一类MIMO状态可观测非线性系统的激励辨识

Identification Approach for MIMO System with Observable States Using Driving Sign

研究一类多输入多输出(MIMO)状态可观测非线性系统的激励辨识问题,输入激励信号采用高斯白噪声,均匀采样获得输出状态数据;根据Girsanov定理获得系统参数的渐近无偏估计;数值仿真试验说明了该方法的有效性并发现多变量耦合辨识中的噪噪比(NNR)现象;最后给出该系统的分步激励辨识算法。

An identification of a multi-input multi-output （MIMO） nonlinear continuous-time system with observable states using driving sign is proposed. The driving sign uses Gauss white noise, and the output is evenly sampled, and the maximum likelihood estimates of the model parameters are derived by using the Girsanov theorem. Numerical simulations illustrate the efficiency of the estimates and the noisenoise-ratio（NNR） phenomenon of coupling multi-variable identification appears. Finally, a step-type identification algorithm is given.