Sensor fault detection and isolation in diesel air path using fuzzy-ARTMAP neural network

Fault detection and isolation have become one of the most important aspects of automotive diagnosis. In this paper, a new approach is proposed dealing with fault detection and isolation problem in diesel engine. Especially, the sensors fault detection and isolation problem in diesel air path is studied. The proposed solution is realized in two stages. In the first one, we classify the unfaulty functioning data of system using the fuzzy-ARTMAP classification in order to model the engine dynamics. In the second stage, a conflict is evaluated between samples of test data based on the hyper-rectangles resulted in the first stage. Two samples are in conflict if their intersection does not belong to the neural model elaborated by fuzzy-ARTMAP. The model is learned and validated using data generated by xMOD software. This tool is also used for test. Finally, to illustrate our approach, some simulation results are given and discussed.