An evaluation of engine faults diagnostics using artificial neural networks

Application of artificial neural network (ANN)-based method to perform engi ne condition monitoring and fault diagnosis is evaluated Back-propagation, feedforward neural nets are employed for constructing engine diagnostic net works. Noise-contained training and testing data ar e generated rising an i nfluence coefficient,matrix and the data scatters. The results indicate tha t under high-level noise conditions ANN fault diagnosis can only achieve a 50-60 percent success rate. For situations where sensor scatters are compar able to those of the normal engine operation, the success rates for both fo ur-input and eight-input ANN diagnoses achieve high scores which satisfy th e minimum 90 per cent requirement. It is surprising to find that the succes s rate of the four-input diagnosis is almost as good as that of the eight-i nput. Although the ANN-based method possesses certain capability in resisti ng the influence of input noise, it is found that a preprocessor that can p erform sensor data validation is of paramount importance. Autoassociative n eural network (AANN) is introduced to reduce the noise level contained it i s shown that the noise call be greatly filtered to result in a higher succe ss rate of diagnosis. This AANN data validation preprocessor call also serv e as an instant trend detector which greatly improves the current smoothing methods ill trend detection. It is concluded that ANN-based fault diagnost ic method is of great potential for future rise, However, further investiga tions using actual engine data have to be done to validate the present find ings.