Nonlinear Predictive Control of Transients in Automotive Variable Cam Timing Engine Using Nonlinear Parametric Approximation

Abstract

The paper considers design of a predictive Linear Time Varying model-based controller with nonlinear feedforward for regulation of transient processes caused by setpoint step changes in a nonlinear plant. An optimal feedforward control sequence is computed based on an empirical Finite Impulse Response model of the process. Though the control techniques developed in this paper are meant to have more general industrial applicability, a specific automotive engine control application—control of Variable Cam Timing automotive engine—is pursued. An advantage of the proposed controller design in this problem is that no first principle models are required. Instead, nonlinear parametric approximations of a neural network type are being used to describe and identify static nonlinear mappings encountered in the problem. A number of simplifying assumptions and approximations are made to make practical implementation of the proposed scheme possible. Validity of the designed controller is verified by simulation. The proposed “model-free” design can potentially increase flexibility and save labor in development and deployment of such controllers for industrial systems.