Оптимизация проточной части осевого насоса с использованием прямых методов

Abstract

The direct methods were developed for optimizing the OP-5 axial pump flow section based on the DesignXplorer tool algorithms of the ANSYS Workbench software package. At the first stage, the MOGA genetic algorithm and the ASO gradient method were used. To increase the optimization efficiency, the input parameters were correlated. Accuracy of the relationship between the input and output parameters was evaluated by the determination coefficients values and the scatter diagram. Such studies made it possible to reduce the number of input parameters from nineteen to seven. It is shown that the best approach for numerical optimization of the axial pump flow section is the consistent use of stochastic and local methods.