非匹配不确定交叉严反馈超混沌系统神经网络反演同步

针对一类具有非匹配不确定性的交叉严反馈超混沌系统,提出一种基于多层前向神经网络的反演自适应同步设计方法.利用神经网络估计系统中的不确定性,运用滑模控制和交叉自适应反演控制处理系统中的非匹配不确定性及神经网络的逼近误差, 当虚拟控制项系数不过零时可保证系统的同步误差趋向于零,过零时可保证同步误差有界. 数值仿真证明了提出的控制方案的有效性.

Neural network-based backstepping design for the synchronization of cross-strict feedback hyperchaotic systems with unmatched uncertainties

For a class of cross-strict feedback hyperchaotic systems with unmatched uncertainties, a multilayer neural network (MNN) based adaptive backstepping design method is proposed. An MNN is introduced to estimate the uncertainties in systems. Sliding mode and adaptive backstepping control are used to deal with the unmatched uncertainties and the MNN approximation errors. If the virtual control coefficients do not pass through zeros, the proposed method guarantees that the synchronization errors of the systems approach zeros. If the virtual control coefficients pass through zeros, the proposed method guarantees that the synchronization errors of the systems are bounded. Numerical simulations are given to demonstrate the efficiency of the proposed control scheme.