基于复合神经网络的感应电机直接转矩控制

提出了一种基于复合神经网络的直接转矩控制器算法,分别对该神经网络控制器各组成部分的结构进行了设计.网络采用固定值和离线训练2种策略获取网络参数,训练样本数少,训练时间短.采用复合网络结构使得直接转矩控制器的数字实现变得容易,充分利用网络的并行处理特性,使得计算时间缩短,响应速度变快.在Matlab/Simulink中结合神经网络工具箱建立了其仿真模型,仿真结果证明了设计的控制器具有可行性、有效性.

Multiple-neural-network-based direct torque control of induction motor

A novel direct torque controller based on multiple-neural-network(MNN) is proposed and the structures of the proposed controller were designed.All parameters of networks were obtained by adopting fixed values or by offline training.The number of training samples is less,and training time is short.The digital implement of direct torque controller based on MNN is easier than the conventional method.The calculation time is shortened and the response speed is high because of the fast parallel computation capability of neural network.A computer simulation model of direct torque control of induction motor is developed in Matlab/Simulink together with the neural network toolbox.The simulation results demonstrate the feasibility and efficiency of the control scheme.