基于RBF NNI的单神经元PID模型参考自适应控制

The SN PID model reference adaptive control based on RBF NNI

RBF神经网络是一种性能良好的前向网络,它具有最佳逼近性能。本文中基于RBF神经网络,以单神经元PID作为控制器NNC,RBF网络作为辨识器,实现对被控对象的雅谷比信息辨识。传统的系统辨识方法包括以脉冲响应、最小二乘法为基础和最大似然法等,普遍存在难以克服的不足。本文的辨识方法及模型参考自适应控制方法具有计算速度快、推广和逼近、收敛特性良好等诸多优点,通过本文中的实例仿真,可以看出运用基于RBF神经网络辨识的单神经元PID模型参考自适应控制的系统输出可以达到对输入精确的跟踪效果。

The RBF neural network is a forward-propagation network with good characteristics and the best-approximation.In this paper,take RBF network as the identification,single neural PID as the neural network controller to accomplish the Jacobian identification of the controlling object.The traditional identify methods,including the impulse reaction,LS and maximum likelihood estimation,etc,all have shortcomings.The identify methods and MRAC used in this paper have the characteristic of fast calculating,good features of promoting,approximating and convergence.The output of the system can accurately trace the input under the SN PID Model Reference Adaptive Control based on RBF NNI.