Mass air flow sensor diagnostics for adaptive fueling control of internal combustion engines

Abstract

Presents an information synthesis (IS) approach for the mass air flow (MAF) sensor diagnosis on internal combustion engines. An information synthesis solution is attractive for diagnostics since the algorithm automatically calibrates itself, reduces the number of false detections and compresses a large amount of engine health information into the model coefficients. There are three primary parts to information synthesis diagnostics. First, an IS model is used to predict the MAF sensor output based on the engine operating condition. The inputs to this IS model include the throttle position sensor (TPS) and the engine speed sensor information. The second part concerns an adaptation process that is used to reduce the errors between the IS model output and the actual MAF sensor output. Finally the adapted model coefficients are used to diagnose the sensor as well as identify the source for changes in the sensor characteristics. This proposed solution is experimentally tested and validated on a Ford 4.6 L V-8 fuel injected engine. The specific MAF sensor faults to be identified include sensor bias and a leak in the intake manifold.