基于BP神经网络的最优励磁控制器

设计了一种新型的基于BP神经网络的最优励磁控制器(NNOEC)。在线性最优励磁控制的基础上，利用4层BP神经网络对发电机的运行方式和系统所遭受的干扰类型进行辨识，通过对网络的训练，使得网络能够实时根据发电机的状态量来调节最优控制的反馈矩阵，以适应当前的运行点和所遭受的干扰。仿真结果表明，所设计的NNOEC在系统运行方式较大的变化范围内都能提供很好的控制性能，在大小扰动下均表现出很好的阻尼特性和良好的电压性能。

OPTIMAL EXCITATION CONTROLLER BASED ON BP NEURAL NETWORK

A new type of optimal excitation controller based on BP neural network (NNOEC) is presented. In the proposedNNOEC, a four--layer BP neural network is used to identify the operation conditions of the generator and the disturbances inthe system. The trained neural network can dynamically adjust the optimal feedback matrix according to the state variables ofthe generator. Simulation results show that the designed NNOEC can provide very good damping over a wide range ofoperating conditions and good voltage performance of the generator can also be guaranteed at the same time.This is a subproject supported by National Science and Technology Key Project: Development of the Three GorgesHydroelectric Generator Excitation Systems and Equipment--Research of the Control Mode in Excitation System (97-312-01if~lb).