基于复合正交神经网络的灰色PID控制

结合传统反馈控制方法和灰色预测控制的预测控制器已在控制系统中获得了成功的应用。由于复合正交神经网络具有学习算法简单、收敛速度快，有逼近线性或非线性函数的优良特性。与灰色预测方法相比，神经网络预测精度高，且误差可控，如果把神经网络作为灰色预测器，建立一种灰色预测控制，那么就会在控制系统中获得良好的控制性能。为此，提出一种结合传统的PID控制和神经网络灰色预测补偿的灰色PID控制器，可对系统进行在线灰色估计和控制，由复合正交神经网络对不确定部分建立的灰色预测模型，可根据系统的参数变化来自动调节预测补偿值，使系统响应具有适应性。仿真结果表明，与传统的PID控制方法相比，该控制器可获得更为优良的动态性能和鲁棒性。

Grey PID Control Based on a Compound Orthogonal Neural Network

The grey prediction control of traditional feedback control combined with grey prediction control is successfully applied to the control systems. Compared with grey prediction, the neural network has high prediction accuracy and can control the prediction error because the compound orthogonal neural network has a simple algorithm, high-speed convergence of learning process, and excellent characteristics in the linear or nonlinear accurate approximation. If the neural network is treated as the grey predictor to form a grey prediction control, it will get good control performances in the control systems. Therefor, traditional PID control and grey prediction compensation of a neural network are integrated together to build a grey PID controller. It obtains on-line grey prediction and control. The grey prediction model built by a compound orthogonal neural network in the undetermined parts of system can automatically adjust the compensation value of grey prediction according to changing parameters, which makes the controller adaptive to the response of systems. Simulation results show that this controller can achieve better dynamic performance and robustness than that of traditional PID control.