液体火箭发动机故障的数值型关联规则挖掘  

Mining quantitative association rules of liquid propellant rocket engine

作　　者：李京浩[1] 胡小平[1] 韩泉东[1] 

机构地区：[1]国防科技大学航天与材料工程学院,湖南长沙410073

出　　处：《火箭推进》2007年第2期7-11,58,共6页Journal of Rocket Propulsion

基　　金：国家自然科学基金资助项目(50376073)。

摘　　要：数值型关联规则的算法大多是将多值属性关联规则挖掘问题转化为布尔型关联规则挖掘问题,而连续属性的离散化是数值型关联规则的核心问题。本文基于数值型关联规则的理论,用一种数理统计的方法进行连续属性的离散化。将该方法应用于某大型液体火箭发动机稳态段的热试车数据,然后利用FP-Growth算法对其进行测试,挖掘出了故障数据,进而验证了其可行性。Most quantitative association rules transform mining association rules of numeric property into boolean property, and the kernel problem is to divide the numeric data into intervals. Based on the theory of quantitative association rules, the numeric data is divided into intervals with statistical method. This method is applied to the measured steady data of a large -scale liquid propellant rocket engine and tested with FP-Growth arithmetic. The fault data is mined and the feasi- bility of the method is verified.

关 键 词：液体推进剂火箭发动机 故障诊断 数值型关联规则 FP-GROWTH算法 

分 类 号：V434.1[航空宇航科学与技术—航空宇航推进理论与工程]