x_1 | x_2 | f_1 | f_2 |
1 | 1 | 0 | 1 |
1 | 0 | 1 | 1 |
0 | 1 | 0 | 0 |
0 | 0 | 0 | 0 |
Recently, the theory of semi-tensor product (STP) method of matrices has received much attention from variety communities covering engineering, economics and industries, etc. This paper describes a detailed survey on some recent applications of the STP method in finite systems. First, some useful mathematical tools on the STP method are provided. Second, many recent developments about robustness analysis on the given finite systems are delineated, such as robust stable analysis of switched logical networks with time-delayed, robust set stabilization of Boolean control networks, event-triggered controller design for robust set stabilization of logical networks, stability analysis in distribution of probabilistic Boolean networks, and how to solve a disturbance decoupling problem by event triggered control for logical control networks. Finally, several research problems in future works are predicted.
Citation: Guodong Zhao, Haitao Li, Ting Hou. Survey of semi-tensor product method in robustness analysis on finite systems[J]. Mathematical Biosciences and Engineering, 2023, 20(6): 11464-11481. doi: 10.3934/mbe.2023508
[1] | Chao Wang, Cheng Zhang, Dan He, Jianliang Xiao, Liyan Liu . Observer-based finite-time adaptive fuzzy back-stepping control for MIMO coupled nonlinear systems. Mathematical Biosciences and Engineering, 2022, 19(10): 10637-10655. doi: 10.3934/mbe.2022497 |
[2] | Shanqing Zhang, Xiaoyun Guo, Xianghua Xu, Li Li, Chin-Chen Chang . A video watermark algorithm based on tensor decomposition. Mathematical Biosciences and Engineering, 2019, 16(5): 3435-3449. doi: 10.3934/mbe.2019172 |
[3] | ZUBOVA Svetlana Petrovna, RAETSKIY Kirill Alexandrovich . Modeling the trajectory of motion of a linear dynamic system with multi-point conditions. Mathematical Biosciences and Engineering, 2021, 18(6): 7861-7876. doi: 10.3934/mbe.2021390 |
[4] | Jian Meng, Bin Zhang, Tengda Wei, Xinyi He, Xiaodi Li . Robust finite-time stability of nonlinear systems involving hybrid impulses with application to sliding-mode control. Mathematical Biosciences and Engineering, 2023, 20(2): 4198-4218. doi: 10.3934/mbe.2023196 |
[5] | Xiaobing Feng, Tingao Jiang . Mathematical and numerical analysis for PDE systems modeling intravascular drug release from arterial stents and transport in arterial tissue. Mathematical Biosciences and Engineering, 2024, 21(4): 5634-5657. doi: 10.3934/mbe.2024248 |
[6] | Iman Mousavian, Mohammad Bagher Shamsollahi, Emad Fatemizadeh . Noninvasive fetal ECG extraction using doubly constrained block-term decomposition. Mathematical Biosciences and Engineering, 2020, 17(1): 144-159. doi: 10.3934/mbe.2020008 |
[7] | Azmy S. Ackleh, Rainey Lyons, Nicolas Saintier . High resolution finite difference schemes for a size structured coagulation-fragmentation model in the space of radon measures. Mathematical Biosciences and Engineering, 2023, 20(7): 11805-11820. doi: 10.3934/mbe.2023525 |
[8] | Kaiyin Zhou, YuxingWang, Sheng Zhang, Mina Gachloo, Jin-Dong Kim, Qi Luo, Kevin Bretonnel Cohen, Jingbo Xia . GOF/LOF knowledge inference with tensor decomposition in support of high order link discovery for gene, mutation and disease. Mathematical Biosciences and Engineering, 2019, 16(3): 1376-1391. doi: 10.3934/mbe.2019067 |
[9] | Jinxin Du, Lei Liu . Adaptive fuzzy fixed time formation control of state constrained nonlinear multi-agent systems against FDI attacks. Mathematical Biosciences and Engineering, 2024, 21(3): 4724-4741. doi: 10.3934/mbe.2024207 |
[10] | M. E. Bakr, Abdulhakim A. Al-Babtain, Zafar Mahmood, R. A. Aldallal, Saima Khan Khosa, M. M. Abd El-Raouf, Eslam Hussam, Ahmed M. Gemeay . Statistical modelling for a new family of generalized distributions with real data applications. Mathematical Biosciences and Engineering, 2022, 19(9): 8705-8740. doi: 10.3934/mbe.2022404 |
Recently, the theory of semi-tensor product (STP) method of matrices has received much attention from variety communities covering engineering, economics and industries, etc. This paper describes a detailed survey on some recent applications of the STP method in finite systems. First, some useful mathematical tools on the STP method are provided. Second, many recent developments about robustness analysis on the given finite systems are delineated, such as robust stable analysis of switched logical networks with time-delayed, robust set stabilization of Boolean control networks, event-triggered controller design for robust set stabilization of logical networks, stability analysis in distribution of probabilistic Boolean networks, and how to solve a disturbance decoupling problem by event triggered control for logical control networks. Finally, several research problems in future works are predicted.
Nowadays, Boolean networks and logical networks, which are two classical finite systems, have been extensively investigated in both theory and applications. The concept of Boolean networks is initiated by Kauffman [1] to model gene regulatory networks. Especially, Boolean network is one kind of logical networks.
Recently, Cheng [2] proposed a new powerful mathematical tool, which is the STP method. From then, many scholars have applied the STP method to model and analyze Boolean networks and logical networks. After the algebraic expressions for Boolean networks and logical networks had been established, Boolean networks and logical networks have been commonly used mathematica models in a variety of communities. Some typical communities includes game theory, networked evolutionary games [3,4], cyber-physical system [5] and gene regulatory networks [6,7,8].
What is noteworthy is that, many great results have been obtained after scholars applied the STP method to solve all kinds of classical control problems, such as stable analysis, stabilization controller design, optimize control, pinning control, etc. Readers can see more details in [2,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36].
Robustness is a system property. It describes the ability of a system to function correctly when unforeseen events appears. Robustness analysis is also a hot research topic for scholars in the control community. This paper focuses on the recent developments on the applications of the STP method in robustness analysis for finite systems and aims to give a comprehensive survey on these results.
The content of this survey covers many recent developments about robustness analysis on finite systems, such as robust stable analysis of switched logical networks with time delays, under impulsive effects of a robust set stabilization of Boolean control networks, event-triggered control for robust set stabilization of logical networks with control inputs, stability analysis in distribution of probabilistic Boolean networks under function perturbation impact, and how to solve a disturbance decoupling problem by event triggered control for of logical control networks. Furthermore, this survey forecasts some research works in the future.
The rest of this manuscript is organized as follows. Section 2 contains some necessary preliminaries on STP and game theory. Section 3.1 introduces the recent developments about robust stable analysis of switched logical networks with time delays. Section 3.2 delineates the idea for under impulsive effects robust set stabilization of Boolean control networks. Section 3.3 describes event-triggered control for robust set stabilization of logical networks with control inputs. Section 3.4 gives stability analysis in distribution of probabilistic Boolean networks under function perturbation impacts. Section 3.5 recalls how to solve disturbance decoupling problems by event triggered control for logical control networks, which is followed by a brief conclusion in Section 4.
Notations: Rm×n denotes the set of m×n real matrices. R+m×n denotes the set of m×n nonnegative real matrices. Δn:={δin|i=1,2,⋯,n}, where δin is the i-th column of the identity matrix In. An n×t matrix M is called a logical matrix, if M=[δi1nδi2n⋯δitn], which is briefly denoted by M=δn[i1i2⋯it]. Define the set of n×t logical matrices as Ln×t. Coli(L) (Rowi(L)) is the i-th column (row) of matrix L. For a set E, |E| denotes the number of elements in E. r=(r1,⋯,rk)T∈Rk is called a probabilistic vector, if ri≥0, i=1,⋯,k, and ∑ki=1ri=1. The set of k dimensional probabilistic vectors is denoted by Υk. If M∈R+m×n and Col(M)⊂Υm, M is called a probabilistic matrix. The set of m×n probabilistic matrices is denoted by Υm×n.
In the beginning, we give the definition of the STP method.
Given two matrices A∈Rm×n and B∈Rp×t. The STP of them is defined as A⋉B=(A⊗Iαn)(B⊗Iαp), where α is the least common multiple of n and p and ⊗ is the Kronecker product (M∗N, as M∗N=[Col1(M)⋉Col1(N)Col2(M)⋉Col2(N)⋯Cols(M)⋉Cols(N)]∈Rpq×s.).
Note that, if n=p holds, the STP method is considered as the ordinary matrix product. In the rest of this paper, we simply call it "product" and omit the symbol "⋉" without confusion.
The theory of the STP method has many useful mathematic tools. We list four of them in the following. They will be useful throughout this paper.
Given X∈Rm, Y∈Rn, and define
Dp,qf=δp[1⋯1⏟q2⋯2⏟q⋯p⋯p⏟q], |
and
Dp,qr=δq[12⋯q⏟12⋯q⏟⋯12⋯q⏟⏟p], |
we have
● W[m,n]XY=YX holds,
● XA=(It⊗A)X holds,
● Dp,qfXY=X holds,
● Dp,qrXY=Y holds,
where W[m,n] denotes the swap matrix (especially W[n,n]:=W[n]).
In additional, we have a method to express a pseudo-logical f(x1,x2, ⋯,xn) is a mapping from Δnk to R function into its algebraic form.
Lemma 2.1. [11] Let f:Δnk→R (or f:Δnk→Δm) be a pseudo-logical function. Then there exists a unique structual matrix Mf, called the structural matrix of f, such that
f(x1,x2,⋯,xn)=Mf⋉ni=1xi, |
where xi∈Δk, i=1,2,⋯,n, Colj(Mf)=f(δjkn), and j=1,2,⋯,kn.
This section delineates robust analysis of switched logical networks with time delays (SDLNs) with all unstable modes. In the beginning, we introduce the description of SDLNs. Then, some new results about them are presented. Future works could generalize these results to switched logical networks with time delays under state-dependent delay and state constraints.
There is a classical SDLN with n state nodes, m disturbance inputs, and s subnetworks:
{x1(t+1)=gϱ(t)1(X(t−τ),⋯,X(t),Υ(t)),⋮xn(t+1)=gϱ(t)n(X(t−τ),⋯,X(t),Υ(t)), | (3.1) |
where τ∈Z+ is the state time delay, the switching signal is denoted by ϱ:N→S:={1,2,⋯,s}, X(t):=(x1(t),x2(t),⋯,xn(t)) is the states, Υ(t):=(γ1(t),γ2(t),⋯,γm(t))∈Dmk represents the disturbance input, and gij:Dn(τ+1)+mk→Dk, i=1,2,⋯,s denote k-valued logical functions.
Especially, the initial state trajectory Y0:=(X(−τ),X(−τ+1),⋯,X(0))∈Dn(τ+1)k, a switching sequence {ϱ(t):t∈N}⊆S, the disturbance signal {Υ(t):t∈N}⊆Dmk are given.
With the STP method, one can obtain the algebraic express of the above systems. Let x(t)=⋉nj=1xj(t)∈Δkn, γ(t)=⋉mj=1γj(t)∈Δkm and y(t)=⋉tj=t−τx(j)∈Δkn(τ+1). SDLN (3.1) is converted into the following equivalent algebraic expression:
x(t+1)=Kϱ(t)γ(t)y(t), | (3.2) |
where Kϱ(t)∈Lkn×kn(τ+1)+m.
Then, we rewrite (3.2) as the following form:
y(t+1)=ˆKϱ(t)γ(t)y(t), |
where
ˆKϱ(t)=Dr[kn,kn](Ikn(τ+1)⊗Kϱ(t))W[km,kn(τ+1)](Ikm⊗Mr,kn(τ+1))∈Lkn(τ+1)×kn(τ+1)+m. |
System (3.1) is equivalent to {the above system}. Given an equilibrium point Xe=(xe1,⋯,xen)∈Dnk, the vector form of Xe is xe=⋉nj=1xej:=δqkn. Let
ye=(xe)τ+1:=δαkn(τ+1). |
Letting ϱ(t)=i∼δis and using logical variables as in the vector form, we rewrite (3.2) into the following form:
x(t+1)=Kγ(t)ϱ(t)y(t), | (3.3) |
where
K:=[K1K2⋯Ks]W[km,s]∈Lkn×skn(τ+1)+m, |
and Ki∈Lkn×kn(τ+1)+m, are obtained from (3.2), and i=1,2,⋯,s.
The two following assumptions are the fundamental bases in this subsection. Further, these two assumptions always hold.
Assumption 1. Assume that:
1) all the modes of (3.1) are not robustly stable, i.e., all the modes of system (3.3) do not satisfy
Rowα(Qχi)=kmχ11kn(τ+1); |
2) the i∗-th mode of (3.3) satisfies ye=ˆKi∗γye, ∀ γ∈Δkm.
Note that, it should be pointed out that the above assumptions are necessary for stability analysis.
Similarly, one rewrite (3.3) into the following form:
x(t+1)=Kγ(t)ϱ(t)y(t), |
where
ˉK:=Dr[kn,kn](Ikn(τ+1)⊗¯K)W[skm,kn(τ+1)](Iskm⊗Mr,kn(τ+1))∈Lkn(τ+1)×skn(τ+1)+m. |
Via (3.4), we can analyze the robust stability for SDLN (3.1) with all unstable modes under by the {following controller}:
ϱ(t)=f(X(t−τ),X(t−τ+1),⋯,X(t)), | (3.4) |
where controller f:Dn(τ+1)k→S is to be constructed.
Firstly, there is the definition of robustly stable for SDLN (3.1) in the following.
Definition 3.1. System (3.1) is said to be robustly stable at Xe, if, there exists a state feedback controller (3.4) and χ∈Z+ such that X(t;Y0,ϱ,Υ)=Xe holds, ∀ initial trajectory Y0∈Dn(τ+1)k, disturbance input sequence {Υ(t):t∈N}⊆Dmk, and integer t≥χ.
Similarly, we convert switching signal (3.4) into the following form:
ϱ(t)=Fy(t), |
where F∈Ls×kn(τ+1) is state feedback gain.
Given μ∈Z+, there is the concept of μ-th step robustly stable for system (3.4).
Definition 3.2. Part with respect to ye, a set Dμ(ye)⊆Δkn(τ+1) is the μ-th step robustly stable, if there exists {ϱ(t):t∈N} such that y(t;y(0),ϱ,γ)=ye holds, ∀y(0)∈Dμ(ye), ∀{γ(t):t∈N}⊆Δkm, and ∀ integer t≥μ.
Definition 3.3. ([38]) Consider system (3.4). D⊆Δkn(τ+1), which is a nonempty set, is said to be one step robustly reachable from y∈Δkn(τ+1), if there exist a switching signal ϱ∈Δs and yj∈D, such that yj=ˆ¯Kδjkmϱy holds for any j=1,2,⋯,km, where ˆ¯K is structural matrix for controller and defined in [38].
In the following, there are five main results for robustly stability for SDLN (3.1) with all unstable modes.
Under Assumption 1, we get that
1) With respect to ye, Dμ(ye)={δψ1kn(τ+1),⋯,δψmμkn(τ+1)} is the μ-th step robustly stable part of system (3.4), if and only if
mμ∏i=1max |
2) With respect to y_e , D_{\mu}(y_{e}) = \{{\delta}_{k^{n(\tau+1)}}^{\psi_{1}}, \cdots, \; {\delta}_{k^{n(\tau+1)}}^{\psi_{m_{\mu}}}\} is the \mu -th step robustly stable part of system (3.4), if
\begin{equation} {\nonumber} \label{eq32.1-3} \prod\limits^{m_{\mu}}_{i = 1}(Q^{\mu})_{\alpha, \psi_{i}} > 0;{\nonumber} \end{equation} |
3) System (3.4) is robustly stable at y_e = {\delta}_{k^{n(\tau+1)}}^{\alpha} under state feedback controller (3.4), if and only if there exists a positive integer \chi\leq k^{n(\tau+1)} such that
\begin{equation} D_\chi(y_{e}) = {\Delta}_{k^{n(\tau+1)}}. \end{equation} | (3.5) |
Furthermore, based on the above results, we have more general results in the following:
1) System (3.1) is robustly stable at X_{e} under the state feedback controller (3.4), if and only if condition (3.5) holds.
2) System (3.1) without disturbances is stable at X_{e} under the state feedback controller (3.4), if and only if there exists an integer 1\leq\chi\leq k^{n(\tau+1)} such that
\begin{equation} {\nonumber} \label{eq4.77}D_\chi(y_{e}) = {\Delta}_{k^{n(\tau+1)}}.{\nonumber} \end{equation} |
For the complete proofs for the aforementioned results in this subsection, readers can see more details in [39].
This subsection introduces some recent works on, under impulsive effects, the robust set stabilization problem of Boolean control networks (BCNs).
For the complete proofs for the results in this section, readers can see more details in [40].
The classical definition for a Boolean control network under impulsive effects is
\begin{equation} \left\{\begin{array}{lll} x_{i}(t+1) = f_{1i}(X(t), U(t), \Xi(t)), \ i = 1, \cdots, n, t_{k-1}\leq t < t_{k}-1;\\ x_{i}(t_{k}) = f_{2i}(X(t_{k}-1), \Xi(t_{k}-1)), \ i = 1, \cdots, n, k\in \mathbb{Z};\\ y_{j}(t) = h_{j}(X(t)), \ j = 1, \cdots, p, \end{array}\right. \end{equation} | (3.6) |
where t_{0} = 0 , \{t_{k}: k\in \mathbb{Z}_+\}\subseteq \mathbb{Z}_+ satisfying 0 = t_0 < t_{1} < t_{2} < \cdots < t_{k} < \cdots is the impulsive time sequence, X(t) = (x_{1}(t), x_{2}(t), \cdots, x_{n}(t))\in\mathcal{D}^{n} , U(t) = (u_{1}(t), \cdots, u_{m}(t))\in\mathcal{D}^{m} , \Xi(t) = (\xi_{1}(t), \cdots, \xi_{q}(t))\in\mathcal{D}^{q} and Y(t) = (y_{1}(t), \cdots, y_{p}(t)) \; \in\mathcal{D}^{p} denote the state variables, the control inputs, the disturbance inputs and the outputs of the system (3.6), respectively, and f_{1i}: \mathcal{D}^{n+m+q}\rightarrow \mathcal{D} , f_{2i}: \mathcal{D}^{n+q}\rightarrow \mathcal{D}, i = 1, \cdots, n and h_{j}: \mathcal{D}^{n}\rightarrow \mathcal{D}, j = 1, \cdots, p are logical functions.
Now, there is the basic definition of robustly stabilizable to the set.
Definition 3.4. Consider a set A\subseteq\mathcal{D}^{n} , which is not empty. To the set A , BCN (3.6) is robustly stabilizable, if, one can find a suitable control sequence \{U(t):t\in \mathbb{N}\} and a integer \tau > 0 such that
\begin{equation} {\nonumber} \label{eq3.2} X(t; X(0), U, \Xi)\in A{\nonumber} \end{equation} |
holds for forall t\geq\tau , \forall X(0)\in\mathcal{D}^{n} , and \forall disturbance \{\Xi(t):t\in \mathbb{N}\}\subseteq{D}^{q} .
Normally, the state feedback control for the above systems can be written as
\begin{equation} {\nonumber} \label{eq3.5} u(t) = Gx(t), {\nonumber} \end{equation} |
where G\in\mathcal{L}_{2^m\times 2^{n}} is called the state feedback gain matrix.
The following definition is about robust L -invariant set.
Definition 3.5. Consider x(t+1) = L\xi(t)x(t) , where x(t)\in{\Delta}_{2^n} , \xi(t)\in{\Delta}_{2^q} , and L\in\mathcal{L}_{2^n\times 2^{n+q}} hold. Let S\subseteq{\Delta}_{2^n} be a nonempty set. S is a robust L -invariant set, if, L\xi x\in S holds for \forall x\in S and \forall \xi\in \Delta_{2^q} .
Then, we present two results. These two results reveal that how to design suitable control sequences under difference situations.
1) Consider a set A\subseteq{\Delta}_{2^n} , which is not empty. We assume that A is both a robust \bar{L}_{1} and a robust L_{2} -invariant set. For A , system (3.6) is said to be robustly stabilizable via controller u(t) = Gx(t) , if and only if, there exists a positive integer \tau s.t.
\begin{eqnarray} {} \label{eq3.9} Col(\tilde{L}_{\tau})\subseteq A{} \end{eqnarray} |
holds, where \tilde{L}_{\tau} is defined as
\begin{eqnarray} {} \label{eq3.8} \tilde{L}_{\tau} = \left\{ \begin{array}{lcr} \bar{L}_{1}(I_{2^q}\otimes\tilde{L}_{\tau-1}), \hbox{when}\ t_{k} < \tau < t_{k+1}, \\ L_{2}(I_{2^q}\otimes\tilde{L}_{\tau-1}), \hbox{when}\ \tau = t_{k+1}. \end{array}\right.{} \end{eqnarray} |
2) Consider a set A\subseteq{\Delta}_{2^n} , which is not empty. We assume that there exists a positive integer \alpha\leq 2^m s.t. A is a both robust Blk_{\alpha}(\hat{L}_{1}) -invariant set and L_{2} -invariant set. For set A , (3.6) is said to be robustly stabilizable under a free-form control sequence, if and only if, there exist two integers \tau > 0 and \beta > 0 such that
\begin{equation} Col(Blk_{\beta}(\hat{L}_{\tau}))\subseteq A. \end{equation} | (3.7) |
Furthermore, if (3.7) holds, then the control sequence is constructed as
\begin{eqnarray} {} \label{eq3.14} u(t) = \left\{ \begin{array}{lcr} u^{*}(t), \ t\in([0, \tau-1]\cap\mathbb{N})\setminus\Lambda(t), \\ \delta_{2^m}^\alpha, \ t\in([\tau, +\infty)\cap\mathbb{N})\setminus\Lambda(t), \end{array}\right.{} \end{eqnarray} |
where u^* is described by
\begin{eqnarray} {} \label{eq3.15} \left\{ \begin{array}{lcr} \ltimes_{i = \tau-1, i\not\in\Lambda(\tau)}^{0}u^{*}(i) = \delta_{2^{(\tau-k+1)m}}^\beta, \ \hbox{when}\ t_{k-1}+1\leq \tau < t_{k};\\ \ltimes_{i = \tau-2, i\not\in\Lambda(\tau)}^{0}u^{*}(i) = \delta_{2^{(\tau-k)m}}^\beta, \ \hbox{when}\ \tau = t_k. \end{array}\right.{} \end{eqnarray} |
This subsection focus on robust partial stabilization problem of (3.6). Letting (x_{1}^{*}, \cdots, x_{r}^{*})\in \mathcal{D}^{r} with r\leq n , one can obtain x^{r} = \ltimes_{i = 1}^{r}x_{i}^{*} = \delta_{2^r}^{\theta} .
The following is the definition of robustly partial stabilizable.
Definition 3.6. To x^{r} , system (3.6) is robustly partial stabilizable, if, one can find a control sequence \{u(t):t\in \mathbb{N}\}\subseteq\mathcal{D}^{m} and an integer \tau > such that
\begin{equation} {\nonumber} \label{eq4.1} x_{i}(t; x(0), u, \xi) = x_{i}^{*}, {\nonumber} \end{equation} |
holds for any integer t\geq\tau , \forall x(0)\in\mathcal{D}^{n} , \forall\{\xi(t):t\in \mathbb{N}\}\subseteq\mathcal{D}^{q} , and i = 1, \cdots, r .
According to the above definition, we present a necessary assumption.
Letting
\begin{equation} {\nonumber} \label{eq4.2} A = \{\delta_{2^r}^{\theta}\ltimes \delta_{2^{n-r}}^{\eta}: \eta = 1, \cdots, 2^{n-r}\}, {\nonumber} \end{equation} |
we assume that A is a robust both \bar{L}_{1} -invariant and robust L_{2} -invariant set. We would get some interesting results.
Theorem 3.7. For the system (3.6), the following statements are equivalent:
\rm 1) To x^{r} , (3.6) is robustly partial stabilizable under the state feedback controller u(t) = Gx(t) ;
\rm 2) To the set A , (3.6) is robustly stabilizable under the state feedback controller u(t) = Gx(t) ;
\rm 3) One can find an integer \tau > 0 such that Col(\tilde{L}_{\tau})\subseteq A holds.
This subsection investigates the output tracking problems of (3.6). Firstly, we give the basic definition of robust output tracking.
Definition 3.8. Given a reference trajectory Y_r = (y_1^r, \cdots, y_p^r)\in\mathcal{D}^{p} . The trajectory of (3.6) is said to robustly track trajectory Y_r , if, one can find a control sequence \{U(t):t\in \mathbb{N}\}\subseteq\mathcal{D}^m and \tau > 0 such that
\begin{equation} {\nonumber} \label{eq4.5} Y(t; X(0), U, \Xi) = Y_r{\nonumber} \end{equation} |
holds for \forall X(0)\in\mathcal{D}^{n} , \forall t\geq\tau , and \forall \{\Xi(t):t\in \mathbb{N}\}\subseteq\mathcal{D}^{q} .
The corresponding result about robust output tracking is in the following.
Theorem 3.9. Assume that \mathcal{O}(\beta) is a robust both \bar{L}_{1} -invariant and L_{2} -invariant set. The output trajectory of (3.6) robustly track y_r = \delta_{2^p}^\beta under the controller u(t) = Gx(t) , if and only if, one can find an integer \tau > 0 such that
\begin{equation} {\nonumber} \label{eq4.8} Col(H\tilde{L}_{\tau}) = \{\delta_{2^p}^\beta\}.{\nonumber} \end{equation} |
This subsection addresses some new developments about the event-triggered control problem for k -valued logical control networks (KVLCNs), and proposes an event-triggered control method.
The first part addresses some results about robust set stabilization of KVLCNs. The second part introduces some recent works about event-triggered control of KVLCNs. For the complete proofs for the results in this section, readers can see more details in [41].
There is a classical definition about k -valued logical control networks. A k -valued logical control network is described as follows:
\begin{equation} \left\{\begin{array}{llll} x_{i}(t+1) = f_{i}(X(t), U(t), \Xi(t)), \ i = 1, \cdots, n;\\ y_{j}(t) = h_{j}(X(t)), \ j = 1, \cdots, p, \end{array}\right. \end{equation} | (3.8) |
where X(t) = (x_{1}(t), x_{2}(t), \cdots, x_{n}(t))\in\mathcal{D}_k^{n} , U(t) = (u_{1}(t), \cdots, u_{m}(t))\in\mathcal{D}_k^{m} , \Xi(t) = (\xi_1(t), \cdots, \xi_q(t)) \in \mathcal{D}_k^{q} and Y(t) = (y_{1}(t), \cdots, y_{p}(t)) \; \in\mathcal{D}_k^{p} are states, control inputs, disturbance inputs and outputs at time t , respectively, and f_{i}: \mathcal{D}_k^{n+m+q}\rightarrow \mathcal{D}_k , i = 1, \cdots, n and h_{j}: \mathcal{D}_k^{n}\rightarrow \mathcal{D}_k are logical functions. Given X(0)\in\mathcal{D}_k^n , a control \{U(t): t\in\mathbb{N}\}\subseteq\mathcal{D}_k^{m} and a disturbance inputs \{\Xi(t): t\in\mathbb{N}\}\subseteq\mathcal{D}_k^{q} , denote the trajectory of system (3.8) by X(t; X(0), U, \Xi) , j = 1, \cdots, p , the numbers of nodes, control inputs, outputs, and disturbances are n , m , p and q .
According to the above definition, we propose the definition of robust set stabilization for k -valued logical control networks (3.8) in the following.
Definition 3.10. Define a set A\subseteq\mathcal{D}_k^{n} , which is nonempty, and let X(0)\in\mathcal{D}_k^{n} . To A , (3.8) is robustly stabilizable, if one can find a control sequence \{U(t):t\in \mathbb{N}\}\subseteq\mathcal{D}_k^{m} and an integer \tau > 0 such that
\begin{equation} {\nonumber} \label{eq2.2} X(t; X(0), U, \Xi)\in A{\nonumber} \end{equation} |
holds, for any t\geq\tau and any \{\Xi(t):t\in \mathbb{N}\}\subseteq\mathcal{D}_k^{q} .
Then, we rewrite k -valued logical control networks (3.8) into an algebraic form step by step.
Using the vector form of logical variables and setting x(t) = \ltimes_{i = 1}^{n}x_{i}(t)\in{\Delta}_{k^{n}} , u(t) = \ltimes_{i = 1}^{m}u_{i}(t)\in{\Delta}_{k^{m}} , \xi(t) = \ltimes_{i = 1}^{q}\xi_i(t)\in{\Delta}_{k^{q}} and y(t) = \ltimes_{i = 1}^{p}y_{i}(t)\in{\Delta}_{k^{p}} , by STP method, k -valued logical control networks (3.8) can be rewritten into the following equivalent algebraic form:
\begin{eqnarray} \left\{ \begin{array}{lcr} x(t+1) = Lu(t)x(t)\xi(t), \\ y(t) = Hx(t), \end{array}\right. \end{eqnarray} | (3.9) |
where L\in\mathcal{L}_{k^{n}\times k^{n+m+q}} is the state transition matrix, and H\in\mathcal{L}_{k^{p}\times k^{n}} is the output matrix.
This subsection considers the following state feedback control:
\begin{equation} u_i(t) = \psi_i(t, X(t)), i = 1, \cdots, m, \end{equation} | (3.10) |
where \psi_i: \mathbb{N}\times\mathcal{D}_k^{n}\rightarrow \mathcal{D}_k are logical functions and i = 1, 2 \cdots, m .
For \forall\psi_i , we can construct a unique structural matrix \Psi_i(t, x(0))\in\mathcal{L}_{k\times k^n} such that u_i(t) = \Psi_i(t, x(0))x(t) . By the Khatri-Rao product of matrices, we construct a time-variant structural matrix \Psi(t, x(0))\in\mathcal{L}_{k^{m}\times k^{n}} such that
\begin{equation} {\nonumber} \label{eq2.5} u(t) = \Psi(t, x(0))x(t), {\nonumber} \end{equation} |
where \Psi(t, x(0)) = \Psi_1(t, x(0))\ast\cdots\ast\Psi_m(t, x(0)) and i = 1, 2, \cdots, m .
A sufficient and necessary criterion for the problem of robust set stabilization for (3.8) under the controller (3.10) is given in the following.
Theorem 3.11. Define a set A\subseteq\mathcal{D}_k^{n} , which is nonempty, and let x(0) = \delta_{k^n}^\alpha and A\subseteq\Upsilon_1(A) . k -valued logical control networks (3.8) is said to be robustly stabilizable under the controller (3.10), if and only if, one can find an integer T > 0 such that x(0)\in\Upsilon_T(A) holds.
We still consider system (3.9). Define a set A\subseteq\mathcal{D}_k^{n} , which is nonempty, and let x(0) = \delta_{k^n}^\alpha be initial state. For a given controller u(t) = \Psi(t, x(0))x(t) , one has
\begin{eqnarray} {} \label{eq2.13}{} &&x(t+1) = Lu(t)x(t)\xi(t)\\{} & = &\ltimes_{i = t}^0(L\Psi(i, x(0))M_{r, k^n})x(0)\ltimes_{j = 0}^{t}\xi(j)\\{} & = &Blk_\alpha\Big(\ltimes_{i = t}^0(L\Psi(i, x(0))M_{r, k^n})\Big)\ltimes_{j = 0}^{t}\xi(j), {} \end{eqnarray} |
where t\in\mathbb{N} , and M_{r, k^n} = {\mbox{Diag}}\{\delta_{k^n}^1, \cdots, \delta_{k^n}^{k^n}\}\in\mathcal{L}_{k^{2n}\times k^n} .
From the arbitrariness of \ltimes_{j = 0}^{t}\xi(j) , x(t+1) forms a set
\begin{equation} {\nonumber} \label{eq2.14} \Omega(t+1) = Col\Big(Blk_\alpha\Big(\ltimes_{i = t}^0(L\Psi(i, x(0))M_{r, k^n})\Big)\Big).{\nonumber} \end{equation} |
Then, we give the event-triggered condition as
\begin{equation} d_H(\Omega(t+1), A) > 0, \end{equation} | (3.11) |
where d_H(\Omega(t+1), A) denotes the typical Hausdorff distance.
For system (3.9), split L into k^m equal blocks as
L = [L_1\; \; L_2\; \; \cdots\; \; L_{k^m}], |
where L_i\in\mathcal{L}_{k^{n}\times k^{n+q}} , i = 1, 2, \cdots, k^m . For any i\in\{1, 2, \cdots, k^m\} , split L_i into k^n equal blocks as
L_i = [L_{i, 1}\; \; L_{i, 2}\; \; \cdots\; \; L_{i, k^n}], |
where L_{i, j}\in\mathcal{L}_{k^{n}\times k^{q}} , j = 1, 2, \cdots, k^n .
In the following, there exist a sufficient condition for the existence of event-triggered controller, and an method to construct the corresponding controller.
Theorem 3.12. Define a set A\subseteq\mathcal{D}_k^{n} , which is nonempty, and let x(0) = \delta_{k^n}^\alpha . k -valued logical control networks (3.8) is said to be robustly stabilizable with the event-triggered condition (3.11), if A\subseteq\Upsilon_1(A) and x(0)\in\Upsilon_1(A) hold.
This subsection introduces some results about robust analysis of probabilistic Boolean networks (PBNs).
For the complete proofs for the results in this subsection, readers can see more details in [42].
There is a classical definition for PBNs with n nodes:
\begin{equation} X(t+1) = f(X(t)), \end{equation} | (3.12) |
where X(t) = (x_1(t), \cdots, x_n(t))\in\mathcal{D}^n is the state vector of PBN (3.12), and f: \mathcal{D}^n\rightarrow \mathcal{D}^n is a logical mapping which is chosen from the set \{f_1, f_2, \cdots, f_r\} with \mathbb{P}\{f = f_i\} = p_i , \sum_{i = 1}^rp_i = 1 .
Using the vector form and setting x(t) = \ltimes_{i = 1}^nx_i(t)\in \Delta_{2^n} , PBN (3.12) can be converted to the following algebraic expression:
\begin{equation} {\nonumber} \label{eq3.11} x(t+1) = Lx(t), {\nonumber} \end{equation} |
where L\in\mathcal{L}_{2^n\times 2^n} is the structural matrix of f , which is chosen from the set \{L_1, \cdots, L_r\} with \mathbb{P}\{L = L_i\} = p_i , and L_i = \delta_{2^n}[\alpha_{i, 1} \; \cdots\; \alpha_{i, 2^n}]\in\mathcal{L}_{2^n\times 2^n} is the structural matrix of f_i .
There are two common definitions. Readers can see more details in [43].
We give the basic definitions of stability and set stability for (3.12) in the following.
Definition 3.13. PBN (3.12) is stable at x_{e} = \delta_{2^n}^\theta in distribution, if, \lim_{t\rightarrow \infty}\mathbb{P}\{x(t; x_0) = x_e\} = 1 holds for \forall x_0\in \Delta_{2^n} .
Definition 3.14. PBN (3.12) is stable at a given nonempty set \mathcal{M}\subseteq \Delta_{2^n} in distribution, if \lim_{t\rightarrow \infty}\mathbb{P}\{x(t ; x_0)\in \mathcal{M}\} = 1 holds for any x_0\in \Delta_{2^n} .
To make this subsection more readable, we give an example to explain the difference between one-bit function and multi-bit function perturbation.
Example 3.15. Consider the following BN:
\begin{eqnarray} {\nonumber} \left\{ \begin{array}{lcr} x_1(t+1) = f_1(x_1(t), x_2(t)), \\{\nonumber} x_2(t+1) = f_2(x_1(t), x_2(t)), {\nonumber} \end{array}\right. \end{eqnarray} |
where f_1 = x_1\wedge\neg x_2 and f_2 = x_1 . The truth table of system (3.13) is given in Table 1.
x_1 | x_2 | f_1 | f_2 |
1 | 1 | 0 | 1 |
1 | 0 | 1 | 1 |
0 | 1 | 0 | 0 |
0 | 0 | 0 | 0 |
The algebraic equation of PBN (3.13) is x(t+1) = Fx(t) , in which F: = {\delta}_4[3\; 1\; 4\; 4] .
First, we define a one-bit perturbation, that is, f_1(0, 1) is changed from 0 to 1 . Then, F is flipped to {\delta}_4[3\; 1\; 2\; 4] , i.e., Col_3(F) is flipped from {\delta}_4^4 to {\delta}_4^2 .
Second, consider a multi-bit perturbation, e.g., f_1(0, 1) and f_2(0, 1) are changed from 0 to 1 , separately. Thus, F is flipped to {\delta}_4[3\; 1\; 1\; 4] . We can find that Col_3(F) is flipped from {\delta}_4^4 to {\delta}_4^1 .
The objective of this section is to propose some criteria. These is very useful in guaranteing the robustness of (3.12), i.e., PBN (3.12) can be still stable at x_e or \mathcal{M} after function perturbation.
We assume that PBN (3.12) is stable at x_e = \delta_{2^n}^\theta in distribution. There is a common assumption for the following results in this subsection.
Assumption 2. For \forall i\in\{1, \cdots, r\} , the \mathcal{B} -th column of L_i is perturbed with \mathcal{B}\neq \theta . We assume that \alpha_{i, \mathcal{B}} flipped to some \gamma\in \{1, \cdots, 2^n\} , where \gamma\neq \alpha_{i, \mathcal{B}} .
According to Assumption 2, we construct a set
\begin{eqnarray} {} \label{eq3.20} \Omega = \{x:\ x\ \hbox{reach}\ x_e\ \hbox{with positive probability}, {}\\ \hbox{meanwhile any path from}\ x\ \hbox{to}\ x_e\ \hbox{must cover}\ \delta_{2^n}^{\mathcal{B}}\}.{} \end{eqnarray} |
Thus, one has the following result.
Theorem 3.16. Given PBN (3.12) is stable at x_e . Under Assumption 2, PBN (3.12) is said to be stable at x_e after function perturbation, if and only if, \delta_{2^n}^{\gamma} \not\in \Omega holds.
One can see from Theorem 3.16 that \delta_{2^n}^{\gamma} \not\in \Omega is very important for the stability in distribution of system (3.12). Then, by the transition probability matrix M , we construct
\begin{eqnarray} {} \label{eq3.29} \varphi(\mathcal{B}, \gamma, \theta)& = &\sum\limits_{k = 2}^{2^n}(M^k)_{\theta, \gamma}-\sum\limits_{k = 2}^{2^n}\sum\limits_{s = 1}^{k-1}(M^{k-s})_{\theta, \mathcal{B}}(M^s)_{\mathcal{B}, \gamma}.{} \end{eqnarray} |
We have the following result on the verification of \delta_{2^n}^{\gamma} \not\in \Omega .
Theorem 3.17. \delta_{2^n}^{\gamma} \not\in \Omega , if and only if
\begin{equation} {\nonumber} \label{eq3.25} \varphi(\mathcal{B}, \gamma, \theta) > 0.{\nonumber} \end{equation} |
This subsection assumes that system (3.12) is stable at a given nonempty set \mathcal{M}\in \Delta_{2^n} . There is a common used assumption in the following.
Assumption 3. For \forall i\in\{1, \cdots, r\} , Col_\mathcal{B}(L_i) doesn't belong to the set I(\mathcal{M}) . We assume that \alpha_{i, \mathcal{B}} changes to some \gamma\in \{1, \cdots, 2^n\} , where \gamma\neq \alpha_{i, \mathcal{B}} .
Define
\begin{eqnarray} {} \label{eq3.30}{} \varphi(\mathcal{B}, \gamma, \mathcal{M})& = &\sum\limits_{{\delta}_{2^n}^\theta \in I(\mathcal{M})}\varphi(\mathcal{B}, \gamma, \theta)\\{} & = &\sum\limits_{{\delta}_{2^n}^\theta \in I(\mathcal{M})}\Big[\sum\limits_{k = 2}^{2^n}(M^k)_{\theta, \gamma}-\sum\limits_{k = 2}^{2^n} \sum\limits_{s = 1}^{k-1}(M^{k-s})_{\theta, \mathcal{B}}(M^s)_{\mathcal{B}, \gamma}\Big].{} \end{eqnarray} |
One has:
Theorem 3.18. Consider PBN (3.12) is said to be stable at a given set \mathcal{M} , which is nonempty, in distribution. Under Assumption 3, PBN (3.12) is still said to be stable at \mathcal{M} after function perturbation, if and only if, \varphi(\mathcal{B}, \gamma, \mathcal{M}) > 0 holds.
This subsection introduces some recent development about event-triggered control for disturbance decoupling problem of mix-valued logical networks (MVLCNs).
There is a classical definition of MVLCNs:
\begin{equation} \left\{ \begin{array}{lcr} x_{1}(t+1) = f_{1}(X(t), U(t), \Xi(t)), \\ \vdots \\ x_{n}(t+1) = f_{n}(X(t), U(t), \Xi(t));\\ y_j(t) = g_j(X(t)), \; \; \; j = 1, \cdots, p, \end{array}\right. \end{equation} | (3.13) |
where X(t) = (x_{1}(t), \cdots, x_{n}(t)) with x_i(t)\in\mathcal{D}_{k_i} denotes states, U(t) = (u_{1}(t) , \cdots, u_{m}(t)) with u_i(t)\in\mathcal{D}_{l_i} denotes controls, \Xi(t) = (\xi_{1}(t), \cdots, \xi_{r}(t)) with \xi_{i}(t)\in\mathcal{D}_{v_i} denoting disturbance inputs, and y_j(t)\in\mathcal{D}_{w_j} , denotes outputs, and j = 1, 2, \cdots, p . Define k: = k_1\cdots k_n , l: = l_1\cdots l_m , v: = v_1\cdots v_r and w: = w_1\cdots w_p .
Define an n -ary logical function h: \mathcal{D}_{k_1}\times\cdots\times\mathcal{D}_{k_n}\rightarrow \mathcal{D}_{k_0} . To convert h into an equivalent algebraic form, we identify \frac{k-i}{k-1}\sim \delta_k^i , i = 1, \cdots, k . Then, we have \mathcal{D}_k\sim \Delta_k . \delta_k^i is called the vector form of logical value \frac{k-i}{k-1}\in\mathcal{D}_k .
Based on the vector form of logical values and defining x(t) = \ltimes_{i = 1}^n x_i(t) , u(t) = \ltimes_{i = 1}^m u_i(t) , \xi(t) = \ltimes_{i = 1}^r \xi_i(t) and y(t) = \ltimes_{j = 1}^p y_j(t) , one gets the following expression of (3.13):
\begin{eqnarray} \left\{ \begin{array}{lcr} x(t+1) = Lu(t)x(t)\xi(t), \\ y(t) = Gx(t), \end{array}\right. \end{eqnarray} | (3.14) |
where L\in\mathcal{L}_{k\times (klv)} and G\in\mathcal{L}_{w\times k} .
In the next, use the method in [44] for coordinate transformation.
We assume that there is a logical coordinate transformation
\begin{equation} \{x_i: i = 1, \cdots, n\}\rightarrow \{z_i: i = 1, \cdots, n\}, \; z_i\in \mathcal{D}_{k_{\alpha_i}}, \end{equation} | (3.15) |
under which system (3.13) becomes
\begin{equation} \left\{ \begin{array}{lcr} z_i(t+1) = \hat{f}_i^1(Z(t), U(t), \Xi(t)), \; i = 1, \cdots, s, \\ z_i(t+1) = \hat{f}_i^2(Z(t), U(t), \Xi(t)), \; i = s+1, \cdots, n;\\ y_j(t) = \hat{g}_j(z_1(t), \cdots, z_s(t)), \; \; \; \; \; \; j = 1, \cdots, p, \end{array}\right. \end{equation} | (3.16) |
where \hat{f}_i^1 , \hat{f}_i^2 , i = s+1, s+2, \cdots, n , \hat{g}_j , j = 1, \cdots, p are logical functions, and i = 1, 2 \cdots, s
Let z(t) = \ltimes_{i = 1}^n z_i(t) . For any \mu\in\{1, \cdots, n\} , set \mathbf{z}_\mu(t) = \ltimes_{i = 1}^\mu z_i(t) , \mathbf{k}_\mu = \ltimes_{i = 1}^\mu k_{\alpha_i} . For any \mu\in\{1, \cdots, n-1\} , set \mathbf{z}_{-\mu}(t) = \ltimes_{i = \mu+1}^n z_i(t) , and \mathbf{k}_{-\mu} = \ltimes_{i = \mu+1}^n k_{\alpha_i} . Next, set \mathbf{k}_{-n} = 1 . Thus, we can find that \mathbf{z}_n(t) = z(t) and \mathbf{k}_n = k . System (3.16) can be rewritten as in the following algebraic form:
\begin{eqnarray} \left\{ \begin{array}{lcr} \mathbf{z}_s(t+1) = \hat{L}_su(t)z(t)\xi(t), \\ \mathbf{z}_{-s}(t+1) = \hat{L}_{-s}u(t)z(t)\xi(t);\\ y(t) = \hat{G}\mathbf{z}_s(t), \end{array}\right. \end{eqnarray} | (3.17) |
where \hat{L}_s\in \mathcal{L}_{\mathbf{k}_s\times (klv)} , \hat{L}_{-s}\in \mathcal{L}_{\mathbf{k}_{-s}\times (klv)} and \hat{G}\in \mathcal{L}_{w\times \mathbf{k}_s} .
When s = n holds, MVLCN (3.17) is
\begin{eqnarray} {} \left\{ \begin{array}{lcr} z(t) = \hat{L}u(t)z(t)\xi(t);\\ y(t) = \hat{G}z(t), \end{array}\right. \end{eqnarray} |
in which \hat{L}\in \mathcal{L}_{k\times (klv)} and \hat{G}\in \mathcal{L}_{w\times k} .
Consider MVLCN (3.16) with z(0)\in \mathcal{D}_{k_{\alpha_1}}\times\cdots\times\mathcal{D}_{k_{\alpha_n}} , we can construct the following controller with respect to z(0) as:
\begin{equation} u_i(t) = k_t^i(z_1(t), \cdots, z_n(t)), \ i = 1, \cdots, m. \end{equation} | (3.18) |
For \forall t\in \mathbb{N} , under controller (3.18), one can find all possible invariant subspaces as \mathcal{Z}_{\sigma_h(t)} = F_l\{z_1, \cdots, z_{\sigma_h(t)}\}\supseteq F_l\{z_1, \cdots, z_s\} , that is, system (3.16) becomes
\begin{eqnarray} {} \label{eq6} \left\{ \begin{array}{lcr} z_i(t+1) = \tilde{f}_i^1(z_1(t), \cdots, z_{\sigma_h(t)}(t)), \; \, \; \; i = 1, \cdots, \sigma_h(t), \\{\nonumber} z_i(t+1) = \tilde{f}_i^2(z_1(t), \cdots, z_n(t), \Xi(t)), \; i = \sigma_h(t)+1, \cdots, n;\\{\nonumber} y_j(t) = \tilde{g}_j(z_1(t), \cdots, z_s(t)), \; \; \; \; \, \; \; \; \, \; \; \; \; \; j = 1, \cdots, p, {\nonumber} \end{array}\right. \end{eqnarray} |
where h\in\{1, \cdots, \lambda_{t}\} , \lambda_{t}\in\{1, \cdots, n-s\} , and \tilde{f}_i^1 , i = 1, \cdots, \sigma_h(t) , \tilde{f}_i^2 , i = \sigma_h(t)+1, \cdots, n , \tilde{g}_j , j = 1, 2, \cdots, p are mix-valued logical functions. For \forall t\in \mathbb{N} , denote
\begin{equation} {\nonumber} \label{eq2} \Gamma_t = \{\sigma_h(t): h = 1, \cdots, \lambda_{t}\}, {\nonumber} \end{equation} |
and set
\begin{equation} {\nonumber} \label{eq28} \sigma(t) = \max\{\sigma_h(t): h = 1, \cdots, \lambda_{t}\}.{\nonumber} \end{equation} |
It is obvious that \sigma_h(t)\in \{s, \cdots, \sigma(t-1)\} with \sigma(-1): = n .
In this subsection, we propose the following concept of disturbance decoupling of logical networks.
Definition 3.19. Given the transformation (3.15) be given. With respect to initial state z(0)\in \Delta_k , the disturbance decoupling problem is solvable, if there exists a controller (3.18) corresponding to z(0) such that \Gamma_t\neq\emptyset holds, \forall t\in \mathbb{N} . The disturbance decoupling problem is solvable, if, with respect to \forall z(0)\in \Delta_k , it is solvable.
We consider (3.14) and assume
\begin{equation} {\nonumber} \label{eq11} G = \delta_{w}[\gamma_1\; \; \gamma_2\; \; \cdots\; \; \gamma_{k}].{\nonumber} \end{equation} |
Denote
\begin{eqnarray} {} \label{eq11-1} \eta_j = \Big|\{i: \gamma_i = j, 1\leq i\leq k\}\Big|, \; \; j = 1, 2, \cdots, w, {} \end{eqnarray} |
where |\cdot| is the number of sets. Then, we have the following definition. For details, please refer to [44].
Definition 3.20. Let H = (h_1, \cdots, h_p): \mathcal{D}_{k_{1}}\times\cdots\times\mathcal{D}_{k_{n}}\rightarrow \mathcal{D}_{w_1}\times\cdots\times\mathcal{D}_{w_p} be a mix-valued logical mapping. The variable x_i is said to be redundant, if H(x_1, \cdots, x_{i-1}, 0, x_{i+1}, \cdots, x_n) = H(x_1, \cdots, x_{i-1}, \frac{\zeta}{k_i-1}, x_{i+1}, \cdots, x_n) holds for any \zeta = 1, \cdots, k_i-1 .
Then, we have
Theorem 3.21. Consider a logical mapping H = (h_1, \cdots, h_p): \mathcal{D}_{k_{1}}\times\cdots\times\mathcal{D}_{k_{n}}\rightarrow \mathcal{D}_{w_1}\times\cdots\times\mathcal{D}_{w_p} . Let M_H\in \mathcal{L}_{w\times k} be the structural matrix of H , and let an integer s\leq n be given. Split M_H into k' = k_1\cdots k_s equal blocks as M_H = [M_H^1\; \; M_H^2\; \; \cdots\; \; M_H^{k'}] , where M_H^1, M_H^2, \cdots, M_H^{k'}\in\mathcal{L}_{w\times \frac{k}{k'}} . Then, (x_{s+1}, \cdots, x_{n}) are redundant variables if and only if \hbox{rank}(M_H^i) = 1 holds for any i = 1, \cdots, k' .
After that, it is easy to obtain that
\begin{equation} \mathbf{z}_s(t) = F[\mathbf{k}_s, \mathbf{k}_{-s}]Tx(t): = T_0x(t). \end{equation} | (3.19) |
This paper has described a comprehensive survey on some recent applications of the STP method on the theory of finite systems. After we introduced some useful mathematical tools on the STP method, some recent developments about robustness analysis on finite systems are delineated, such as robust stable analysis of switched logical networks with time-delayed, under impulsive effects robust set stabilization of Boolean control networks, event-triggered control for robust set stabilization of logical networks with control inputs, stability analysis in distribution of probabilistic Boolean networks under functional perturbation impact and how to solve disturbance decoupling problems by event triggered control of logical control networks have been presented.
Furthermore, the STP method is a generalization of ordinal products of matrices. It is inevitable to keep some shortcomings in ordinal product of matrices. One of them is that the dimensions of the structural matrices increase too rapidly. Thus, it leads to the calculation complexity's exponential growth. There are only a few results [45,46] about that. Our future plan is to solve it. In addition, we can extend the existing results about logical control networks to the other systems. It also is a great research area for scholars, and readers can see [47,48,49,50,51], such as finite-time stability and settling-time estimation of nonlinear impulsive systems, nonlinear systems with delayed impulses and so on.
The National Natural Science Foundation of China under grants 62273216 and 61873150, the Natural Science Fund for Distinguished Young Scholars of Shandong Province under grant JQ201613, and the Young Experts of Taishan Scholar Project.
We declare that we have no financial and personal relationships with other people or organizations that can inappropriately influence our work, there is no professional or other personal interest of any nature or kind in any product, service and/or company that could be construed as influencing the position presented in, or the review of, the manuscript entitled, "survey of semi-tensor product method in robustness analysis on finite systems".
[1] |
S. Kauffman, Metabolic stability and epigenesis in randomly constructed genetic net, J. Theor. Biol., 22 (1969), 437. https://doi.org/10.4327/jsnfs1949.22.437 doi: 10.4327/jsnfs1949.22.437
![]() |
[2] | D. Cheng, H. Qi, Z. Li, Analysis and Control of Boolean Networks, Springer London, 2011. https://doi.org/10.1007/978-0-85729-097-7 |
[3] |
G. Zhao, Y. Wang, Formulation and optimization control of a class of networked evolutionary games with switched topologies, Nonlinear Anal. Hybrid Syst., 22 (2016), 98–107. https://doi.org/10.1016/j.nahs.2016.03.009 doi: 10.1016/j.nahs.2016.03.009
![]() |
[4] |
G. Zhao, Y. Wang, H. Li, Invertibility of higher order k-valued logical control networks and its application in trajectory control, J. Franklin Inst. Eng. Appl. Math., 353 (2016), 4667–4679. https://doi.org/10.1016/j.jfranklin.2016.07.004 doi: 10.1016/j.jfranklin.2016.07.004
![]() |
[5] |
G. Zhao, H. Li, T. Hou, Input-output dynamical stability analysis for cyber-physical systems via logical networks, IET Control Theory Appl., 14 (2020), 2566–2572. https://doi.org/10.1049/iet-cta.2020.0197 doi: 10.1049/iet-cta.2020.0197
![]() |
[6] | B. Li, J. Lou, Y. Liu, Z. Wang, Robust invariant set analysis of Boolean networks, Complexity, 2019 (2019). https://doi.org/10.1155/2019/2731395 |
[7] |
H. Li, Y. Wang, L. Xie, Output tracking control of Boolean control networks via state feedback, constant reference signal case, Automatica, 59 (2015), 54–59. https://doi.org/10.1016/j.automatica.2015.06.004 doi: 10.1016/j.automatica.2015.06.004
![]() |
[8] |
H. Li, Y. Wang, Controllability analysis and control design for switched Boolean networks with state and input constraints, SIAM J. Control Optim., 53 (2015), 2955–2979. https://doi.org/10.1137/120902331 doi: 10.1137/120902331
![]() |
[9] | D. Cheng, H. Qi, Analysis and control of Boolean networks, a semi-tensor product approach, Acta Autom. Sin., 37 (2011), 529–540. |
[10] |
B. Li, J. Lu, Y. Liu, Z. Wu, The outputs robustness of Boolean control networks via pinning control, IEEE Trans. Control Network Syst., 7 (2019), 201–209. https://doi.org/10.1109/TCNS.2019.2913543 doi: 10.1109/TCNS.2019.2913543
![]() |
[11] |
H. Li, Y. Wang, Z. Liu, A semi-tensor product approach to pseudo-Boolean functions with application to Boolean control networks, Asian J. Control, 16 (2014), 1073–1081. https://doi.org/10.1002/asjc.767 doi: 10.1002/asjc.767
![]() |
[12] |
H. Li, Y. Wang, Lyapunov-based stability and construction of Lyapunov functions for Boolean networks, SIAM J. Control Optim., 55 (2017), 3437–3457. https://doi.org/10.1137/16M1092581 doi: 10.1137/16M1092581
![]() |
[13] |
H. Li, X. Ding, A control Lyapunov function approach to feedback stabilization of logical control networks, SIAM J. Control Optim., 57 (2019), 810–831. https://doi.org/10.1137/18M1170443 doi: 10.1137/18M1170443
![]() |
[14] |
H. Li, X. Xu, X. Ding, Finite-time stability analysis of stochastic switched Boolean networks with impulsive effect, Appl. Math. Comput., 347 (2019), 557–565. https://doi.org/10.1016/j.amc.2018.11.018 doi: 10.1016/j.amc.2018.11.018
![]() |
[15] |
H. Li, Y. Zheng, F. E. Alsaadi, Algebraic formulation and topological structure of Boolean networks with state-dependent delay, J. Comput. Appl. Math., 350 (2019), 87–97. https://doi.org/10.1016/j.cam.2018.10.003 doi: 10.1016/j.cam.2018.10.003
![]() |
[16] |
J. Lu, M. Li, Y. Liu, D. W. C. Ho, J. Kurths, Nonsingularity of Grain-like cascade FSRs via semi-tensor product, Sci. China Inf. Sci., 61 (2018), 1–12. https://doi.org/10.1007/s11432-017-9269-6 doi: 10.1007/s11432-017-9269-6
![]() |
[17] |
J. Lu, M. Li, T. Huang, Y. Liu, J. Cao, The transformation between the Galois NLFSRs and the Fibonacci NLFSRs via semi-tensor product of matrices, Automatica, 96 (2018), 393–397. https://doi.org/10.1016/j.automatica.2018.07.011 doi: 10.1016/j.automatica.2018.07.011
![]() |
[18] |
H. K. Ji, Y. Y. Li, X. Y. Ding, J. Q. Lu, Stability analysis of Boolean networks with Markov jump disturbances and their application in apoptosis networks, Electron. Res. Arch., 30 (2022), 3422–3434. https://doi.org/10.3934/era.2022174 doi: 10.3934/era.2022174
![]() |
[19] | D. Y. Zhong, Y. Y. Li, J. Q. Lu, Feedback stabilization of Boolean control networks with missing data, IEEE Trans. Neural Networks Learn. Syst., 2022 (2022). https://doi.org/10.1109/TNNLS.2022.3146262 |
[20] |
S. Liang, G. Zhao, H. Li, X. Ding, Structural stability analysis of gene regulatory networks modeled by Boolean networks, Math. Methods Appl. Sci., 42 (2019), 2221–2230. https://doi.org/10.1002/mma.5488 doi: 10.1002/mma.5488
![]() |
[21] | X. Li, J. Lu, J. Qiu, X. Chen, X. Li, F. E. Alsaadi, Set stability for switched Boolean networks with open-loop and closed-loop switching signals, Sci. China Inf. Sci., 61 (2018), 092207. |
[22] |
X. Li, J. Lu, X. Chen, J. Qiu, J. Cao, F. E. Alsaadi, Robust output tracking of delayed Boolean networks under pinning control, IEEE Trans. Circuits Syst. II Express Briefs, 65 (2018), 1249–1253. https://doi.org/10.1109/TCSII.2018.2827931 doi: 10.1109/TCSII.2018.2827931
![]() |
[23] |
Y. Liu, B. Li, H. Chen, J. Cao, Function perturbations on singular Boolean networks, Automatica, 84 (2017), 36–42. https://doi.org/10.1016/j.automatica.2017.06.035 doi: 10.1016/j.automatica.2017.06.035
![]() |
[24] |
Y. Liu, Y. Zheng, H. Li, F. E. Alsaadi, B. Ahmad, Control design for output tracking of delayed Boolean control networks, J. Comput. Appl. Math., 327 (2018), 188–195. https://doi.org/10.1016/j.cam.2017.06.016 doi: 10.1016/j.cam.2017.06.016
![]() |
[25] |
Y. Li, B. Li, Y. Liu, J. Lu, Z. Wang, F. E. Alsaadi, Set stability and set stabilization of switched Boolean networks with state-based switching, IEEE Access, 6 (2018), 35624–35630. https://doi.org/10.1109/ACCESS.2018.2851391 doi: 10.1109/ACCESS.2018.2851391
![]() |
[26] | Y. Li, J. Zhong, J. Lu, Z. Wang, On robust synchronization of drive-response Boolean control networks with disturbances, Math. Probl. Eng., 2018 (2018). https://doi.org/10.1155/2018/1737685 |
[27] |
Y. Mao, L. Wang, Y. Liu, J. Lu, Z. Wang, Stabilization of evolutionary networked games with length-r information, Appl. Math. Comput., 337 (2018), 442–451. https://doi.org/10.1016/j.amc.2018.05.027 doi: 10.1016/j.amc.2018.05.027
![]() |
[28] |
L. Tong, Y. Liu, Y. Li, J. Lu, Z. Wang, F. E. Alsaadi, Robust control invariance of probabilistic Boolean control networks via event-triggered control, IEEE Access, 6 (2018), 37767–37774. https://doi.org/10.1109/ACCESS.2018.2828128 doi: 10.1109/ACCESS.2018.2828128
![]() |
[29] |
X. Xu, H. Li, Y. Li, F. E. Alsaadi, Output tracking control of Boolean control networks with impulsive effects, Math. Methods Appl. Sci., 41 (2018), 1554–1564. https://doi.org/10.1002/mma.4685 doi: 10.1002/mma.4685
![]() |
[30] |
X. Xu, Y. Liu, H. Li, F. E. Alsaadi, Robust set stabilization of Boolean control networks with impulsive effects, Nonlinear Anal. Modell. Control, 23 (2018), 553–567. https://doi.org/10.15388/NA.2018.4.6 doi: 10.15388/NA.2018.4.6
![]() |
[31] |
X. Xu, Y. Liu, H. Li, F. E. Alsaadi, Synchronization of switched Boolean networks with impulsive effects, Int. J. Biomath., 11 (2018), 1850080. https://doi.org/10.1142/S1793524518500808 doi: 10.1142/S1793524518500808
![]() |
[32] |
J. Yang, J. Lu, L. Li, Y. Liu, Z. Wang, F. E. Alsaadi, Event-triggered control for the synchronization of Boolean control networks, Nonlinear Dyn., 96 (2019), 1335–1344. https://doi.org/10.1007/s11071-019-04857-2 doi: 10.1007/s11071-019-04857-2
![]() |
[33] |
Q. Yang, H. Li, Y. Liu, Pinning control design for feedback stabilization of constrained Boolean control networks, Adv. Differ. Equations, 2016 (2016), 1–16. https://doi.org/10.1186/s13662-015-0739-5 doi: 10.1186/s13662-015-0739-5
![]() |
[34] |
Q. Yang, H. Li, P. Song, Y. Liu, Global convergence of serial Boolean networks based on algebraic representation, J. Differ. Equations Appl., 23 (2017), 633–647. https://doi.org/10.1080/10236198.2016.1270275 doi: 10.1080/10236198.2016.1270275
![]() |
[35] |
J. Zhong, Y. Liu, K. I. Kou, L. Sun, J. Cao, On the ensemble controllability of Boolean control networks using STP method, Appl. Math. Comput., 358 (2019), 51–62. https://doi.org/10.1016/j.amc.2019.03.059 doi: 10.1016/j.amc.2019.03.059
![]() |
[36] | S. Zhu, J. Lou, Y. Liu, Y. Li, Z. Wang, Event-triggered control for the stabilization of probabilistic Boolean control networks, Complexity, 2018 (2018). https://doi.org/10.1155/2018/9259348 |
[37] |
D. Cheng, F. He, H. Qi, T. Xu, Modeling, analysis and control of networked evolutionary games, IEEE Trans. Autom. Control, 60 (2015), 2402–2415. https://doi.org/10.1109/TAC.2015.2404471 doi: 10.1109/TAC.2015.2404471
![]() |
[38] |
H. Li, Y. Wang, Lyapunov-based stability and construction of Lyapunov functions for Boolean networks, SIAM J. Control Optim., 55 (2017), 3437–3457. https://doi.org/10.1137/16M1092581 doi: 10.1137/16M1092581
![]() |
[39] |
X. Kong, H. Li, X. Lu, Robust stability of switched delayed logical networks with all unstable modes, J. Franklin Inst., 359 (2021), 12–26. https://doi.org/10.1016/j.jfranklin.2020.12.036 doi: 10.1016/j.jfranklin.2020.12.036
![]() |
[40] |
X. Xu, Y. Liu, H. Li, F. E. Alsaadi, Robust set stabilization of Boolean control networks with impulsive effects, Nonlinear Anal. Modell. Control, 23 (2018), 553–567. https://doi.org/10.15388/NA.2018.4.6 doi: 10.15388/NA.2018.4.6
![]() |
[41] |
Y. Li, H. Li, W. Sun, Event-triggered control for robust set stabilization of logical control networks, Automatica, 95 (2018), 556–560. https://doi.org/10.1016/j.automatica.2018.06.030 doi: 10.1016/j.automatica.2018.06.030
![]() |
[42] |
X. Li, H. Li, Y. Li, X. Yang, Function perturbation impact on stability in distribution of probabilistic Boolean networks, Math. Comput. Simul., 177 (2020), 1–12. https://doi.org/10.1016/j.matcom.2020.04.008 doi: 10.1016/j.matcom.2020.04.008
![]() |
[43] |
Y. Guo, R. Zhou, Y. Wu, W. Gui, C. Yang, Stability and set stability in distribution of probabilistic Boolean networks, IEEE Trans. Autom. Control, 64 (2018), 736–742. https://doi.org/10.1109/TAC.2018.2833170 doi: 10.1109/TAC.2018.2833170
![]() |
[44] |
Z. Liu, Y. Wang, Disturbance decoupling of mix-valued logical networks via the semi-tensor product method, Automatica, 48 (2012), 1839–1844. https://doi.org/10.1016/j.automatica.2012.05.053 doi: 10.1016/j.automatica.2012.05.053
![]() |
[45] |
J. Zhong, D. W. C. Ho, J. Lu, A new approach to pinning control of Boolean networks, IEEE Trans. Control Network Syst., 2021 (2021), 415–426. https://doi.org/10.1109/TCNS.2021.3106453 doi: 10.1109/TCNS.2021.3106453
![]() |
[46] |
J. Zhong, Y. Liu, J. Lu, W. Gui, Pinning control for stabilization of Boolean networks under knock-out perturbation, IEEE Trans. Autom. Control, 67 (2021), 1550–1557. https://doi.org/10.1109/TAC.2021.3070307 doi: 10.1109/TAC.2021.3070307
![]() |
[47] |
X. Li, D. W. C. Ho, J. Cao, Finite-time stability and settling-time estimation of nonlinear impulsive systems, Automatica, 99 (2019), 361–368. https://doi.org/10.1016/j.automatica.2018.10.024 doi: 10.1016/j.automatica.2018.10.024
![]() |
[48] |
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
![]() |
[49] |
X. Li, D. Peng, J. Cao, Lyapunov stability for impulsive systems via event-triggered impulsive control, IEEE Trans. Autom. Control, 65 (2020), 4908–4913. https://doi.org/10.1109/TAC.2020.2964558 doi: 10.1109/TAC.2020.2964558
![]() |
[50] |
X. Li, X. Yang, S. Song, Lyapunov conditions for finite-time stability of time-varying time-delay systems, Automatica, 103 (2019), 135–140. https://doi.org/10.1016/j.automatica.2019.01.031 doi: 10.1016/j.automatica.2019.01.031
![]() |
[51] |
X. Li, X. Yang, J. Cao, Event-triggered impulsive control for nonlinear delay systems, Automatica, 117 (2020), 108981. https://doi.org/10.1016/j.automatica.2020.108981 doi: 10.1016/j.automatica.2020.108981
![]() |
1. | Pattrawut Chansangiam, Arnon Ploymukda, Riccati equation and metric geometric means of positive semidefinite matrices involving semi-tensor products, 2023, 8, 2473-6988, 23519, 10.3934/math.20231195 | |
2. | Tianxu Cui, Ying Shi, Wenge Li, Rijia Ding, Qing Wang, Said El Kafhali, IntFedSV: A Novel Participants’ Contribution Evaluation Mechanism for Federated Learning, 2025, 2025, 0884-8173, 10.1155/int/3466867 |
x_1 | x_2 | f_1 | f_2 |
1 | 1 | 0 | 1 |
1 | 0 | 1 | 1 |
0 | 1 | 0 | 0 |
0 | 0 | 0 | 0 |