A method of multiobjective optimization on the basis of the NSGA-II algorithm with approximate models for an optimized object is proposed. Artificial neural networks with radial basis functions (RBF networks) are used to construct the approximate models whose parameters are determined with an evolutionary algorithm. The multiobjective optimization of the working process of a gas turbine engine is studied as an example.

Keywords: approximation, response surface model (RSM), RBF networks, multiobjective optimization, gas turbine engine parameters

• PostScript (in Russian) (7.82MB)
• PDF (in Russian) (1.19MB)



• PostScript. zip (in Russian) (2.32MB)
• PDF. zip (in Russian) (1.14MB)