Engine fault diagnosis based on multi-resolution analysis and dual microphones

Abstract

In order to diagnose the abnormal sound fault of the diesel without disassemble, dual microphones are used to collect noise signal of the engine in this paper, and the sound pressure signal is decomposed with wavelet decomposition technique. By comparing different wavelet decomposition techniques, it determines the sound feature parameter for fault diagnosis, and establishes a BP neural network for pattern recognition. The test result shows that diagnosing engine fault with dual microphones is accuracy, and it can be applied to diagnose the abnormal sound fault of engine.