基于变精度粗糙集-支持向量机的断路器故障诊断

Breaker Fault Diagnosis Based on Variable Precision Rough Set-Support Vector Machine

由于断路器运行时监测数据量大,以前采用的反向传播算法(back propagation,BP)神经网络、径向基函数(radical basis function,RBF)神经网络的故障诊断存在网络结构复杂、诊断速率慢等缺点,为了快速准确地得出断路器故障原因,提出一种基于变精度粗糙集-支持向量机的断路器故障诊断算法。利用变精度粗糙集约简决策表去除断路器庞大监测数据里的冗余信息,降低过程数据的维度,并结合支持向量机对粗糙集处理后的信息进行故障诊断,可减少诊断时的主观因素,并具有容错性和解释性,该诊断方法是可以有效实施的。

For reason of great numbers of monitoring data for breaker running,there are disadvantages of complicated net-work structure and slow diagnosis speed for fault diagnosis by using back propagation neural network method and radical ba-sis function neural network method.For rapidly and correctly finding reason for breaker fault,this paper proposes a kind of breaker fault diagnosis algorithm based on variable precision rough set-support vector machine which uses variable precision rough set reduction decision table to remove redundant information in mass monitoring data of the breaker and reduce di-mensionality of process data.Meanwhile,combining support vector machine,it is able to conduct fault diagnosis on informa-tion after being processed by rough set which may reduce subjective factors for diagnosis and be provided with fault tolerance and explanatory characteristic.This diagnosis method was proved to be effectively implementary.