Transformer Fault Diagnosis Based on Improved PSO-RBF Neural Network
刘锐①LIU Rui;胡宗义①HU Zong-yi;殷嘉伟②YIN Jia-wei;杨彪①YANG Biao
①国电南瑞南京控制系统有限公司,南京211111;
②南瑞集团有限公司(国网电力科学研究院有限公司),南京210000)
①NARI-TECH Nanjing Control Systems Limited ,Nanjing 211111,China;
②NARI Group Corporation ( State Grid Electric Power Research Institute ) ,Nanijing 210000 ,China
摘要:针对电力变压器故障多发且类型多样等问题,文章通过改进传统粒子群算法,来训练径向基函数神经网络,构建了IPSORBF神经网络模型,又依据电力变压器油内气体含量关系,对其故障类型进行分类,并通过文章中的网络模型进行仿真测试。仿真结果表明,IPSO-RBE神经网络对各类故障类型的诊断正确率高达94.6%。相较于PSO-RBE神经网络模型,输出误差、运行时间和检测正确率均得到提升,有较好的效果,能满足实际应用中对电力变压器的故障检测需求。
Abstract: In order to solve the problem of power transformer faults occurring frequently and of various types, this paper improves thetradlitional particle swam optimirzation algorithm to train the Radial basis function neural network, and constructs the IPSO-RBF neuralnetwork modiel. Based on the relationship between the gas content in the oil of power transfomers， their fault types are classified andsimulated using the network model in the article. Simulation shows that the IPSO-RBF neural network has a diagnostic accuracy of up t94.6% for various types of fauls. Compared with the PSO-RBF neural network model, the output eror，nunning time,and delectionaccuracy have all been improvedl, with good results, which can meet the fault detection needs of power transformers in practical applications.
关键词:电力变压器;神经网络;粒子群算法;仿真测试;故障检测
Key words: power transformer; neural network; particle swarm optimization algorithm; simulation testing; fault detection
价值工程-文章出处：刘锐;胡宗义;殷嘉伟;杨彪.基于改进PSO-RBF神经网络的变压器故障诊断[J].价值工程,2023,42(21):149-151. |
