数学形态学在航空串联故障电弧检测中的应用研究

Application of Mathematical Morphology in the Detection of Aviation Series Arc Fault

为提高航空串联故障电弧判别的可靠性以及实现对故障模式的准确分类,提出了一种数学形态学梯度和极端学习机相结合的航空交流故障电弧检测方法。通过数学形态学梯度提取线路中电流信号的相关故障特征参数,采用分形维数的思想对特征参数进行处理,引入极端学习机对故障进行分类识别,实现了最终的诊断。试验计算结果表明,串联故障电弧诊断率高达97.5%,所设计的诊断方法具有良好的泛化能力。

In order to improve the reliability of aviation series arc fault identification and realize the accurate classification of the fault modes, an AC arc fault detection method was proposed based on mathematical morphological gradient and extreme learning machine. Extracting the fauh characteristic parameters of current signal in lines through mathematical morphology, the fractal dimension thought is adopted to process the characteristic parameters, introducing extreme learning machine to classify the fault, the final diagnosis was achieved. The experimental results show that the series arc fault diagnosis rate is as high as 97.5% ,and this diagnosis method has good generalization ability.