基于BP神经网络的多变量PID解耦控制

基于神经网络的智能PID控制策略，以经典的PID控制理论为基础，并通过具有多变量解耦控制自学习功能的神经网络参数整定来实现。本文给出了网络的结构和算法，示出了一组二元变量强耦合时变系统的实时仿真结果。通过计算机仿真证明，基于神经网络的PID控制具有良好的自学习和自适应解耦控制能力。该系统融解耦器和控制器于一体，易于实现，适用于非线性多变量系统的解耦控制。它使解耦后的系统具有较好的动态和静态性能，特别是当根据BP控制规律确定了网络连接权系数的初值时，还能使系统参数快速收敛。

Multivariable PID Decoupling Control Based On Back Propagation Neural Network

Intelligent PID control strategy, which is based on neural network, is according to classical PID control theory, and it is realized through neural network parameter setting, with a self study function on multivariate decoupling control .The structure and the algorithm of the network were given and the real-time simulation results of a time-varying system with a double variable and strong-coupling characteristics were shown. It proved that PID control based on neural network has preferable self study and adaptive decoupling control ability through computer simulation .In this system, the de-coupler inosculates with the controller, the system is easy to implement and applicable for decoupling control of the nonlinear multivariate system. It makes the decoupled system had better dynamic performance and static characteristic. Especially, it makes the system parameters astringed fast when determined the initial value of link-weighted coefficient of the network according to BP control law.