基于Bayes-CNN的变压器故障分类方法研究

Research on transformer fault classification method based on Bayes-CNN

电力变压器发生故障的表现和根本原因具有一定程度的模糊性和随机性,在复杂情境下,传统方法往往难以准确识别变压器故障,其精度存在一定提升空间。因此,本文提出一种新的变压器故障识别方法。该方法采用结合贝叶斯理论和卷积神经网络(CNN)的算法,具体为利用卷积神经网络处理特征气体数据,并采用贝叶斯算法对模型参数进行寻优,旨在提高故障检测的准确率。研究通过对故障类型进行编码和预处理,构建变压器故障分类模型,应用Bayes-CNN模型对变压器故障进行分类,结合实例验证,并将其与SVM、DBN及CNN模型进行对比实验。结果表明,采用贝叶斯优化的CNN算法显著提升了模型的收敛速度和拟合精度,证明该变压器故障分类方法具有较优性能,为电力变压器故障诊断领域提供了新的方法与思路。

The performance and root cause of power transformer failure have a certain degree of ambiguity and randomness,in complex situations,the traditional method is often difficult to accurately identify transformer faults,and there is certain amount of room for improvement in its accuracy.Therefore,a new method of transformer fault identification is proposed in this paper.In this method,a combination of Bayesian theory and convolutional neural network(CNN)algorithm is used to process characteristic gas data by convolutional neural network,and Bayesian algorithm is used to optimize model parameters,aiming at improving the accuracy of fault detection.By coding and preprocessing the fault types,the transformer fault classification model is constructed,and the Bayes-CNN model is applied to classify the transformer faults,which is verified by examples and compared with SVM,DBN and CNN models.The results show that the convergence speed and fitting accuracy of the model are significantly improved by using the Bayesian optimization CNN algorithm,which proves that the transformer fault classification method has better performance and provides a new method and idea for power transformer fault diagnosis.