针对电力设备红外图像批量诊断中故障特征参量提取及参数配置难题,采用粒子群算法(PSO)与Niblack算法相结合的方法,将设备热像从背景中分割出来并提取出设备的最低、最高及平均温度等参量,通过计算设备各温升特征,构建支持向量机(SVM)样本特征空间.采用优化的蝙蝠算法(BA)对SVM参数进行寻优,并利用最优参数配置下的SVM实现设备故障诊断.对220组图像样本测试结果表明:该红外图像故障诊断方法在电力设备热故障缺陷检测方面的效率及准确率较高,适用于电力大数据中非结构化红外图像的批量分析与处理.%Aiming at the problem of defect test and parameter assignment in the batch diagnosis of power equipment infrared image,PSO and Niblack algorithm are used to separate the equipment thermal image from the background and extract the lowest,highest and average temperature.Then,the SVMsample feature space can be constructed by calcu-lating the temperature rise characteristics of the equipment.The support vector machine(SVM)parameters are opti-mized by using the optimized bat algorithm(BA),and the equipment defects diagnosis is realized by SVM under the optimal parameter configuration.According to the 220 groups of image sample testing results,the proposed method has high efficiency and accuracy in thermal defects detection of power equipment,and is suitable for batch analysis and processing of unstructured infrared images in large power data.
展开▼