改进梯度提升树算法的输电线路故障识别

Transmission line fault identification based on improved extreme gradient boosting algorithm

为精准识别输电线路的短路故障类型,提高输电线路短路故障诊断精度,提出一种贝叶斯优化梯度提升树的输电线路短路故障识别方法。通过变分模态分解和对称分量法,提取故障特征,构建特征集。采用贝叶斯优化梯度提升树挖掘特征集与短路故障类型之间的关系,建立短路故障识别模型,利用Simulink识别输电线路的故障精度。结果表明,该诊断模型能够快速且准确地识别短路故障类型,识别准确率高达99.75%。与传统方法相比,该方法显著减少了过渡电阻、故障距离和故障初始角对模型识别准确率的影响。

This paper aims to identify the type of short-circuit faults accurately in transmission lines and improve the diagnostic accuracy of short-circuit faults,and proposes a Bayesian optimized gradient boosting tree algorithm as a short-circuit fault identification method of transmission lines.The study involves extracting fault features and constracting the feature set with the variational modal decomposition and symmetric component method;using Bayesian optimization gradient boosting tree to study the relationship between the feature set and the short circuit fault type;developing a short circuit fault identification model;and simulating the transmission line to test the accuracy of fault identification by Simulink.The results show that this diagnostic model can identify the short-circuit fault types quickly and accurately with the accuracy by 99.75%.Compared with the traditional method,the proposed method significantly reduces the effects of transition resistance,fault distance and initial angle of fault on the model recognition accuracy.