摘要
提出一种基于思维进化算法的模糊神经网络变压器故障诊断方法。该方法利用思维进化算法中的趋同和异化操作,对模糊神经网络中输入变量的隶属度函数位置参数和宽度参数以及神经网络的连接权值进行全局优化,可有效地克服常规模糊神经网络BP算法收敛速度慢、精度不高和遗传算法训练模糊神经网络速度缓慢、易陷入局部极小等缺点,有利于更快地收敛于全局最优解。并将其应用到基于溶解气体分析的变压器故障诊断中,实例表明,采用该方法具有较快的收敛速度和较高的诊断准确度,说明了该方法的正确性和有效性。
A new transformer fault diagnostic method using fuzzy neural network based on mind evolutionary algorithm is proposed.According to the similartaxis and dissimilation of mind evolutionary algorithm,the method optimizes the membership function parameter of input variable and connection weight in fuzzy neural network,and benefits to find the global optimal solution quickly.It can avoid the defects of conventional BP algorithm which has slow convergence and low precision.At the same time,it can remove the defects of GA which has slow training speed and local minimum.Simulation results of transformer fault diagnosis based on the dissolved gas analysis show that this method improved convergence speed and diagnosis accuracy to some extent and it is effective.
出处
《科学技术与工程》
2011年第13期2957-2961,共5页
Science Technology and Engineering
关键词
变压器
故障诊断
模糊神经网络
思维进化算法
transformer fault diagnosis fuzzy neural network mind evolutionary algorithm
