【摘要】 针对经验模态分解应用于故障检测中存在的模态混叠问题,提出一种基于变分模态分解-近似熵与支持向量机结合的航空串联型电弧故障检测方法。以实验线路中不同负载类型的电流信号为研究对象,利用变分模态分解算法将其分解为4组固有模态分量,分别计算前三组固有模态分量的近似熵值作为特征向量,输入支持向量机进行电弧故障识别。将所用特征提取方法与基于经验模态分解-近似熵的方法进行比较,实验结果表明,基于变分模态分解-近似熵的特征提取方法以其抑制模态混叠产生的性质弥补了经验模态分解算法在信号处理方面的不足,与支持向量机相结合可以准确可靠地识别电弧故障;在负载已知的情况下,其识别准确率高达98%以上,负载未知时,其识别准确率分布在93. 75%～97. 5%之间。更多还原

【Abstract】 Aiming at the problem of modal aliasing in empirical mode decomposition in fault detection,an aviation series arc fault detection method based the variation modal decomposition-approximate entropy and support vector machine is proposed. Used the current signals of different load types in the experimental line as the research object,decomposed it into 4 groups of intrinsic modal components using the variation mode decomposition. The approximate entropy values of the first three groups of intrinsic modal components were calculated as feature vectors. The support vector machine was used for arc fault recognition.The experimental results show that the feature extraction method based on the variation mode decomposition-approximate entropy makes up for the deficiency of empirical mode decomposition algorithm in signal processing due to its property of suppressing modal aliasing. Combined with support vector machines,it can accurately and reliably identify arc faults. When the load is known,the recognition accuracy rate is as high as 98% or more; and when the load is unknown,the recognition accuracy rate distribution is between 93. 75% and 97. 5%.更多还原