基于单步预测的模型参考自适应逆控制

Model Reference Adaptive Inverse Control Based on Single-step Prediction

针对非线性系统,提出一种基于单步预测的模型参考自适应逆控制方案.该方案将预测思想引入自适应逆控制中,采用Elman网络作为模型辩识器,利用它建立非线性系统输出的预测模型;采用模糊神经网络作为自适应逆控制器,由预测误差进行参数的在线自适应寻优.仿真结果表明,该方案具有很好的跟踪效果和较强的鲁棒性.

A scheme of model reference adaptive inverse control based on single-step prediction is proposed for the nonlinear system. The thought of prediction is introduced to adaptive inverse control. The predictive model of the nonlinear system＇ s output is established by model identifier, which adopts Elman network. A fuzzy neural network is adopted for adaptive inverse controller, and the parameters of controller are trained by predictive error, so that they can have optimum searching on line. The simulation result indicates that the scheme has good traceability and robustness.