Improving tracking performance of automotive air conditioning system via μ synthesis

Abstract

In order to improve the performances of air conditioning systems, it is desirable to track time-varying trajectories generated by optimization algorithms, which H∞ synthesis techniques have been proved to successfully solve. However, the control-oriented models of vapor compressor cycles used for algorithm development, even if built from first-principles, suffer from model uncertainties introduced by modeling assumptions, calibration inaccuracies, and linearization errors. The differences between the actual plant and the control-oriented model, mainly in the form of unmodeled dynamics and parameter uncertainty, undermine the stable margin as well as the performance of the closed-loop system with H∞ controllers. In order to solve the problem, the concept of the structured singular value μ is used to analyze the influences of model uncertainties on robust stability and robust performance. Based on μ analysis results, μ synthesis techniques compared to H∞ methods achieve better stability and performance margins over the same set of uncertainties. Furthermore, simulation results show that the μ controller achieves better performances of output tracking and disturbance rejection than the H∞ controller for the automotive air conditioning system studied.