基于改进GWO-PNN算法的电力变压器故障诊断  

Fault Diagnosis of Power Transformer Based on Probabilistic Neural Network Optimized by Improved Grey Wolf Optimizer

作　　者：徐勇 路小娟[1] 张学玉 XU Yong;LU Xiao-juan;ZHANG Xue-yu(School of Automation and Electrical Engineering,Lanzhou Jiaotong University,Lanzhou 730070,China)

机构地区：[1]兰州交通大学自动化与电气工程学院,兰州730070

出　　处：《兰州交通大学学报》2023年第1期54-61,共8页Journal of Lanzhou Jiaotong University

基　　金：甘肃省科技重大专项(20ZD7GF011);甘肃省高校产业支撑计划项目(2022CYZC-34)。

摘　　要：针对电力变压器以油中气体含量作为故障输入特征量,诊断结果准确性不理想、模型不稳定的问题,提出了一种基于改进灰狼算法优化概率神经网络的混合智能故障诊断方法.首先,对灰狼算法的控制因子和加权距离进行修改,以提高算法收敛精度及稳定性;其次,通过6个常用的测试函数测试改进后灰狼算法的性能,并将改进后的灰狼算法与其它智能算法进行对比;最后,将概率神经网络和改进的灰狼算法相结合并用于故障诊断.仿真结果表明:该模型在变压器故障诊断方面具有一定的有效性.Aiming at the problem that the gas content in the oil is used as the fault input characteristic of the power transformer, the accuracy of the diagnosis results is not ideal and the model is unstable.A hybrid intelligent fault diagnosis method is proposed based on probabilistic neural network optimized by improved grey wolf optimizer.Firstly, the control factor and weighted distance of the gray wolf optimizer are modified to improve the convergence accuracy and stability of the algorithm.Secondly, the performance of the improved gray wolf optimizer is tested through six commonly used test functions, and the improved gray wolf optimizer is compared with other intelligent algorithms.Finally, the probabilistic neural network and improved gray wolf optimizer are combined to carry out fault diagnosis.The simulation results show that the model has certain validity in transformer fault diagnosis.

关 键 词：变压器 故障诊断 控制因子 加权距离 灰狼算法 概率神经网络 

分 类 号：TU225.4[建筑科学—建筑设计及理论]