基于单神经元自适应PID的矢量控制系统研究

Research of Single Neuron Adaptive PID Control for Vector Control System

转子磁场定向矢量控制中,转速PI调节器在控制对象参数变化时其鲁棒性能较差,而单神经元具有自学习、自适应能力,为了进一步改善矢量控制系统的性能,文章提出了用单神经元自适应PID控制器代替传统PI调节器。为了提高单神经元PID控制器的学习能力,将无监督的Hebb学习规则与有监督的Delta学习规则相结合,实现单神经元控制器的参数优化与在线自调。仿真结果表明,该系统不仅具有很好的静、动态性能,而且还具有很强的自适应性和鲁棒性。

In the rotor flux field oriented vector control system, the robust performance for the speed PI regulator is worse when the parameters are changed. The single neuron has the abilities of self-learning and adaptive. It is presented that single neuron adaptive PID controller substitutes the speed PI controller. In order to enhance the self-study ability of single neuron PID controller, the supervisory Delta study rule is combined with non-supervisory Hebb study rule, which realizes parameter optimization and self-tuning on line for single neuron controller. Results of simulation show that the system not only has good static, dynamic performance, but also has strong self-adaptability and robustness.