Robust Multi-Objective Optimization for Gas Turbine Operation Based on Kriging Surrogate Model

This paper presents a systematic method for the optimal settings of gas turbine operation. After building the high-fidelity model of the gas turbine, the fuel flow, variable inlet guide-vane(VIGV) of the high and low pressure compressors, ambient air pressure and temperature are selected as the decision variables, while the pollutant emissions and output power of whole system are chosen as response variables to be optimized. In order to guarantee the accurate response prediction, the performance of three different types of surrogate models, namely polynomial response surface (PRS), Kriging and radial basis function (RBF) models are built with sampling points generated by Latin hypercube design (LHD) method. With Kriging surrogate model best fitting the data, it is chosen as the model for performance prediction during the optimization process. It is also shown that the NSGA-II algorithm is suitable for the multi-objective optimization of gas turbine. The robustness of the Pareto solutions are checked under the varying ambient pressure and temperature. By considering the solution robustness, and the nature of gas turbine performance change under perturbation, the optimal operating conditions are identified.