目前，高压断路器的机械特性测量及缺陷诊断大多是停电后利用传感器获取其机械特性参数从而进行缺陷诊断。本文以弹簧操动机构高压断路器为对象展开了研究，提出了一种无需停电的高压断路器机械特性非接触式测量及其缺陷诊断方法，主要研究内容和研究成果如下：

首先，本文提出了一种高压断路器分闸电磁铁行程的非接触式测量方法，基于主动视觉标定的原理设计了安全电气距离下的测量系统，利用K-Means图像聚类算法结合CLSD直线检测算法计算出了电磁铁行程像素值。同时结合激光传感器精准定位的功能，利用游标卡尺进行标定得到固定距离下的像素当量，从而计算出电磁铁行程实际物理值。通过计算结果与塞尺测量结果的对比，验证了该方法的有效性。

其次，搭建了一种包括电磁铁行程缺陷在内的高压断路器机械特性模拟实验平台，考虑到动触头密封于灭弧室内，且当光照不足时易发生目标物局部缺失，本文将深度学习方法应用于与动触头刚性连接的标志物行程曲线测量上，先利用YOLOv5框架对图像中的标志物进行了检测，其次基于DeepSort算法对其运动轨迹进行跟踪得到了标志物行程曲线，根据标志物与动触头的刚性连接关系最终得到动触头行程曲线。利用小波变换在突变点检测方面的优点提取出分闸电流波形，经低通滤波后与行程曲线相结合，得出了高压断路器常见缺陷状态的曲线变化规律。通过特征定义法和导数定义法分别计算出每种状态的行程与电流曲线特征值，为高压断路器缺陷诊断提供了样本数据集。

最后，提出了一种基于随机森林特征优选的灰狼优化支持向量机高压断路器缺陷诊断方法，先利用随机森林算法结合支持向量机模型确定最优特征子集数据库，再基于灰狼算法对支持向量机进行参数寻优来实现高压断路器缺陷诊断。计算结果表明，经随机森林算法特征优选后，可以一定程度的提升模型诊断准确率，降低诊断时间，其次结合灰狼优化算法对支持向量机参数寻优可以进一步提升诊断准确率。

At present, the mechanical characteristic measurement and defect diagnosis of high voltage circuit breakers are mostly realized by using sensors to obtain mechanical characteristic parameters after power outage. In this paper, a non-contact measurement and defect diagnosis method for mechanical characteristics of high voltage circuit breakers with spring actuator is proposed. The main research contents and results are as follows:

Firstly, this paper proposes a non-contact measurement method for the solenoid stroke of high voltage circuit breaker. Based on the principle of active visual calibration, the measurement system under safe electrical distance is designed. The pixel value of the solenoid stroke is calculated by using K-Means image clustering algorithm and CLSD linear detection algorithm. Combined with the precise positioning function of laser sensor, the vernier caliper is used to calibrate the pixel equivalent at a fixed distance, and the actual physical value of the electromagnet stroke is calculated. The calculation results are compared with the field measurement results, and the relative error is small, which verifies the effectiveness of the method.

Secondly, a simulation experiment platform for mechanical characteristics of high voltage circuit breakers, including the stroke defect of electromagnet, is built. Considering that the dynamic contact is sealed in the arc extinguishing chamber and the target is prone to local missing when the light is insufficient, this paper applies the deep learning method to the measurement of the marker stroke curve rigidly connected with the dynamic contact. Firstly, the YOLOv5 framework is used to detect the markers in the image. Secondly, the DeepSort algorithm is used to track its trajectory to obtain the marker stroke curve. According to the rigid connection relationship between the marker and the dynamic contact, the dynamic contact stroke curve is finally obtained. At the same time, the breaking current waveform is extracted by using the advantages of wavelet transform in the detection of mutation points. After low-pass filtering, it is combined with the stroke curve to obtain the curve variation law of each defect state of high voltage circuit breaker, and its influencing factors are analyzed. The characteristic values of stroke and current curves of each state are calculated by the feature definition method and the derivative definition method, respectively, which provides a sample data set for the defect diagnosis of high voltage circuit breakers.

Finally, a grey wolf optimization support vector machine defect diagnosis method for high voltage circuit breaker based on random forest feature optimization is proposed. Firstly, the random forest algorithm is combined with the support vector machine model to determine the optimal feature subset database, and then the parameters of support vector machine are optimized based on the grey wolf algorithm to realize the defect diagnosis of high voltage circuit breaker. The calculation results show that after the feature optimization of the random forest algorithm, the diagnostic accuracy of the model can be improved to a certain extent, and the diagnostic time can be reduced. Secondly, the parameter optimization of the support vector machine combined with the grey wolf optimization algorithm can further improve the diagnostic accuracy.

[1]Tang, G F, He Z Y, Pang H. Research application and development of VSC-HVDC engineering technology [J]. Automation of Electric Power Systems, 2013, 37(15):3-14.

[2]Li, Y B, Bu, G Q, Wang, S S, et al. Analysis of DC overvoltage caused by DC short-circuit fault in Zhangbei VSC-based DC grid [J]. Proceedings of the CSEE, 2017, 37(12):3391-3399.

[3]鲁敦科,张新雨,兰云生,等.基于径向基函数分类器的断路器机械故障诊断[J].高压电器,2021,57(04):142-148+155.

[4]石万里,夏武.一起500kV断路器拒动故障原因分析及改进措施[J].电工电气,2017(02):59-62.

[5]Mazza G. The first international enquiry on circuit-breaker failures and defects in service[J]. Electra, 1981, 79.

[6]Huang N, Chen H, Cai G, et al. Mechanical fault diagnosis of high voltage circuit breakers based on variational mode decomposition and multi-layer classifier[J]. Sensors, 2016, 16(11): 1887.

[7]B Feizifar, Z Muller , O Usta. A New Algorithm for Detecting Failure to Clear Mode of Circuit Breakers Using Fundamental Frequency Component of Voltage Signals[J].IEEE Transactions on Power Delivery, 2020,35(02):794-801.

[8]刘银.基于Mate30高速摄影的断路器分合闸速度测试方法研究[J].机电信息,2022(04):72-74.

[9]赵靖波.高压断路器型式试验机械行程特性曲线的监测[J].电气时代,2015(01):93-95.

[10]张生成,陈志新.真空断路器的速度特性测量方法[J].甘肃科技,2009,25(12):10-11.

[11]王立军,张拓,刘光伟,等.基于机器视觉技术的高压断路器机械特性诊断[J].高电压技术,2020,46(06):2148-2154.

[12]董华军,马丽婷,李杰,等.基于图像处理技术的真空开关分闸速度检测[J].高电压技术,2020,46(07):2539-2544.

[13]Cui G, Zhao S, Zhao C, et al. Novel HV Circuit Breaker Reliability Testing Measure Based on Image Processing[C]//2008 Joint International Conference on Power System Technology and IEEE Power India Conference. IEEE, 2008: 1-4.

[14]吴玲玲.高压断路器故障诊断及维修决策研究[D].天津大学,2019.

[15]董镝,李恒真,易林,等.基于形态小波和支持向量机的高压断路器状态监测和故障诊断[J].电测与仪表,2017,54(04):92-99.

[16]A I Khalyasmaa, M D Senyuk, S A Eroshenko. High-Voltage Circuit Breakers Technical State Patterns Recognition Based on Machine Learning Methods[J]. IEEE Transactions on Power Delivery, 2019, 34(04):1747-1756.

[17]杨成刚.高压断路器分合闸速度现场测试方法的研究与应用[D].华北电力大学,2018.

[18]赵晨光.基于图像测量技术的高压开关机械特性测试方法研究[D].华北电力大学（河北）,2009.

[19]白云飞,王代远,朱庆洋,等.基于合闸行程曲线的12 kV真空断路器合闸速度在线监测研究[J].高压电器,2021,57(10):59-65+76.

[20]M Ishii et al. Multistory transmission tower model for lightning surge analysis[J].IEEE Transactions on Power Delivery, 1991,6(03):1327-1335.

[21]易慧,尹项根,郑浩,等.基于DSP的高压断路器综合在线监测装置[J].高压电器,2007, 43(01):35-39.

[22]武建文,白洪超,张丹丹,等.适用于新标准的断路器机械特性测试装置[J].高压电器,2006, 42(04):298-300.

[23]Huang H, Wang F, Lu Y, et al. A new test method of circuit breaker spring telescopic characteristics based image processing[C]//E3S Web of Conferences. EDP Sciences, 2018, 38: 04022.

[24]刘建寅,刘亚南,王常龄,等.高压断路器操动机构典型故障模拟及其特征参量变化分析[J].高压电器,2021,57(02):7-13.

[25]李斌,郝建成,刘一涛,等.断路器机械特性特征数据稳定性和故障机理[J].中国电力,2015,48(02):59-65.

[26]栾乐,刘俊翔,王红斌,等.基于线圈电流曲线与行程曲线的断路器操作机构故障诊断的对比研究[J].四川电力技术,2019,42(06):9-13+35.

[27]张佩，赵书涛，申路，等.基于改进EEMD的高压断路器振声联合故障诊断方法[J].电力系统保护与控制,2014,42(08):77-81.

[28]夏小飞,芦宇峰,苏毅,等.基于相空间重构与改进GSA-SVM的高压断路器机械故障诊断[J].中国电力,2021,54(10):169-176.

[29]代洁,马骊群,高廷,等.线激光位移传感器像素当量标定方法[J].计测技术,2019,39(06):31-35.

[30]赵美丹,余桂英,瞿永顺,等.微小型光学元件尺寸的视觉测量系统研制[J].激光与光电子学进展,2021,58(08):372-380.

[31]贾国邦,宋丽丽,曹博文,等.基于机器视觉的微小孔零件尺寸检测研究[J].工具技术,2021,55(07):105-109.

[32]王运哲,郑凯.机器视觉中相机标定方法的研究[J].机械管理开发,2019,34(05):81-82.

[33]Li J B, Huang W Q, Zhao C J. Machine vision technology for detecting the external defects of fruits—A review[J]. The Imaging Science Journal, 2015, 63(05): 241-251.

[34]谢俊,李玉萍,左飞飞,等.基于机器视觉的孔类零件尺寸在线检测[J].电子测量技术,2021,44(02):93-98.

[35]姜苗苗,史国友,许拴梅,等.基于自适应K均值聚类和霍夫变换的船舶干舷视觉检测[J].上海海事大学学报,2021,42(02):34-39.

[36]曹义亲,何恬,刘龙标.基于改进LSD直线检测算法的钢轨表面边界提取[J].华东交通大学学报,2021,38(03):95-101.

[37]黎淏钧.复杂背景下的无人机配电线检测[D].广东工业大学,2020.

[38]左川.基于图像识别的输电线路绝缘子检测方法研究[D].华北电力大学,2019.

[39]郭文诚,崔昊杨,马宏伟,等.基于Zernike矩特征的电力设备红外图像目标识别[J].激光与红外,2019,49(04):503-506.

[40]胡漫,曹继华,李士心,等.改进YOLOv3和DeepSort的多目标跟踪算法[J].天津职业技术师范大学学报,2022,32(01):52-57.

[41]金立生,华强,郭柏苍,等.基于优化DeepSort的前方车辆多目标跟踪[J].浙江大学学报(工学版),2021,55(06):1056-1064.

[42]黄兴泉,王伟,张欲晓,等.断路器机械特性的高速摄像法检测[J].华北电力大学学报(自然科学版),2011,38(05):43-47.

[43]王森.高压断路器故障诊断算法的研究与应用[D].浙江大学,2018.

[44]宋亚凯,张一茗,张文涛,等.断路器分合闸线圈电流波形特征值提取算法研究[J].高压电器,2020,56(01):181-187.

[45]Dou L, Wan S, Zhan C. Application of Multiscale Entropy in Mechanical Fault Diagnosis of High Voltage Circuit Breaker[J].Entropy (Basel). 2018,20(5):325.

[46]张帅,彭在兴,李锐海,等.断路器分合闸线圈电流波形的差异机制研究[J].高压电器,2020,56(06):165-172.

[47]Philip J.Moore.Radiometric measurement of circuit breaker interpole switching times[J].IEEE Transscation on power delivery,2004,19(03).987-992.

[48]Toshie T, Kenichi K, Mitsuru T. Electromagnetic analysis coupled with motion for high-speed circuit breakers of eddy current repulsion using the tableau approach[J]. Electrical Engineering in Japan，2005,152(04):8-16.

[49]Landry M, Leonard F, Landry C, et al. An improved vibration analysis algorithm as a diagnostic tool for detecting mechanical anomalies on power circuit breakers[J]. IEEE Transactions on Power Delivery, 2008, 23(04):1986-1994.

[50]Runde M，Ottesen G E, Skyberg B, et al. Vibration analysis for diagnostic testing of circuit-breakers[J].IEEE Transaction on Power Delivery, 1996,11(4):1816-1823.

[51]Okutsu N, Takahashi Y, Matsuda S, et al. Pattern Recognition of Vibrations in Metal Enclosures of Gasnsulated Equipment and its Application[J].IEEE Transactions on Power Apparatus and Systems,2007,100(06):2733-2739.

[52]陈伟根.以油中多种气体为特征量的变压器绝缘在线监测及故障诊断技术研究[D].重庆大学,2003.

[53]孙一航,武建文,廉世军,等.结合经验模态分解能量总量法的断路器振动信号特征向量提取[J].电工技术学报,2014,29(03):228-236.

[54]李斌,李爽,鲁旭臣.高压断路器机械特性振动信号特征提取和故障诊断方法研究[J].高压电器,2015,51(10):138-144.

[55]杜小康,陈志英.高压断路器合(分)闸线圈电流信号特征值提取方法研究[J].电气应用,2018,37(15):26-30.

[56]孙银山,张文涛,张一茗,等.高压断路器分合闸线圈电流信号特征提取与故障判别方法研究[J].高压电器,2015,51(09):134-139.

[57]李聪.基于分合闸线圈电流特征的断路器机械故障诊断研究[D].华北电力大学,2019.

[58]逯浩坦,伊力哈木·亚尔买买提,刘鹏伟,等.基于GWO-SVM的高压断路器故障诊断[J].组合机床与自动化加工技术,2022(01):103-107.

[59]杨秋玉,彭彦卿,庄志坚.基于ADAMS的高压断路器弹簧机构动力学仿真与故障分析[J].高压电器,2018,54(06):67-73+80.

[60]刑锋,钟声,梁胜乐,等.基于行程曲线的高压断路器典型故障特征参数提取[J].华北电力大学学报(自然科学版),2021,48(04):56-62+90.

[61]李德阁,武建文,马速良,冯英,杨景刚,赵科.基于行程信息的断路器弹簧故障程度诊断[J].高压电器,2018,54(04):20-27.

[62]Dheeb Al Bashish, Malik Braik, Sulieman Bani-Ahmad. Detection and Classification of Leaf Diseases using K-means-based Segmentation and Neural-networks-based Classification[J]. Information Technology Journal, 2011, 10(02):267-275.

[63]杨珺雯,张锦水,朱秀芳,等.随机森林在高光谱遥感数据中降维与分类的应用[J].北京师范大学学报(自然科学版),2015,51(S1):82-88.

[64]安国庆,史哲文,马世峰,等.基于RF特征优选的WOA-SVM变压器故障诊断[J].高压电器,2022,58(02):171-178.

[65]Aydadenta H, Adiwijaya A. A clustering approach for feature selection in microarray data classification using random forest[J]. Journal of Information Processing Systems, 2018, 14(5): 1167-1175.