(2)若g(x,t)=I(x)+f(x)sin(ωt),且a=1,则可得到方程
1确定学习理论
考虑如下的非线性动态系统:
(3)
式中:x=[x1,x2,…,xn]T∈Rn,是可测的系统状态;p是系统的常值参数向量,不同的p产生不同的动态行为;F(x;p)=[f1(x;p),f2(x;p),…,fn(x;p)]T,是未知的光滑非线性向量场.假定系统状态x保持一致有界,即x(t)∈Θ⊂Rn,∀t≥t0,其中Θ是一个紧集,且以x0为起点的系统轨迹φζ(x0)是回归轨迹.
考虑如下的动态径向基函数(RBF)神经网络模型:
(4)
(5)
(6)
2Sine-Gordon方程的有限维近似
文中采用有限差分法的一种特殊形式——直线法[21]来实现SG方程的有限维逼近,针对SG方程,将空间偏导数用有限差分来代替.为叙述方便,考虑如下的一维有阻尼受迫SG方程:
(7)
(8)
其中,h是空间等分间隔.
注1由于j=0和j=N时,式(8)会产生两个虚拟点-1和N+1,因此当j=0和j=N时分别使用如下公式:
在此基础上,记U=[u0,u1,…,uN]T,系统(7)可离散为如下形式:
(9)
(10)
(11)
2.1解的存在性和唯一性分析
对任意给定的初始条件Z(0)=Z0,首先在距离空间C[0,T]中构造一个到其自身的映射Q:
其距离定义为
ρ(x,y)=‖x-y‖∞,x,y∈C[0,T].
为简化起见,下文中把下标∞省略.显然Z∈C[0,T],任取S∈C[0,T],有
(12)
对式(12)两边求微分易知,Z(t)为系统(11)的解.这就证明了系统(11)解的存在性和唯一性.
2.2解的收敛性
(13)
做适当变换可以得到如式(11)的形式:
由拉格朗日中值定理可知:
另外,
(14)
又由于
因此可以得到:
由Gronwall不等式可知:
这就证明了系统(11)的解收敛到系统(10)的解.
3SG方程近似系统的动态辨识
为方便分析,将系统(10)改写为
(15)
(16)
(17)
设计如下基于Lyapunov的权值学习律:
(18)
其中,Γ=ΓT>0,σ>0是一个很小的设计参数.
考虑由式(16)-(18)组成的自适应系统,由确定学习理论[18,22]可知,对任意始于U0=U(0)的周
4仿真结果
5结论
文中对满足一定条件下的有阻尼受迫SG方程的系统动态进行了辨识,由于SG方程是一个无穷维的分布参数系统,文中采用有限差分方法对其进行降维处理,将其近似为一组常微分方程组,并对近似系统解的存在唯一性和收敛性进行了研究,最后证明了其系统动态可利用确定学习理论实现准确辨识.仿真结果表明,具有回归轨迹的有阻尼受迫SG方程的系统动态确实得到了准确辨识.
图1 系统(2)在条件(I)下的仿真结果
图2 系统(2)在条件(II)下的仿真结果
参考文献:
[1]Josephson B D.Possible new effects in super conductive tunneling [J].Physics Letters,1962,1(7):251-253.
[2]Johnson S,Biswas A.Breather dynamics of the Sine-Gordon equation [J].Communications in Theoretical Phy-sics,2013,59(6):664-670.
[3]Gulevich D R,Gaifullin M B,Kusmartsev F V.Controlled dynamics of sine-Gordon breather in long Josephson junctions[J].The European Physical Journal B:Condensed Matter and Complex Systems,2012,85(1):1-5.
[4]Li Fei,Pan Chang-ning,Zhang Dong-xia,et al.Chaotic dynamics of a Josephson junction with nonlinear damping [J].Chinese Physics Letters,2010,27(5):050501-050503.
[5]Gutman S,Ha J H.Identification problem for damped Sine-
Gordon equation with point sources [J].Journal of Mathematical Analysis and Applications,2011,375(2):648-666.
[6]Ha J H,Nakagiri S.Identification of constant parameters in perturbed Sine-Gordon equations [J].Journal of the Korean Mathematical Society,2006,43(5):931-950.
[7]Roblesr-Pérez A M.Almost periodic solutions of forced Sine-Gordon equation [C]∥Proceedings of 2003 International Conference on Differential Equations.Singapore:World Scientific Publishing Co Pte Ltd,2005:258-263.
[8]Sun Y.New travelling wave solutions for Sine-Gordon equation [J].Journal of Applied Mathematics,2014,2014:841416.
[9]Yang Z,Zhong W P.Analytical solutions to Sine-Gordon equation with variable coefficient [J].Romanian Reports in Physics,2014,66(2):262-273.
[10]Jiang Z W,Wang R H.Numerical solution of one-dimensional Sine-Gordon equation using high accuracy multiquadric quasi-interpolation[J].Applied Mathematics and Computation,2012,218 (15):7711-7716.
[11]套格图桑.Sine-Gordon型方程的无穷序列新精确解[J].物理学报,2011,60(7):070203.
Tao Ge-tu-shang.New infinite sequences exact solutions to Sine-Gordon-type equations [J].Acta Physica Sinica,2011,60(7):070203.
[12]Demetriou M A,Rosen I G.On the persistence of excitation in the adaptive excitation of distributed parameter systems [J].IEEE Transactions on Automatic Control,1994,39(5):1117-1123.
[13]Demetriou M A,Rosen I G.Adaptive identification of se-cond-order distributed parameter systems [J].Inverse Problems,1994,10(2):261-294.
[14]Hong K S,Bentsman J.Direct adaptive control of parabolic systems:algorithm synthesis and convergence and stability analysis [J].IEEE Transactions on Automatic Control,1994,39(10):2018-2033.
[15]Narendra K S,Annaswamy A.Stable adaptive systems [M].Englewood Cliffs:Prentice-Hall,1989.
[16]Sastry S S,Bodson M.Adaptive control:stability convergence and robustness [M].Englewood Cliffs:Prentice-Hall,1989.
[17]Gevers M.A personal view of the development of system identification:a 30-year journey through an exciting field [J].Control Systems,2006,26(6):93-105.
[18]Wang C,Hill D J.Deterministic learning theory for identification,recognition,and control [M].Boca Raton:CRC Press,2009.
[19]朱澍,周盛凡.有阻尼受迫Sine-Gordon方程的全局周期吸引子[J].数学学报,1999,42(5):809-814.
Zhu Shu,Zhou Sheng-fan.Global periodic attractor for the damped and driven Sine-Gordon equation [J].Acta Mathematic Sinica,1999,42(5):809-814.
[20]Powell M J D.The theory of radial basis function appro-ximation in 1990 [J].Advances in Numerical Analysis,1992,2:105-210.
[21]Schiesser W E.The numerical method of lines [M].San Diego:Academic Press,1991.
[22]Wang C,Hill D J.Deterministic learning and nonlinear observer design [J].Asian Journal of Control,2010,12(6):714-724.
Identification of Damped and Driven Sine-Gordon
Equation Based on Deterministic Learning
DongXun-de1,2WangCong1
(1.School of Automation Science and Technology, South China University of Technology, Guangzhou 510640, Guangdong, China;
2.School of Mathematics,South China Unirersity of Technology,Guangzhou 510640,Guangdong,China)
Abstract:Discussed in this paper is the identification of dynamics of a class of damped and driven Sine-Gordon (SG) equation.Firstly,SG equation described by partial differential equation(PDE), which is infinite dimensional,is approximated by a set of ordinary differential equation with finite dimension by means of finite difference method. Then, the existence, uniqueness and convergence of the solution of the approximated system are proofed. Finally, the dynamics of the approximated system is identified on the basis of deterministic learning. Experimental results show that the proposed method helps achieve locally accurate identification of SG equation dynamics.
Key words: Sine-Gordon equation;system identification;deterministic learning;dynamics