The Teaching Application of Deep Learning in Diagnosis of Automobile Abnormal Sound Faults

Traditional manual diagnosis methods are often used in teaching and practice of vehicle faults diagnosis. It leads to low accuracy in detecting abnormal sound in vehicles and low student interest in learning. With the development of deep learning and artificial intelligence technology, using the above methods to identify the types of abnormal sound of vehicles can effectively improve the accuracy of vehicle faults diagnosis and achieve the purpose of improving the teaching quality of faults diagnosis. In this paper, microphone array is used to achieve abnormal sound location, and transfer learning for the Audio domain is used to classify and diagnose abnormal sound. From the experimental results, transfer learning has demonstrated significant feasibility and high accuracy in diagnosing vehicle abnormal sound faults. It provides innovative directions and has a certain leading role in the teaching and practice of vehicle fault diagnosis.