Abstract: Aiming at the condition monitoring data’s features of large-volume,high-dimension and complex-types in complex equipment,a mutual information-based association detecting and modeling method was proposed to mining the association between each monitoring data.The concept of monitoring data’s association was defined based on mutual information,and the associative data in monitoring data was found by association discovery algorithm concentrately. An experiment using real condition monitoring data showed that many obvious associations could be easily detected and modeled by proposed approach,and some potential underlying relationships between different dimensions of the data were found.

Key words: complex equipment, condition monitoring data, mutual information, association, modeling

摘要： 针对数据量大、维度高、类型复杂的复杂装备状态监测数据,为挖掘各维监测数据之间的关联关系,提出一种以互信息为基础的相关性发现及建模方法。基于互信息定义了监测数据相关度的概念,使用一种相关性发现计算方法找到监测数据集中有相关性的数据。使用实际状态监测数据对该方法进行了实验验证,结果表明,该方法可以找到很多具有明确物理意义的直接关联关系,也可以发现部分潜在关联关系。

关键词: 复杂装备, 状态监测数据, 互信息, 关联关系, 建模