A Fuzzy Decision-Making System for Automotive Application

Fault diagnosis for automotive systems is driven by government regulations, vehicle repairability, and customer satisfaction. Several methods have been developed to detect and isolate faults in automotive systems, subsystems and components with special emphasis on those faults that affect the exhaust gas emission levels. Limit checks, model-based, and knowledge-based methods are applied for diagnosing malfunctions in emission control systems. Incipient and partial faults may be hard to detect when using a detection scheme that implements any of the previously mentioned methods individually; the integration of model-based and knowledge-based diagnostic methods may provide a more robust approach. In the present paper, use is made of fuzzy residual evaluation and of a fuzzy expert system to improve the performance of a fault detection method based on a mathematical model of the engine. The fuzzy residual evaluation reduces the detection errors in the model-based residual generator, while the fuzzy expert system allows integration of exhaust gas emissions data (engine-out or tailpipe) into the diagnostics. The paper presents an overview of the theory of the method, and illustrates its application by way of an experimental study conducted on a production engine.