基于模式矩阵的液体火箭发动机试车台故障关联规则挖掘

Association Rules of Liquid-Propellant Rocket Engine Test-Bed Based on Pattern Matrix

为发现试车台状态变量与其工作状况的关联关系,本文对液体火箭发动机试车台历史测试数据进行关联规则挖掘,对试车台状态参数进行选择和编码,使用自适应阈值算法对测试数据进行离散化,引入模式矩阵用于计算事务数据库中的频繁项集,提出了基于模式矩阵的关联规则挖掘方法,并将该方法应用于对多组历史测试数据进行频繁项集挖掘,对挖掘结果的分析表明:与传统方法相比,本文方法效率高,计算量小,挖掘结果能正确反映试车台状态变量与其工作状况间的关联规则,为试车台的测试数据挖掘提供了一种有效方法。

In order to discover the associated relation between state variables and operating conditions of LRE test-bed,the association rules are mined from historical test data of test-bed.The state parameters of test-bed are elected and encoded.The test data are discretized by using an adapting threshold algorithm.The pattern matrix is introduced to calculate the frequent item set in the transaction database,and a method for mining association rules based on pattern matrix is presented.this method was applied in mining frequent item set from several historical test sample and the result is analyzed.The result shows that compared with traditional methods,the method in this paper is efficient and has smaller computational effort.The mining result can reflect the association rules between state variables and operating conditions of test-bed correctly.The method in this paper is validated to mine test-data of LRE test-bed.