基于聚类、粗糙集和支持向量机的故障诊断

故障诊断有多种方法,利用支持向量机进行故障诊断是其中一种比较有效的方法。但是,故障征兆往往有多个,很难确定哪些征兆是关键的,哪些征兆是冗余的,这样就要搜集处理大量的数据,使支持向量机的结构变得复杂,诊断效率不高。为了提高故障诊断的效率,文中提出一种将K均值聚类、粗糙集、支持向量机相结合进行故障诊断的方法。这种方法首先利用K均值聚类对数据进行预处理,然后利用粗糙集对属性进行约简,最后再用支持向量机进行故障诊断。这样可以充分发挥粗糙集与支持向量机各自的优势,实例证明它可以提高故障诊断的速度和精确度,是一种较好的故障诊断方法。

Default Diagnosis Based on Clustering,RS and SVM

There are many ways to diagnose default, support vector machine is better one.However, it is difficult to make certain which symptom is key or redundancy when default systems are a great lot.So,a mass of data are have to be collected and dealt with,which makes the structure of SVM complex and diagnosis inefficient. In order to improve efficiency ,a new default diagnosis method that use K-means clustering,rough sets and support vector machine is presented.First,data are pretreated through K-means clustering. Then,attributes are reduced by using rough sets. At last , SVM diagnoses default.The method can take full advantage of superiority of RS and SVM and can boost the speed and precision of default diagnosis.The application of an example proves it is an excellent default diagnosis means.