基于深度学习的整车辐射骚扰故障诊断方法研究OA

In automotive electromagnetic compatibility(EMC)testing,excessive radiated emissions present a significant challenge,making the rapid localization and accurate diagnosis of fault sources a critical technical issue.To address this problem,this paper proposed a deep learning-based fault diagnosis method for radiated emissions in complete vehicles,aiming to identify radiated emission faults that exceed the limits specified under regulatory limit requirements.Four typical fault sources responsible for excessive electromagnetic radiated emissions were considered,namely air conditioning controllers,DC-DC converters,engine ignition systems,and switch tubes at high frequencies.First,the test signals were preprocessed to obtain the preprocessed signals.Subsequently,the short-time Fourier transform was used to convert the preprocessed signals into two-dimensional time-frequency image datasets,and the MobileNetV3 deep learning model was selected for fault classification.Experimental results show that the proposed method achieves 93%training accuracy in model training,showing good diagnostic effects and high robustness,which proves that the signal preprocessing method can effectively improve the classification accuracy of the model.