Research Of Aviation Circuit Board Fault Diagnosis Based On Infrared Technology

The traditional fault detection of aviation circuit boards are a method that voltage,current and other signals are manually measured by instruments and engineers analyse faults based on experience.Not only could not instruments detect high-frequency real-time fault,bring secondary faults to precision circuits.Traditional methods are complex and inefficient.So this technology of fault diagnosis based on infrared detection and health state test method are researched,which improve accuracy,efficiency,generality and reduce cost.First,the system of aviation circuit fault detection that is based on infrared instrument could collect data,process images,extract data and so on.Advanced infrared technology is integrated into detection system by high-performance infrared instrument.Sensitivity and accuracy of fault diagnosis are improved by hardware.There are 3 experiments about circuit model that are carried out to find relationship between working states and infrared characteristics.Second,fault codes of circuit infrared cases are defined to build a standard database including 81 cases.On the basis of fault diagnosis experiments by method of thermal difference and sequences analysis a new fault diagnosis method is proposed to establish relationship between temperature and fault symptom based on multi-level SVM.The 8eigenvalues are extracted to form vectors of working state,and the first-level fault diagnosis SVM model and the second-level data fusion SVM model are established from 8 dimensions.The accuracy of fault diagnosis is improved by genetic algorithm that optimizes parameters c and g.At last,on the basis of a fault prediction experiment by gray model a new fault prediction method is proposed to predict next working cycle's temperature sequences of circuit based on SVM regression algorithm.And temperature sequences are returned to first-level and second-level SVM models to predict fault codes.The failure mechanism of circuit is studied to build the relationship between thermal model and aging degree.The temperature sequences of 19586 flight hours are collected to build an aging model based on infrared temperature that is divided to 6 stages.The actual flight hours and remaining life are calculated by SVM aging model trained.WRT-2100 aircraft weather radar circuit boards are verified by test platform that isdeveloped in a project.Fault diagnosis method based on multi-level SVM that integrates multi-dimensional information is used to classify faults with high accuracy and efficiency.The MSE of fault prediction is low,and the aging model of circuits could calculate aging degree and remaining life.