Study on Improved Multidiscipline Feasible Strategy for Complicated Turbine Component Optimization

Abstract

A new improved multidiscipline feasible (MDF) strategy based on the updating approximate function constructed by Response Surface Method (RSM) is firstly pointed out and denoted as MDF-update-RSM which involves double cycles. Before the start of inner loop on MDF-update-RSM strategy, the approximate function is initialized by RSM on a set of sampling points which are generated from the Design of Experiment (DOE). With the process of outer loop, the coefficients and one high-order term of polynomial are updated and upgrade by the Least Square Regression. The ability of this new strategy promotes the efficiency and saves huge calculational cost incurred by the MDF strategy which always runs on the multidisciplinary analysis (MDA). The MDF-update-RSM is good procedure to ameliorate the accuracy that has been tested by the Rosenbrock function in this paper. Finally, the optimization strategy is employed on executing the low pressure turbine shrouded blade design and provides a true optimal result compared to the classical MDF strategy.