| Nowadays,the Internet and the automotive industry are developing rapidly.There are more and more electronic devices in vehicles.The electromagnetic environment of vehicles is becoming more and more complex.The electromagnetic compatibility of vehicles has attracted widespread attention.At present,in the diagnosis,optimization and suppression of automobile electromagnetic interference,it mainly relies on the practical experience and actual vehicle testing of electromagnetic compatibility engineers.How to extract electromagnetic compatibility development experience accumulated by electromagnetic compatibility engineers for many years into intelligent analysis rules is the focus of research.Based on the existing electromagnetic compatibility test data,this thesis proposes an EMC test data feature extraction algorithm,and based on the feature data,it is combined with the actual vehicle test rectification report and the engineer’s experience to establish a small sample data set and learning mechanism.Reverse diagnosis model,classify and predict the source of EMC faults,effectively provide the direction for EMC fault diagnosis,and based on the research of this thesis,design and develop a B /S-based EMC test data analysis system.The EMC test data feature extraction algorithm and the constructed model were applied.The details are as follows:First of all,the related background and significance of the reverse diagnosis of vehicle EMC faults are elaborated,and the domestic and foreign research status of EMC test,fault diagnosis and fault diagnosis based on machine learning are introduced in detail.Then,briefly introduces the related concepts of electromagnetic compatibility,analyzes the method of extracting the characteristics of electromagnetic compatibility test data,briefly introduces the previous fault diagnosis methods,and proposes the framework of the fault diagnosis method used in this study.The classifiers required for reverse diagnostic modeling are introducedThen,analyze the existing electromagnetic compatibility test data,design and propose an EMC test data feature extraction algorithm,and then combine the feature data with the engineer ’s existing experience and actual vehicle test experience to correlate machine learning The algorithm is applied to the establishment of a model for reverse diagnosis of vehicle electromagnetic compatibility testing.The SVM algorithm and the Naive Bayes algorithm are used to establish models for comparative experiments.Finally,the results of the experiments using the two classifiers are displayed and analyzed.The experimental results show that the reverse diagnosis model based on SVM is more suitable for the research in this thesis and can provide good support for EMC fault diagnosis.The experiment finally shows the actual application developed based on the model studied in this thesis,namely the EMC test data analysis system,and uses the actual vehicle EMC test data to verify the results of the system. |
