基于信息熵及模糊熵聚类算法的油液监测数据关联性

应用信息熵及模糊熵聚类算法对油液监测数据间的关联关系进行考察．根据系统输出的数据较为简化地表征系统内部因素间的相互联系程度，对得到的关联关系进行分级、评价，以便对设备的磨损状况提供可信、可靠的分析手段．通过应用Shannon信息论的互信息熵理论表征数据序列之间的关联性，通过模糊熵理论进行阈值的选取，以标准5函数为隶属函数，对关联数据进行挖掘．实例验证了算法的有效性，并对聚类结果进行了解释．对设备故障的定位与磨损状况的评级划分提出了一种量度手段．

The Correlativity of Oil Monitoring Information Based on Mutual Information and Fuzzy Entropy Clustering

A Shannon's mutual information and fuzzy entropy clustering based oil analysis method was presented. According to the results of the clusters, the relationship of all the oil monitoring information is expressed. The interpretation of these clusters can provide us a new viewpoint on analyzing the materials of main wear component and a new measure on wear state and identify probable fault location. Tools based on this technique can bring a useful extra understanding of the oil monitoring data. The method of using MaxEPD to estimate the probability density distribution of oil monitoring data is adopted. The standard S membership function is selected and the threshold is determined by the local maximum fuzzy entropy of the fuzzy set. Through the analysis and discussion of an example, the algorithm is validated to be an effective way of studying of the correlativity of oil monitoring information.