无量纲免疫支持向量机的复合故障诊断方法

Complex fault diagnosis using dimensionless immune support vector machine

针对机组多故障并发时,故障特征互相干扰,产生模式混淆,难以准确分类,提出一种无量纲免疫支持向量机的复合故障诊断方法。由于五种无量纲指标对不同频段复合故障的敏感性不同,将无量纲指标和人工免疫的阴性选择算法相结合,通过选择合适的编码位数来提取故障特征,多分类支持向量机(MSVM)的参数经过免疫优化算法训练后获得最优解,把五种时域特征的无量纲指标提取的故障特征向量输入到MSVM,学习后的MSVM应用于故障诊断。实验结果表明优化后的支持向量机对小样本具有良好的分类性能和实时性,无量纲免疫MSVM与MSVM模型相比能够更加快速、准确进行复合故障诊断。

Due to the problem of difficult classification, fault features interfering with each other, generating mode confusion when concurrent multi-fault for unit occur, it is necessary to establish a suitable dimensionless parameters immune Multi-Support Vector Machine（MSVM） method for complex fault diagnosis. This paper presents a new method that 5 types of dimensionless parameters with different sensitivity and artificial immune negative selection are combined to extract fault feature by selecting the appropriate number of encoded bits. MSVM＇ s parameter is optimized by immune optimization algorithm to obtain the optimal solution. Multi-class support vector machine is learned after 5 types of dimensionless parameters input to it. Then it can be directly applied to the fault diagnosis. Experimental results show that the MSVM has a good classification performance for small sample, and comolex fault diaanosis has a rapid speed and a high accuracy can be achieved by this aoDroach.