpha _{1c}} = \frac{{{{\underline{\kappa }}_{1}}{{\bar \kappa }_1}{x_1}}}{{\left( {{{\underline{\kappa }}_{1}} + {x_1}} \right)\left( {{{\bar \kappa }_1} - {x_1}} \right)}} - \frac{{{{\underline{\kappa }}_{1}}{{\bar \kappa }_1}{y_r}}}{{\left( {{{\underline{\kappa }}_{1}} + {y_r}} \right)\left( {{{\bar \kappa }_1} - {y_r}} \right)}} \hfill \\ = \left( { - \frac{{\underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{\kappa } _1^2{{\bar \kappa }_1}}}{{{{\underline{\kappa }}_{1}} + {{\bar \kappa }_1}}}\frac{1}{{{{\underline{\kappa }}_{1}} + {x_1}}} + \frac{{{{\underline{\kappa }}_{1}}\bar \kappa _1^2}}{{{{\underline{\kappa }}_{1}} + {{\bar \kappa }_1}}}\frac{1}{{{{\bar \kappa }_1} - {x_1}}}} \right) - \left( { - \frac{{\underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{\kappa } _1^2{{\bar \kappa }_1}}}{{{{\underline{\kappa }}_{1}} + {{\bar \kappa }_1}}}\frac{1}{{{{\underline{\kappa }}_{1}} + {y_r}}} + \frac{{{{\underline{\kappa }}_{1}}\bar \kappa _1^2}}{{{{\underline{\kappa }}_{1}} + {{\bar \kappa }_1}}}\frac{1}{{{{\bar \kappa }_1} - {y_r}}}} \right) \hfill \\ = \frac{{\underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{\kappa } _1^2{{\bar \kappa }_1}}}{{{{\underline{\kappa }}_{1}} + {{\bar \kappa }_1}}}\left( {\frac{1}{{{{\underline{\kappa }}_{1}} + {y_r}}} - \frac{1}{{{{\underline{\kappa }}_{1}} + {x_1}}}} \right) + \frac{{{{\underline{\kappa }}_{1}}\bar \kappa _1^2}}{{{{\underline{\kappa }}_{1}} + {{\bar \kappa }_1}}}\left( {\frac{1}{{{{\bar \kappa }_1} - {x_1}}} - \frac{1}{{{{\bar \kappa }_1} - {y_r}}}} \right) \hfill \\ = \frac{{\underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{\kappa } _1^2{{\bar \kappa }_1}}}{{{{\underline{\kappa }}_{1}} + {{\bar \kappa }_1}}}\frac{{{x_1} - {y_r}}}{{\left( {{{\underline{\kappa }}_{1}} + {y_r}} \right)\left( {{{\underline{\kappa }}_{1}} + {x_1}} \right)}} + \frac{{{{\underline{\kappa }}_{1}}\bar \kappa _1^2}}{{{{\underline{\kappa }}_{1}} + {{\bar \kappa }_1}}}\frac{{{x_1} - {y_r}}}{{\left( {{{\bar \kappa }_1} - {x_1}} \right)\left( {{{\bar \kappa }_1} - {y_r}} \right)}} \hfill \\ = \left[ {\frac{{\underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{\kappa } _1^2{{\bar \kappa }_1}}}{{\left( {{{\underline{\kappa }}_{1}} + {{\bar \kappa }_1}} \right)\left( {{{\underline{\kappa }}_{1}} + {y_r}} \right)\left( {{{\underline{\kappa }}_{1}} + {x_1}} \right)}} + \frac{{{{\underline{\kappa }}_{1}}\bar \kappa _1^2}}{{\left( {{{\underline{\kappa }}_{1}} + {{\bar \kappa }_1}} \right)\left( {{{\bar \kappa }_1} - {x_1}} \right)\left( {{{\bar \kappa }_1} - {y_r}} \right)}}} \right]\left( {{x_1} - {y_r}} \right) \hfill \\ \end{gathered} . | (B.1) |
{x_1} and {y_r} are constrained in the open sets \left({ - {{\underline{\kappa }}_{i}}, {{\bar \kappa }_i}} \right). Thus, 0 < {\underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{\kappa } _1} + {y_r} < {\underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{\kappa } _1} + {\bar \kappa _1}, 0 < {\underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{\kappa } _1} + {x_1} < {\underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{\kappa } _1} + {\bar \kappa _1}, 0 < {\bar \kappa _1} - {x_1} < {\underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{\kappa } _1} + {\bar \kappa _1} and 0 < {\bar \kappa _1} - {y_r} < {\underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{\kappa } _1} + {\bar \kappa _1}. One obtains
\begin{gathered} 0 < \frac{{\underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{\kappa } _1^2{{\bar \kappa }_1}}}{{\left( {{{\underline{\kappa }}_{1}} + {{\bar \kappa }_1}} \right)\left( {{{\underline{\kappa }}_{1}} + {y_r}} \right)\left( {{{\underline{\kappa }}_{1}} + {x_1}} \right)}} + \frac{{{{\underline{\kappa }}_{1}}\bar \kappa _1^2}}{{\left( {{{\underline{\kappa }}_{1}} + {{\bar \kappa }_1}} \right)\left( {{{\bar \kappa }_1} - {x_1}} \right)\left( {{{\bar \kappa }_1} - {y_r}} \right)}} \hfill \\ < \frac{{\underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{\kappa } _1^2{{\bar \kappa }_1}}}{{{{\left( {{{\underline{\kappa }}_{1}} + {{\bar \kappa }_1}} \right)}^3}}} + \frac{{{{\underline{\kappa }}_{1}}\bar \kappa _1^2}}{{{{\left( {{{\underline{\kappa }}_{1}} + {{\bar \kappa }_1}} \right)}^3}}} = \frac{{{{\underline{\kappa }}_{1}}{{\bar \kappa }_1}}}{{{{\left( {{{\underline{\kappa }}_{1}} + {{\bar \kappa }_1}} \right)}^2}}} \hfill \\ \end{gathered} . | (B.2) |
The transformed output tracking error {z_1} = {s_1} - {\alpha _{1c}} strictly converges in the prescribed performance region - \underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{k} \mu \left(t \right) < {z_1}\left(t \right) < \bar k\mu \left(t \right) . Hence, it has
- \frac{{\underset{\raise0.3em\hbox{$\smash{\scriptscriptstyle-}$}}{k} \mu \left( t \right){{\left( {{{\underline{\kappa }}_{1}} + {{\bar \kappa }_1}} \right)}^2}}}{{{{\underline{\kappa }}_{1}}{{\bar \kappa }_1}}} < {x_1} - {y_r} < \frac{{\bar k\mu \left( t \right){{\left( {{{\underline{\kappa }}_{1}} + {{\bar \kappa }_1}} \right)}^2}}}{{{{\underline{\kappa }}_{1}}{{\bar \kappa }_1}}} . | (B.3) |
This finishes the proof.
The authors would like to appreciate all the editors and reviewers for improving the quality of this article. This work was supported by the National Key Research and Development Program of China (2018YFB1304800) and Key Research and Development Program of Guangdong Province (2020B090926002).
All authors declare no conflicts of interest in this paper.
[1] |
W. He, X. Gao, W. Zhong, Secure impulsive synchronization control of multi-agent systems under deception attacks, Inf. Sci., 459 (2018), 354–368. https://doi.org/10.1016/j.ins.2018.04.020 doi: 10.1016/j.ins.2018.04.020
![]() |
[2] |
Z. Xu, X. Li, V. Stojanovic, Exponential stability of nonlinear state-dependent delayed impulsive systems with applications, Nonlinear Anal., 42 (2021), 101088. https://doi.org/10.1016/j.nahs.2021.101088 doi: 10.1016/j.nahs.2021.101088
![]() |
[3] |
Q. Zhu, J. Cao, R. Rakkiyappan, Exponential input-to-state stability of stochastic Cohen-Grossberg neural networks with mixed delays, Nonlinear Dyn., 79 (2015), 1085–1098. https://doi.org/10.1007/s11071-014-1725-2 doi: 10.1007/s11071-014-1725-2
![]() |
[4] |
G. Stamov, E. Gospodinova, I. Stamova, Practical exponential stability with respect to h-manifolds of discontinuous delayed Cohen-Grossberg neural networks with variable impulsive perturbations, Math. Model. Control, 1 (2021), 26–34. https://doi.org/10.3934/mmc.2021003 doi: 10.3934/mmc.2021003
![]() |
[5] |
C. Aouiti, E. A. Assali, Nonlinear Lipschitz measure and adaptive control for stability and synchronization in delayed inertial Cohen-Grossberg-type neural networks, Int. J. Adapt. Control Signal Process., 33 (2019), 1457–1477. https://doi.org/10.1002/acs.3042 doi: 10.1002/acs.3042
![]() |
[6] |
X. Li, X. Zhang, S. Song, Effect of delayed impulses on input-to-state stability of nonlinear systems, Automatica, 76 (2017), 378–382. https://doi.org/10.1016/j.automatica.2016.08.009 doi: 10.1016/j.automatica.2016.08.009
![]() |
[7] |
B. Jiang, J. Lu, X. Li, J. Qiu, Event-triggered impulsive stabilization of systems with external disturbances, IEEE Trans. Autom. Control, 67 (2021), 2116–2122. https://doi.org/10.1109/TAC.2021.3108123 doi: 10.1109/TAC.2021.3108123
![]() |
[8] |
Y. Xing, X. He, X. Li, Finite-time stabilization of nonlinear time-varying systems involving impulsive action, Commun. Nonlinear Sci. Numer. Simul., 131 (2024), 107818. https://doi.org/10.1016/j.cnsns.2024.107818 doi: 10.1016/j.cnsns.2024.107818
![]() |
[9] |
M. Yu, S. Wu, X. Li, Exponential stabilization of nonlinear systems under saturated control involving impulse correction, Nonlinear Anal., 48 (2023), 101335. https://doi.org/10.1016/j.nahs.2023.101335 doi: 10.1016/j.nahs.2023.101335
![]() |
[10] |
Y. Wang, J. Lu, Y. Lou, Halanay-type inequality with delayed impulses and its applications, Sci. China Inf. Sci., 62 (2019), 192206. https://doi.org/10.1007/s11432-018-9809-y doi: 10.1007/s11432-018-9809-y
![]() |
[11] |
J. Cao, L. Li, Cluster synchronization in an array of hybrid coupled neural networks with delay, Neural Networks, 22 (2009), 335–342. https://doi.org/10.1016/j.neunet.2009.03.006 doi: 10.1016/j.neunet.2009.03.006
![]() |
[12] |
G. Chesi, Robustness analysis of genetic regulatory networks affected by model uncertainty, Automatica, 47 (2011), 1131–1138. https://doi.org/10.1016/j.automatica.2010.10.012 doi: 10.1016/j.automatica.2010.10.012
![]() |
[13] |
L. Liu, Z. Wu, Intermittent stochastic stabilization based on discrete-time observation with time delay, Syst. Control Lett., 137 (2020), 104626. https://doi.org/10.1016/j.sysconle.2020.104626 doi: 10.1016/j.sysconle.2020.104626
![]() |
[14] |
W. Chen, W. Zheng, Delay-independent minimum dwell time for exponential stability of uncertain switched delay systems, IEEE Trans. Autom. Control, 55 (2010), 2406–2413. https://doi.org/10.1109/TAC.2010.2054845 doi: 10.1109/TAC.2010.2054845
![]() |
[15] |
X. Li, J. Wu, Stability of nonlinear differential systems with state-dependent delayed impulses, Automatica, 64 (2016), 63–69. https://doi.org/10.1016/j.automatica.2015.10.002 doi: 10.1016/j.automatica.2015.10.002
![]() |
[16] |
H. Wang, S. Duan, C. Li, L. Wang, T. Huang, Stability of impulsive delayed linear differential systems with delayed impulses, J. Franklin Inst., 352 (2015), 3044–3068. https://doi.org/10.1016/j.jfranklin.2014.12.009 doi: 10.1016/j.jfranklin.2014.12.009
![]() |
[17] | K. Gu, V. L. Kharitonov, J. Chen, Stability of time-delay systems, Springer Science Business Media, 2003. https://doi.org/10.1007/978-1-4612-0039-0 |
[18] |
X. Qi, H. Bao, J. Cao, Exponential input-to-state stability of quaternion-valued neural networks with time delay, Appl. Math. Comput., 358 (2019), 382–393. https://doi.org/10.1016/j.amc.2019.04.045 doi: 10.1016/j.amc.2019.04.045
![]() |
[19] |
B. Jiang, J. Lu, J. Lou, J. Qiu, Synchronization in an array of coupled neural networks with delayed impulses: average impulsive delay method, Neural Networks, 121 (2020), 452–460. https://doi.org/10.1016/j.neunet.2019.09.019 doi: 10.1016/j.neunet.2019.09.019
![]() |
[20] |
A. Seuret, C. Edwards, S. K. Spurgeon, E. Fridman, Static output feedback sliding mode control design via an artificial stabilizing delay, IEEE Trans. Autom. Control, 54 (2009), 256–265. https://doi.org/10.1109/TAC.2008.2010889 doi: 10.1109/TAC.2008.2010889
![]() |
[21] |
J. Lu, B. Jiang, W. X. Zheng, Potential impacts of delay on stability of impulsive control systems, IEEE Trans. Autom. Control, 67 (2022), 5179–5190. https://doi.org/10.1109/TAC.2021.3120672 doi: 10.1109/TAC.2021.3120672
![]() |
[22] |
X. Li, S. Song, Stabilization of delay systems: delay-dependent impulsive control, IEEE Trans. Autom. Control, 62 (2017), 406–411. https://doi.org/10.1109/TAC.2016.2530041 doi: 10.1109/TAC.2016.2530041
![]() |
[23] |
X. Li, P. Li, Stability of time-delay systems with impulsive control involving stabilizing delays, Automatica, 124 (2021), 109336. https://doi.org/10.1016/j.automatica.2020.109336 doi: 10.1016/j.automatica.2020.109336
![]() |
[24] |
W. Liu, X. He, X. Li, Lyapunov conditions for exponential stability of nonlinear delay systems via impulsive control involving stabilizing delays, Nonlinear Anal., 51 (2024), 101411. https://doi.org/10.1016/j.nahs.2023.101411 doi: 10.1016/j.nahs.2023.101411
![]() |
[25] |
X. Li, S. Song, J. Wu, Exponential stability of nonlinear systems with delayed impulses and applications, IEEE Trans. Autom. Control, 64 (2019), 4024–4034. https://doi.org/10.1109/TAC.2019.2905271 doi: 10.1109/TAC.2019.2905271
![]() |
[26] |
B. Jiang, J. Lu, Y. Liu, Exponential stability of delayed systems with average-delay impulses, SIAM J. Control Optim., 58 (2020), 3763–3784. https://doi.org/10.1137/20M1317037 doi: 10.1137/20M1317037
![]() |
[27] |
H. Zhu, P. Li, X. Li, Input-to-state stability of impulsive systems with hybrid delayed impulse effects, J. Appl. Anal. Comput., 9 (2019), 777–795. https://doi.org/10.11948/2156-907X.20180182 doi: 10.11948/2156-907X.20180182
![]() |
[28] |
X. Yang, Z. Yang, Synchronization of TS fuzzy complex dynamical networks with time-varying impulsive delays and stochastic effects, Fuzzy Sets Syst., 235 (2014), 25–43. https://doi.org/10.1016/j.fss.2013.06.008 doi: 10.1016/j.fss.2013.06.008
![]() |
[29] |
X. Li, J. Cao, An impulsive delay inequality involving unbounded time-varying delay and applications, IEEE Trans. Autom. Control, 62 (2017), 3618–3625. https://doi.org/10.1109/TAC.2017.2669580 doi: 10.1109/TAC.2017.2669580
![]() |
[30] |
J. Cao, K. Yuan, H. Li, Global asymptotical stability of recurrent neural networks with multiple discrete delays and distributed delays, IEEE Trans. Neural Networks, 17 (2006), 1646–1651. https://doi.org/10.1109/TNN.2006.881488 doi: 10.1109/TNN.2006.881488
![]() |
1. | Yankui Song, Gong Cheng, Yaoyao Tuo, Adaptive output feedback control for chaotic PMSMs stochastic system considering constraints, 2024, 187, 09600779, 115321, 10.1016/j.chaos.2024.115321 |