基于支持向量机的机载产品延长日历寿命方法  

Research on SVM-based Predicting Method on Calendar Life Extension of Airborne Products

作　　者：李郑琦[1] 何宇廷[2] 邵青[2] 魏鹏[2] 

机构地区：[1]中国飞行试验研究院,陕西西安710089 [2]空军工程大学工程学院,陕西西安710038

出　　处：《空军工程大学学报（自然科学版）》2010年第4期6-10,共5页Journal of Air Force Engineering University(Natural Science Edition)

基　　金：国防预研基金资助项目(513270501)

摘　　要：针对现有的外场数据统计法在确定飞机机载产品寿命指标上的局限性,建立了机载产品使用影响因素体系,利用某型机载产品在不同典型使用环境下的故障率数据,建立支持向量机回归分析模型,通过机器学习掌握已知机载产品使用影响因素向量和故障率数据的相互关系,根据已知的产品故障率数据对未知寿命进行预测。利用8个单位的产品故障率来预测另一单位的产品故障率,并给出了算例分析。计算结果与实际情况相吻合,表明该方法具有一定的应用价值。In this paper an influencing factor system of airborne products is established based on a comprehensive consideration of factors like geography, climate, service situation and maintenance level. SVM regression analysis model and forecasting model of airborne products based on SVM are introduced. Then a life prediction model based on SVM is built according to the failure rate data of an airborne product equipped in some representative conditions. It can predict the service life of the product by finding out the relationship between the failure rate and the influencing factors, accordingly it can make up the localization of outfield data statistical method. A case shown in this paper presents the life forecasting process of before - mentioned method and the result indicates that this method is of some value for practical application.

关 键 词：机载产品 日历寿命 延寿 支持向量机 回归分析模型 

分 类 号：V241.01[航空宇航科学与技术—飞行器设计]