Research On Health Warning Technology And Methods For Aircraft Based On QAR Data

With the development of the aviation industry, the continuing civil aviation of security has become a hot issue of growing concern.To monitor the health status of the aircraft;airlines have to install QAR recorders to provide data support for the health warning of aircraft. But, how to make good use of QAR data to assess the health status of aircraft system and predicting failures before the system failure are the main purpose of this paper.The main research contents are as follows:First, although the QAR data contains a large number of operating parameters of the aircraft system, it is difficult to find the fault samples. Through introducing related methods of early healthy warning for aircraft and analyzing profoundly QAR data,the SVM displays excellent advantages of the intelligent fault diagnosis when lacking a large number of fault samples. SVM uses structural risk minimization principle, taking into account the training error and generalization capabilities. In dealing with the small sample data sets and nonlinear problems it has unique advantages, especially for the establishment of fault diagnosis model.Considered of the above problems, this paper will make use of SVM in fault diagnosis of aircraft system.Second, this paper describes the principles of SVM and two typical multi-class classification algorithms, the“one-against-rest” and “one-against-one”. Firstly the article makes experimental analysis of IRIS data set to verify good classification performance of the SVM. Then two typical multi-class classification algorithms(more than one pair of Class and pair classification) are constructed in fault diagnosis of aircraft engine when making the QAR data as research object, which can obtain accurate classification results. Therefore the validity and superiority of SVM in dealing with small sample problems are verified.Third, in order to achieve the purpose of early health warning for aircraft, this article establishes a type of SVM regression and prediction model. Based on the classification result of on aircraft engine fault diagnosing, and uses SVM regression algorithm to make the regression and prediction of a certain kind of fault data. The experiment indicates this algorithm is able to fit the sample data precisely. So the health warning of aircraft system is realized.