Intelligent vehicle fault diagnosis system using Neural Networks

Abstract

Diagnosis has become a very complex and critical task in determining the condition of vehicle engine. Sound emitted by the engine is always considered to be an annoying noise but a detaiedl analysis of the sound signal shows that noise emnated from a vehicle engine may vary for different fault conditions. In order to clearly diagnose the faults present in a vehicle engine the real time data are collected from various vehicle engines. Simple methods are proposed for recording the vehicle engine sound signal emanated using microphones. By applying auto regressive modeling method, features are extracted from the power spectral variations of the vehicle engine noise. The features are then associated to the expert's opinion to formulate a neural network model that can identify the faults automatically. A simple feed forward neural network model trained by back propagation procedure is proposed