Model-based air leak detection for turbocharged gasoline engines without a hot-film air mass flow meter sensor

Abstract

This paper presents an observer-based fault detection and isolation strategy to detect and isolate an air leakage in the compressor–throttle path and a fault in the boost pressure sensor in a turbocharged gasoline engine. The strategy is applicable to engines not equipped with a hot-film air mass flow meter sensor. A novel combination of a turbocharger model and a control volume model for the pressure dynamics of the intercooler is used to design an asymptotically stable non-linear observer which estimates the unmeasured turbocharger variables. The turbocharger observer’s estimation error is used as a residual in the fault detection and isolation process where the effects of the measurement noise and the model simplifications are cancelled using the maximum-likelihood ratio as the test statistic. An engine charge estimation observer is added to the turbocharger observer for detection and isolation of the air leakage from the sensor fault. Validation of all models and the fault detection and isolation strategy is carried out for a modern gasoline turbocharged engine. The experimental results reveal the ability of the control-oriented strategy to perform online detection and isolation of air leakage through a hole with a diameter of 5 mm or larger.