无刷直流电动机的新型自适应模糊神经控制

 为无刷直流电动机提出了一种自适应模糊神经控制方法.这是一种建立在开关控制、模糊控制和自适应控制相结合基础上的控制方法,并用神经网络实现了模糊控制器和自适应机构.在无刷直流电动机的双闭环调速系统中,电流控制器是PI控制器;转速控制器是由1个开关控制器和1个包括自适应机构在内的模糊控制器相结合组成的,且用1个3层前向神经网络离线学习实现了模糊控制器,学习算法采用的是改进的BP算法.用1个单神经元通过在系统运行过程中的动态学习实现了自适应机构,学习算法选用了有监督的Hebb学习算法.由电机所处的运行阶段决定哪一个控制器工作.此控制算法的仿真结果说明,它使系统具有良好的动、静态特性和自适应性.

New Control Method of Adaptive Fuzzy-Neural Networks for Brushless D.C. Motor

This control method was based on combining fuzzy control with adaptive and bang-bang control. Fuzzy controller and mechanism are realized by neural networks. In double closed loop system the current regulator consists of proportional and integral controller, the speed regulator consists of bang-bang controller and fuzzy controller combining with adaptive mechanism. Fuzzy controller is realized by three layers of neural networks, which learns off the line. Improved BP algorism is used to learn. Adaptive mechanism is realized by an artificial neuron, which dynamically learns on the line during the operation of the system. The superintendence Hebb algorism is used to learn. Operation stage of the motor resolves which controller to work. Its shown by the results of digital simulation that the system has good dynamic and static performance, it also has excellent adaptive performance.