两电机同步系统的神经网络控制

Neural network control of two-motor synchronous system

在对两台感应电机同步系统模型分析的基础上,依据同步系统的结构特点和控制要求,结合人工神经网络的非线性映射、自适应、自学习等能力,提出一种新的基于神经网络的两电机同步系统控制方案,其中神经网络控制器由基于RBF网络整定的自适应PID控制器和神经元解耦补偿器两部分组成.两个自适应PID控制器分别对速度控制回路和张力控制回路进行自适应控制,使系统具有更强的适应能力、更好的实时性和鲁棒性;神经元解耦补偿器综合两控制回路的耦合作用,通过训练网络权值,补偿各回路之间的耦合影响,实现速度和张力的解耦.试验结果表明:采用神经网络控制方法可以实现两电机同步系统中速度和张力的解耦控制,系统具有良好的动静态性能.

On the basis of model analysis of the two-motor synchronous system, according to the structural characteristic and control request of the system, a new control strategy based on neural networks is presented combined with its nonlinear mapping and adaptive and self-learning capabilities. The neural network controller is composed of adaptive PID controller, which uses the RBF network to modulate and the neuron decoupling compensator. The two adaptive PID controllers are used in the velocity loop and tension loop respectively, which make the system possess stronger adaptive capability and other better performances. The neuron decoupling compensator integrates the coupling effects of the two loops and realizes the decoupling control between velocity and tension by training the weights of networks to compensate the coupling effects. The experimental results show that the two-motor synchronous system is decoupled based on neural network control with better dynamic and static characteristics.