基于SO-PAA-GAF和AdaBoost集成学习的高压断路器故障诊断

针对在小样本和复杂工况下高压断路器故障诊断识别精度不高的问题，提出一种基于振动信号处理和Ada Boost集成学习的高压断路器故障诊断方法。首先，搭建高压断路器实验平台并采集8种工况下的分闸振动信号。其次，对振动信号进行绝对值处理后，使用分段聚合近似(piecewise aggregate approximation, PAA)进行分段平均，将输出的新序列采用格拉姆角场(Gramian angular field, GAF)转换成图片，并使用Relief F方法对提取的高维图片特征进行重要度排序。最后，将保留的重要特征输入到Ada Boost集成学习模型进行故障诊断，并用蛇优化算法确定最优PAA分段步长和输入分类器特征数量，以进一步提高故障诊断精度。通过分析多种信号处理方式及分类模型可知，图片信号和Ada Boost集成学习模型能够有效处理振动信号并准确判断故障类型，为准确、可靠地诊断高压断路器故障提供了新途径。

Fault diagnosis of high-voltage circuit breaker based on SO-PAA-GAF and AdaBoost ensemble learning

Aiming at the low accuracy of high-voltage circuit breaker fault diagnosis under small samples and complex working conditions, a fault diagnosis method of high-voltage circuit breaker based on vibration signal processing and AdaBoost ensemble learning is proposed. First, the high-voltage circuit breaker test platform is built and the switching vibration signals are collected under 8 working conditions. Second, after absolute value processing of vibration signals, piecewise aggregate approximation (PAA) is used to do piecewise averaging, and Gramian angular field (GAF) is used to convert new output sequences into pictures. The Relief F method is used to sort the importance of the extracted high-dimensional image features. Finally, the retained important features are input into the AdaBoost ensemble learning model for fault diagnosis, and the snake optimization algorithm is used to determine the optimal PAA step size and the number of input classifier features to further improve the fault diagnosis accuracy. The comparison and analysis results with various signal processing methods and classification models indicate that picture signal and AdaBoost ensemble learning model can deal with vibration signal effectively and judge fault type accurately, which provides a new way to diagnose high-voltage circuit breaker fault accurately and reliably.