一种改进型BP算法在自适应控制系统中应用的研究

Study on the Application of An Improved BP Algorithm in Self-adaptive Control System

介绍BP算法在自适应控制中应用 ,即在离线学习过程中每一个输入样本对网络不立即产生作用 ,而是等到全部输入样本学习完成 ,将全部误差求和累加 ,再集中修改权值一次。对非线性控制系统在线控制过程中 ,采用一定算法可避免神经网络线性化 ,提高控制效果。

The application of BP algorithm in self adaptive control is i ntroduced, i.e. each sample in off line learning procedure does not affect the n etwork instantly, while after all the s amples have been learning completely the errors are accumulated and the weight v alue is corrected once. In on line contr ol of non linear control system the line arization of neural network can be avoid ed by using certain algorithm to obtain better control result.