基于IMDE和ORF模型的断路器工况识别  

Working Condition Identification of Circuit Breaker Based on IMDE and ORF Model

作　　者：车一鸣 王冬梅 王国兴 管华 CHE Yi-ming;WANG Dong-mei;WANG Guo-xing;GUAN Hua(Skill Training Center,State Grid Jibei Electric Power Company Limited,Baoding Hebei 071051,China;Zunhua Power Supply Company,State Grid Jibei Electric Power Company Limited,Zunhua Hebei 064200,China;不详)

机构地区：[1]国网冀北电力有限公司技能培训中心,河北保定071051 [2]国网冀北电力有限公司遵化市供电公司,河北遵化064200 [3]安徽同华新能源动力股份有限公司,安徽宣城242500

出　　处：《组合机床与自动化加工技术》2019年第12期85-90,共6页Modular Machine Tool & Automatic Manufacturing Technique

基　　金：河北省科技型中小企业技术创新基金:基于传感器网络的大型变压器状态监测与诊断(E2017080532)

摘　　要：针对高压断路器状态监测及早期故障诊断问题,提出了一种基于改进多尺度散布熵和优化随机森林模型的工况识别方法。首先,在传统多尺度散布熵算法基础上提出计算稳定性及精确度更为良好的改进多尺度散布熵算法,并利用其提取不同状态下断路器振动信号的多尺度特征,由此构造特征向量后,对遗传算法优化的随机森林模型进行训练,最后使用训练完成的最佳随机森林模型进行工况分类。实验测试结果表明,结合改进多尺度散布熵和优化随机森林模型可以实现断路器不同运行工况的准确识别,对实际工程应用具有一定参考借鉴价值。To solve the problem of condition monitoring and early fault diagnosis of high-voltage circuit breaker, a working condition identification method based on improved multiscale dispersion entropy(IMDE) and optimized random forest(ORF) model is presented. Firstly, on the basis of traditional multiscale dispersion entropy(MDE) algorithm, the improved multiscale dispersion entropy(IMDE) with better stability and accuracy is proposed, the multiscale features of different condition vibration signals of circuit breakers are obtained by IMDE algorithm, and the feature vectors are constructed to train the genetic algorithm optimized random forest model. Finally, the train-completed optimized random forest model is used to identify the different conditions. The test result shows that the proposed method combining IMDE and ORF model can accurately identify the different conditions of circuit breaker, and has a certain reference value for engineering application.

关 键 词：多尺度散布熵 随机森林模型 断路器 工况识别 

分 类 号：TH113.1[机械工程—机械设计及理论] TG506[金属学及工艺—金属切削加工及机床]