论文部分内容阅读
神经网络可用来建立非线性动态系统的模型，其辨识模型可分为串并联辨识模型和并联辨识模型两种．后者的思想源于基于参考模型自适应方案的输出误差辨识模型，对观测扰动有较强的抑制能力．本文对这种神经网络并联辨识结构的收敛性进行了研究，指出在网络参数满足一定条件时并联预测过程收敛，且并联辨识算法具有局部收敛性．仿真实验验证了上述结论．
The neural network can be used to establish the model of nonlinear dynamic system. The identification model can be divided into two models: serial-parallel identification model and parallel identification model. The latter idea originates from the output error identification model based on the reference model adaptive scheme, which has a strong ability to restrain the observed disturbance. In this paper, we study the convergence of this neural network parallel identification structure, and point out that the parallel prediction process converges when the network parameters satisfy certain conditions, and the parallel identification algorithm has local convergence. Simulation results verify the above conclusion.