【摘要】 高压断路器是电网中的关键设备，机械缺陷是其主要故障类型，尽早发现潜在缺陷并及时进行维修意义重大。文中基于计算机视觉技术，提出一种采用角点跟踪法从动态图像中提取运动轨迹、支持向量机进行缺陷识别的高压断路器机械缺陷检测方法。首先，利用高速摄像机获取断路器操动机构的动作过程视频，通过角点跟踪算法获取拐臂角位移曲线；接着，通过主成分分析将高维特征向量降低为新空间中的低维特征向量；最后，利用支持向量机进行缺陷分类。实验测试结果表明，28组训练集、8组测试集的准确率分别为92.86%、87.5%，达到了良好的诊断效果。该方法作为一种非接触式测量方法，无需安装传感器，测量速度快，准确率高，对于现场检修具有重要意义。更多还原

【Abstract】 High-voltage circuit breaker is the key equipment in the power grid,mechanical defect is its main type of fault and it is important to diagnose the mechanical defect and perform maintenance in time. In this paper,a mechanical defect detection method based on computer vision,applying corner tracking method to extract motion trajectory and support vector machine to classify the defects,is proposed. Firstly,the high-speed camera is used to obtain the operation video of the operating mechanism of circuit breaker,and the angular displacement curve of the lever is obtained by corner tracking algorithm. Then,the high-dimensional feature vector is reduced to the low-dimensional feature vector by principal component analysis(PCA). Finally,the defect classification for reduced-dimensional feature vector is implemented by the support vector machine(SVM). The experimental test results show that the accuracy rate of 28 training sets and 8 test sets are 92.86%,87.5% respectively,and good detection results are achieved. The proposed method,as a kind of non-contact measurement method without installation of sensor,is fast in measurement speed and high in accuracy,and has important significance for the maintenance at site.更多还原