基于动作过程振动检测的低压断路器机械寿命预测  

Mechanical life prediction of low-voltage circuit breaker based on vibration detection during operation

作　　者：孙曙光 张伟 王景芹[1] 杜太行 高辉[3] Sun Shuguang;Zhang Wei;Wang Jingqin;Du Taihang;Gao Hui(State Key Laboratory of Reliability and Itelligence of Elecrical Equipment,Hebei Universily of Technology,Tianjin 300130,China;School of Arificial Intelligence,Hebei University of Technology,Tianjin 300130,China;Tianjin Benefo Elecrie Co.,lud.,Tianjin 300385,China)

机构地区：[1]河北工业大学省部共建电工装备可靠性与智能化国家重点实验室,天津300130 [2]河北工业大学人工智能与数据科学学院,天津300130 [3]天津市百利电气有限公司,天津300385

出　　处：《仪器仪表学报》2020年第12期146-157,共12页Chinese Journal of Scientific Instrument

基　　金：国家自然科学基金项目(51777057);河北省自然科学基金创新研究群体项目(E2020202142);河北省教育厅项目(ZD2016108)资助。

摘　　要：针对低压断路器机械寿命预测的问题,提出了一种基于振动检测的低压断路器剩余机械寿命在线预测方法。在对断路器分闸过程振动事件产生原因分析的基础上,提出以机构动作时间作为寿命评估模型的退化参量。首先,利用变分模态分解和奇异值分解联合降噪方法对振动信号进行降噪处理;然后,基于短时能量的双门限法实现关键振动事件时刻的检测,进而完成机构动作时间参数的提取;最后,基于Wiener理论构建断路器机械性能退化模型,并实现其剩余机械寿命预测。实验结果表明,所提方法能够有效地评估低压断路器的剩余机械寿命,预测平均相对误差仅为4.92%,具有一定的工程实用性。To solve the problem of on-line mechanical life prediction of low-voltage circuit breaker, propose an on-line remaining mechanical life prediction method for low-voltage circuit breaker based on vibration detection. Based on the analysis of the causes of vibration events during the opening process of the circuit breaker, the mechanism action time is utilized as the degradation parameter of life prediction model. Firstly, vibration signal is de-noised by the fusion of variation mode decomposition and singular value decomposition. Then, based on short-term energy, the dual-threshold method is used to detect moments for key vibration events and complete the extraction of mechanism action time parameter. Finally, the mechanical performance degradation model of the circuit breaker is formulated, which utilizes Wiener theory to realize prediction of the remaining mechanical life. Experimental results show that the proposed method can effectively evaluate the remaining mechanical life of low-voltage circuit breakers. The average relative error of prediction is only 4.92%, which has an engineering practicability.

关 键 词：振动信号 变分模态分解 奇异值分解 短时能量 Wiener理论 在线预测 

分 类 号：TM561[电气工程—电器] TH165.3[机械工程—机械制造及自动化]