LPV Modeling and Mixed Constrained <inline-formula><tex-math>${{H_2}/{H_\infty}}$</tex-math> </inline-formula> Control of an Electronic Throttle

Abstract

Engine electronic throttle control (ETC) is challenging due to its high system nonlinearities and the required fast response time. In this paper, an electronic throttle system was modeled as a linear parameter varying (LPV) system in discrete-time domain, where the vehicle battery voltage is modeled as the measurable time-varying parameter; the nonlinear friction is modeled as a function of the measurable throttle position; and the limp-home spring nonlinearity is compensated through feedforward control. Mixed constrained ${{H_2}/{H_\infty}}$ LPV controller was designed for the LPV throttle control system using the linear matrix inequality convex optimization approach. The system output and control weights were optimized through simulation studies to achieve the best performance; and the finalized LPV controller was experimentally validated on an ETC test bench. Comparing with the baseline well-tuned fixed gain proportional, integral, and derivative controller, the LPV controller reduces the 2% settling time from 0.30 to 0.15 s. Especially, smooth transient response was achieved when the throttle plate crosses the region with heavy spring nonlinearity required by the limp-home operation.