一类未知非线性受控对象快速神经网络辨识方法的研究

A fast neural network identification method researched on an unknown nonlinear control object

介绍一种用于辨识未知非线性系统的改进神经网络模型结构与算法。该方法具有辨识精度高、速度快、简单有效的特点，为设计非线性对象控制器提供了一条思路，从而摆脱了用线性模型近似被控对象的粗略做法。在此算法中，学习率采用随误差变化率而改变的做法，从而减小了学习率先取的盲目性，加速了网络训练过程。

A modified neural network model and algorithm, which can be used to identify an undnown nonlinear system is introduced. It provides a way for designing nonlinear object control device. Accuracy and fastness are its main features, so it can effectively avoid inaccurate identification and decrease the blindness in selecting the learning rate which vales according to the error difference and it also speeds up the learning procedure.