Gas Turbine Control Using Mixed Sensitivity H∞-Optimisation

The purpose of this paper is to present an industrial application of the polynomial approach to multivariable mixed sensitivity H∞ optimisation for feedback systems introduced by Kwakernaak [8]. The software implemented was supplied by the Dept. of Applied Mathematics, Twente University [7] based on the design algorithm of [2].The multivariable process model was derived from non-linear simulation data for a 1.5 MW gas turbine supplied by Hawker Siddeley Dynamics [6] as part of a collaborative research project at the Industrial Control Unit,University of Strathclyde.A previous scalar design study relating to the gas turbine problem has been presented by Biss [1].The design results for the MISO gas turbine problem with respect to the controller will demonstrate the robustness properties which can be produced by judicious choice of weighting function matrices V,W1 and W2 within a criterion to be minimized,with respect to the stabilising compensator transfer matrix G, of the form ¦¦ Z(s) ¦¦ ∞ where $${\text{Z}}\;:\; = \;{{\text{V}}^ * }\;({{\text{S}}^ * }{{\text{W}}_1}^ * {{\text{W}}_1}{\text{S}}\;{\text{ + }}\;{{\text{T}}^ * }{{\text{W}}_2}^ * {{\text{W}}_2}{\text{T}}){\text{V}}$$ and S : = (I+HG)-1 and T : = GS