基于多目标遗传算法的屏蔽泵叶轮水力优化

In order to improve the comprehensive hydraulic characteristic of the cannedmotor pump, a joint optimization platform of ANSYS-Workbench and iSIGHT was built. The sensitivity analysis of 28 alternative parameters, such as the blade bone line, hub and shroud of impeller and blade thickness et al were carried out by optimized Latin cube design. Based on the influence degree of various parameters on the objective function, 9 parameters were determined at the blade bone line and the shroud of impeller as the final input parameters. Kriging surrogate model and NSGA-? were selected to iteratively optimize the efficiency and head. Finally, two blades optimization schemes were obtained according to different weight distribution. Through the verification of numerical simulation, the efficiency of Scheme 1 and Scheme 2 under rated working condition was increased by 1 .98% and 2.83%, respectively, and the head was increased by 15. 73 m and 13. 39 m, respectively. The hydraulic characteristics in the operation range were significantly improved. The results show that the hub parameter z3 has the greatest influence on the efficiency and head, reaching -18.99% and -30.10%, respectively. Kriging surrogate model has the highest prediction accuracy, with the total error of 3. 393%. In the operating range of 0.83Q BEP - 1 . 12Q BEP, the head of Scheme 1 and Scheme 2 is significantly higher than that of the original scheme, and Scheme 1 has the most significant optimization efficiency under the optimal flow condition, reaching 13.89%.