粗糙集在滚动轴承故障诊断中的应用

摘要

The characteristic values of vibration signal of several common pitting fault of the rolling bearing are analyzed in this paper. Discretization algorithm of rough set (RS)based on entropy is used for discretization, and heuristic reduction algorithm based on attribute importance is used for attribute reduction. Then the feature vectors after attribute reduction are input support vector machine (SVM)with RBF kernel function to training,SVM model is built for fault recognition and diagnosis.Experimental results show that by applying the hybrid intelligent diagnostic of rough set combined with support vector machine (RS-SVM),RS is used as the front system of SVM to realize the pre-processing of data,while RS is utilized to reduce the attribute number of information express and the rule number of decision systems of fault diagnosis, through which the input data of SVM can be greatly reduced and the system processing speed is also well improved. Hence ,good results of the failure identification of vibration signal of rolling bearing are finally obtained,which verifies the efficiency and value of rough set theory for the fault diagnosis of rolling bearing.%对滚动轴承几种常见点蚀故障的振动信号特征值进行分析,使用粗糙集基于熵的离散化算法进行属性离散化,并采用基于属性重要度的启发式约简算法进行属性约简,然后选用径向基核函数的支持向量机将经过属性约简的特征向量输入支持向量机训练,建立支持向量机模型并进行故障识别与诊断.实验分析结果表明,应用粗糙集和支持向量机相结合的混合智能诊断方法,将粗糙集作为支持向量机的前置系统对数据进行预处理,利用粗糙集可以减少信息表达的属性数量和故障诊断决策系统的规则数,使支持向量机输入端数据量大大减少,提高系统的处理速度,对于滚动轴承振动信号的故障模式识别可以获得良好的效果.从而验证了粗糙集理论对滚动轴承故障诊断的有效性以及其应用价值.