基于差分进化小波神经网络的配电变压器故障检测  

Distribution Transformer Fault Diagnosis Based on Differential Evolution Wavelet Neural Networks

作　　者：王武[1] 杨晓博[1] 王红艳[1] Wang Wu;Yang Xiaobo;Wang Hongyan(School of Electrical and Mechanical Engineering,Xuchang University,Xuchang,Henan 461000,China)

机构地区：[1]许昌学院电气与机械工程学院,河南许昌461000

出　　处：《机电工程技术》2023年第7期155-158,共4页Mechanical & Electrical Engineering Technology

基　　金：河南省教育厅项目(18A510005,21B470009)。

摘　　要：配电变压器内部结构复杂,运行环境不确定,故障时有发生。将差分进化算法和小波神经网络结合,应用于配电变压器的故障检测。给出了小波神经网络的结构和原理,分析了差分进化算法的具体实现,在此基础上,给出了差分进化小波神经网络的原理和具体实施步骤。影响变压器运行的变量很多,对故障诊断中用到的状态变量进行了类型划分,找到动态变量、准动态变量和静态变量的状态数据和变量应用场景,进行了故障诊断前的故障状态数据挖掘。结合状态数据,进行了神经网络的结构设计和样本设计。学习算例表明,差分进化小波神经网络能够表达故障检测的映射关系,具有更高的故障诊断效率。Distribution transformer failure was inevitable for its complex internal structure and the uncertain operating environment.The differential evolution algorithm and wavelet neural network are combined to apply to fault detection of distribution transformers.The structure and principle of wavelet neural network are given,the specific implementation of differential evolution algorithm is analyzed,and on this basis,the principle and specific implementation steps of differential evolution wavelet neural network are given.There are many variables affecting the operation of the transformer,and the state variables used in fault diagnosis are divided into types,the state data and variable application scenarios of dynamic,quasi-dynamic and static variables are found,and the fault state data mining before fault diagnosis is carried out.Combined with the state data,the structural design and sample design of the neural network were carried out.The learning example shows that the differential evolutionary wavelet neural network can express the mapping relationship of fault detection,and has higher fault diagnosis efficiency.

关 键 词：配电变压器 故障诊断 小波神经网络 差分进化算法 

分 类 号：TM407[电气工程—电器]