Nonlinear System Identification of Variable Oil Pump for Model-Based Controls and Diagnostics

This paper presents nonlinear system identification of a variable oil pump for model-based controls and diagnostics of advanced internal combustion engines. The variable oil pump offers great benefits over the conventional fixed displacement oil pump in terms of fuel efficiency and functional optimality. However, to fully benefit from the variable oil pump, an accurate mathematical model that describes its dynamic behavior is foundational to develop an accurate and robust oil pressure control and diagnostic. Toward this end, Hammerstein and Wiener models that consist of a nonlinear static block followed by a linear dynamic block and a linear dynamic block followed by a nonlinear static block, respectively are developed. Under different operating conditions (oil temperature and engine speed), the oil pressure (output) is measured with the multilevel duty cycle (input) of the flow control valve. The model parameters are estimated sequentially, the nonlinear static block first and linear dynamic block next, through decoupling and implemented as lookup tables. The experimental validation shows the nonlinear model outperforms the linear model in terms of prediction accuracy. The Hammerstein model is demonstrated to have better performance than the Wiener model.