惯性导航状态数据分析与基于SVM的故障诊断模型构建

为更好地利用当前航空装备系统积累的大量飞行数据, 从中挖掘信息来更有效地服务于空军装备可靠性为中心的维修保障工作, 将数据分析技术应用到当前的装备故障诊断中, 设计并实现了某型飞机惯性导航系统状态数据分析和故障诊断系统. 分析了整个系统的数据处理过程, 对数据自动判读处理、数据清理和修正进行了研究. 设计了支持向量机模型, 提出了通过对惯导状态信息数据进行回归预测, 用预测值与实际值之间误差诊断故障; 通过状态数据和与之对应的故障信息库, 用支持向量机进行故障分类两种方法来实现故障的诊断和决策. 最后对模型进行了验证和分析.

Inertial navigation condition data analysis and establishment of diagnostic model based on SVM

In order to take full advantages of the mass aviation data from the air force equipment,and find useful information which could serve the RCM(reliability centered maintenance) better,the modern data analysis technologies is introduced to current diagnostics,and an INS(inertial navigation system) condition data analysis and diagnostic system of one kind of plane is designed.This paper describes the data analysis procedure of the system,studies on the auto-reading and judgment,data filter and modification of it.Then an SVM model is presented,which contains the regression and the classification model.Based on these models,two methods are putted forward;one diagnoses the equipments by comparing the forecast value from the regression model with real value,the other by classifying the faults through condition data and corresponding fault information database.Finally a group of real data are used to verify and evaluate these methods.