Modeling, fault detection and diagnosis of an automotive engine using artificial neural networks

In this thesis, we studied the feasibility of using artificial neural network (ANN) models to develop model-based techniques for on-board failure detection and isolation (FDI) in spark ignition (SI) engines. Due to the preliminary nature of this study, a subset of engine subsystems was selected for this research. This included subsystems that involved air dynamics, i.e., throttle body, intake manifold, and exhaust gas recirculation processes. These processes are highly nonlinear, and as a result are difficult to model. In this thesis, data from an instrumented Buick Regal, equipped with a V6 3800 engine, were used to develop ANN based models for the above systems. The models were then used in a decision-making process to detect and isolate faults in the manifold pressure sensor and in the exhaust gas recirculating (EGR) valve.; This study revealed that the ANNs have great potential for developing techniques for the OBD II. We believe that ANN-based techniques are superior in general to other engine diagnostic approaches, since the former have the potential for systematically detecting and isolating a variety of soft incipient failures under different engine operating conditions. (Abstract shortened by UMI.)...