基于改进Elman网络的最优励磁控制器  

Linear Optimal Excitation Controller Based on Modified Elman Neural Network

作　　者：陈前[1] 毛承雄[1] 陆继明[1] 余翔[1] 

机构地区：[1]华中科技大学电气与电子工程学院,湖北武汉430074

出　　处：《大电机技术》2007年第3期51-55,共5页Large Electric Machine and Hydraulic Turbine

摘　　要：本文在线性最优励磁控制的基础上,将线性最优控制理论与改进Elman神经网络有机结合,设计了一种新型的基于改进Elman网络(Modified Elman Neural Network)的最优励磁控制器。由于Elman网络具有特殊结构层,形成有“记忆”能力的神经网络的特点,并在原有结构上将高斯径向基函数引入Elman网络隐含层,因此可以更好地映射系统的非线性和动态特性。对单机无穷大系统进行仿真研究的结果表明,所设计的控制方式能精确地反映系统动态变化过程并提供良好的电压调节性能。This paper presents the design of excitation controUer based on linear optimal control theory and modified Elman neural network. A special correlation layer is appended to hidden layer of B15 network and its hidden layer is changed into radial basis functions to form an Elman neural network with memorial ability, with which the nonlinearity and the dynamic behavior of the system can be perfectly mapped. The digital simulation results in a single machine infinite-bus power system show that the designed excitation controller can reflect the dynamic changing process precisely and provide good voltage regulating performance for the generator.

关 键 词：最优控制 ELMAN网络 电力系统 智能控制 励磁控制 

分 类 号：TM571[电气工程—电器]