System Identification for Air/Fuel Ratio Modeling Using Switching Sensors
Modeling the internal combustion engine for air-to-fuel ratio (AFR) control has been widely studied and several methodologies have been adopted toward the end goal of applying model based control schemes. In this paper, an online binary sensor identification (BID) technique using switching sensors is adopted for modeling the response from fuel input to AFR output of a spark-ignited, internal combustion engine, to be used in AFR control. In general terms, the algorithm identifies the impulse response of a linear time invariant (LTI) system by choosing an optimal sequence of inputs. The entire modeling process is done online with a four-cylinder engine in a test cell, using typical production switching sensors. Finite impulse response (FIR) linear time invariant (LTI) models are identified at prescribed operating points of the engine (specified by engine speed and the manifold air pressure). The validity of the resulting model is then tested on separate data streams with AFR measured from a wide-range sensor output. By scheduling the coefficients of the FIR models based on the operating condition, it is possible to identify a linear parameter varying AFR model for the appropriate operating regions of the engine.