Application Research Of Data Mining In Fault Diagnosis Of Key Parts Of Satellite

Satellites usually pass ground tests before they are launched.Most failures are caused by changes in the physics of the materials used to make the parts.A satellite is a complex system.In case of failure in orbit flight,only telemetry data of some parameters can be used for failure analysis on the ground,and they are stored in the database in a certain order.These satellite telemetry data contain a large number of objective laws and knowledge that can be used for satellite fault diagnosis,and it is of great significance for the decision-making and management of satellite fault diagnosis to effectively understand,grasp and utilize the laws of each component and device of the satellite mined from them.Based on the research of the data mining algorithm,this paper USES the data mining algorithm to study the fault diagnosis of the key parts of the satellite-T he main research contents include the following:(1)Aiming at the problem that satellite fault data are easy to be missing,a missing value filling algorithm based on information entropy data fusion is proposed.Using the information entropy in information theory,the prediction results of SVM and RBF are fused by the entropy weight fusion method.The experiment shows that the prediction accuracy of this method is more accurate than that of the single prediction method.Aiming at the problem that outliers are difficult to detect,an improved k-means method is proposed to diagnose outliers.(2)For the problems with more satellite parameters and higher data dimensions,principal component analysis(PCA),core principal component analysis(KPCA)and local linear embedding(LLE)were used to extract the characteristics of experimental data.The k-nearest neighbor(KNN)algorithm was used to test the extracted data.The KPCA method was used to detect the fault.The proj ection matrix of the original data was obtained through the KPCA method,and the principal component model of the original data was established.Then the statistical detection index(T2,SPE)was used to detect the operation process.(3)SVM multi-classification algorithm is adopted to carry out fault diagnosis for problems with many types of failures in key parts of satellites;Particle swarm optimization algorithm(PSO)was adopted to optimize the SVM parameter penalty factor and the problem that it is difficult to select the kernel parameters of the radial basis function.(4)An intelligent aided decision-making system for fault diagnosis of satellite integrated electronic system is developed by using MATLAB and C#hybrid programming.