基于神经网络集成模型在高压断路器机械故障诊断中的应用  

Application of Neural Network Ensemble Model in Mechanical Fault Identification of High Voltage Circuit Breaker

作　　者：赵科 杨景刚 马速良[2] 王昱皓 武建文[2] 梁传涛 ZHAO Ke;YANG Jinggang;MA Suliang;WANG Yuhao;WU Jianwen;LIANG Chuantao(Power Research Institute, State Grid Jiangsu Electric Power Company, Nanjing 211103, China;School of Automation Science and Electrical Engineering, Beihang University, Beijing 100191, China;Shandong Taikai High Voltage Limited Company, Shandong Tai＇ an 271000, China)

机构地区：[1]国网江苏省电力公司电力科学研究院,南京211103 [2]北京航空航天大学自动化科学与电气工程学院,北京100191 [3]山东泰开高压开关有限公司,山东泰安271000

出　　处：《高压电器》2018年第7期217-223,共7页High Voltage Apparatus

摘　　要：高压断路器健康情况严重影响着电网的安全、稳定运行。文中提出一种基于高压断路器合闸过程振动信号时频特征下的集成学习神经网络模型,满足高压断路器故障情况高精度诊断的要求。首先,分析高压断路器在多测量位置下合闸振动信号特征,并在时、频两域定义合闸过程多测量位置振动信号的广义能量和小波能量比进行特征空间描述;然后,设计基于集成学习思想的神经网络算法划分特征空间,诊断故障类别;最后,通过实验数据分析和多种诊断方法的对比验证文中所述的诊断过程合理、诊断结果精确,有利于高压断路器故障排查。The health of high voltage circuit breaker seriously affects the safe and stable operation of power grid. In this paper, a neural network ensemble model based on time-frequency characteristics of vibration information of highvoltage circuit breaker is proposed in view of the requirement of high-precision circuit breaker fault diagnosis. Firstly, the characteristics of the vibration signal of the high-voltage circuit breaker under the multi-measurement position are analyzed. The generalized energy of the vibration signal and the wavelet energy ratio are described in the feature space. Then, neural network ensemble model is used to divide the feature space, and the fault category is diagnosed. Finally, by comparing the experimental data with various diagnostic methods, it is verified that the diagnosis process described in this article is reasonable and the diagnosis result is accurate, and it is beneficial to the troubleshooting of high voltage circuit breakers.

关 键 词：高压断路器 故障诊断 集成学习 人工神经网络 机械振动 

分 类 号：TM561[电气工程—电器]