一种新型改进单神经元控制器

A novel improved single neuron controller

本文通过理论分析重点指出了目前单神经元控制器普遍存在的权值恶化问题,分析了产生原因,并在此基础上给出了改进的神经元算法,这种改进算法较好地抑制了权值参数恶化,可以使神经元权值按照系统的实际状态进行调整。与其它单神经元控制器比较表明,该单神经元控制具有良好的稳定性及较强的鲁棒性。

By the theoretical researching,the limitations of traditional single neuron,especially on the weight value's deterioration,are pointed out in this article and the causations are analyzed.Based on these analyses,an improved single neuron algorithm is designed.This improved single neuron algorithm obviates the weight value's deterioration,and the weight's value can be adjusted according to the actual state of system.Comparing with the others single neuron controller,it has the advantages of stableness and robustness.