Nonparametric Design of Robust Linear Controllers and their Experimental Application to Idle Control

S. Zhao, A. Abass, and A.T. Shenton (UK)

Keywords

nonparametric, optimisation, powertrain, robust control,spectral analysis , STFT

Abstract

Several approaches to the nonparametric design of linear robust controllers based on nonparametric frequency response identification using the short-time Fourier transform (STFT) are presented. The analytic function optimisation techniques and algorithms of Helton and Merino[1] used as a nonparametric (NP) controller design method are described. The theory for applying the approach using Helton’s multiple sensitivity-function split-frequency ’disk-method’ and the mixed sensitivity analytic function optimisation technique are presented. The methods are all applied to an engine idle-speed control problem using dynamometer based experimental test data. The resulting controller is then experimentally validated and compared to the experimental results for a similarly designed controller obtained by a parametric approach. It is found that the nonparametric techniques based on a nominal design model developed from the centre of the frequecy response functions from multiple time STFTs give significantly better delay margins.

Important Links:



Go Back