基于数据挖掘的航空装备故障成品预测模型

Prediction Model of Aviation Equipment Failure Components Based on Data Mining

中文关键词:  航空装备  数据挖掘  文本聚类  故障预测

英文关键词:aviation equipment  data mining  text clustering  failure prediction

中文摘要:

      因航空装备系统复杂度高，故障成因复杂，在保障时根据故障现象判断故障成品的难度很大，同时大量的历史故障记录数据未能有效利用。通过分析航空装备成品故障记录的特点，提出了一种基于数据挖掘的航空装备故障成品预测模型。该模型将历史故障记录数据作为输入，通过文本聚类将大量故障描述聚类得到故障现象簇，并建立 “故障现象”与“故障成品”之间的多对多关系，提出了故障成品概率分布算法，并通过匹配新发生故障现象与故障现象簇，计算出故障成品概率分布。实验验证结果表明，该模型可有效地根据故障描述预测可能发生故障的成品的概率分布，且预测准确率可随数据量的增加而提高，满足实际保障需求。

英文摘要:

      Due to the high complexity of the aviation equipment system and the complicated causes of failure,it is very difficult to judge the failure component based on the failure phenomenon during the troubleshooting,and a large amount of historical fault record data cannot be effectively used.By analyzing the characteristics of the failure records of aviation equipment components,a prediction model of failure components of aviation equipment based on data mining is proposed.Taking historical failure record data as input,a large number of failure descriptions are clustered to obtain failure phenomenon clusters through text clustering,and a many to many relationship between “failure phenomenon” and “failure components” is established,and the probability distribution of failure component algorithm is proposed.By matching the newly occurring failure description and the failure phenomenon cluster,the probability distribution of the failure components is calculated.Experimental verification results show that the model can effectively predict the probability distribution of components that may fail based on the description of the failure,and the accuracy of the prediction can be improved with the increase of the amount of data to meet the actual guarantee requirements.