基于改进PSO-BP神经网络的变压器故障诊断方法
Modified PSO-BP Neural Network-based Transformer Fault Diagnosis
战泓廷 1张王瑞 1蒋景熙 1张宏阳 1杨帅 1赵臻 1闻新1
作者信息
- 1. 南京航空航天大学航天学院,江苏 南京 211106
- 折叠
摘要
针对电力变压器以油中气体含量作为判别故障特征,诊断结果可信度不理想的问题,提出一种基于改进PSO-BP算法的变压器神经网络故障诊断方法.对PSO-BP算法的粒子初始搜索范围进行修改,以提高算法收敛精度及稳定性.将变压器气体溶解分析(DGA)数据作为输入,通过改进的PSO-BP算法训练神经网络,并将诊断结果与传统PSO-BP故障诊断神经网络结果进行对比分析,结果表明改进的PSO-BP算法提高了变压器故障诊断的精度.
Abstract
Aiming at the problem that the reliability of the diagnosis results of power transformer is not ideal when the gas content in oil is used as the fault characteristic,a transformer neural network fault diagnosis method based on modified PSO-BP algorithm is proposed.The initial particle search range of PSO-BP algorithm is modified to improve the conver-gence accuracy and stability of the algorithm.The transformer gas dissolution analysis(DGA)data is taken as input,and the improved PSO-BP algorithm is used to train the neural network.The diagnosis results are compared with those of the conventional PSO-BP fault diagnosis neural network,whose results show that the modified PSO-BP algorithm achieves improved accuracy of transformer fault diagnosis.
关键词
变压器/油中溶解气体分析/故障诊断/神经网络/粒子群算法Key words
transformer/dissolved gas analysis/fault diagnosis/neural network/particle swarm optimization引用本文复制引用
出版年
2024
NETL
NSTL
万方数据