水轮机调节系统的神经网络控制

Neural Network-Based Control of Hydraulic Turbine Regulating System

设计了一个三层ＢＰ神经网络，对有功功率反馈参与控制的水轮机调节系统，以其典型工况下的最优ＰＩＤ系数作为训练样本，对所设计的ＢＰ神经网络进行离线训练，进而构成一个基于ＢＰ神经网络的变参数ＰＩＤ控制器；利用ＢＰ神经网络的函数逼近能力来实现ＰＩＤ控制器在线调整，以达到优化控制的目的。对简单电力系统的仿真结果表明，这种控制器与常规ＰＩＤ控制器相比可以取得较好的控制效果，是实现水轮机调节系统自适应控制的一种可行的方法。

A PID control with variable parameters has been constructed based on a three-layer BP neural network for hydraulic turbine regulating system. The neural network takes the optimal PID parameters in typical operating modes as its training samples. Taking advantage of function-approximating ability of the BP neural network, on-line tuning of the PID controller has been realized. Simulation results have shown that the proposed scheme is feasible to realize the adaptive control of the hydraulic turbine regulating system.