To solve the problem of the narrow high-efficiency operation zone and low overall energy efficiency of a multistage centrifugal pump,an optimal design study of the impeller and volute of a multistage double-suction cen-trifugal pump with a specific speed of 64 was conducted.The applicability of different surrogate models in the opti-mization of the hydraulic performance of the multistage centrifugal pump was investigated and compared,and a GA-BP neural network was selected as the optimal surrogate model.Nine main design parameters were chosen as opti-mization variables,and the efficiency of the pump under partial-load and nominal conditions 0.6Qd and 1.0Qd were set as the optimization objectives.The Pareto-optimal solution of the multiobjective optimization problem was ob-tained using the NSGA-II algorithm with the Latin hypercubic sampling method and an automatic numerical analysis program to construct a sample database,and appropriate combinations of parameters were selected according to ac-tual engineering requirements.The analysis results showed that the efficiency of the model pump was increased by 2.49%and 3.09%under partial-load and nominal conditions,respectively,and the issue of the steep drop in the head under overload conditions was alleviated.This method can be a reference for the positive design of multistage centrifugal pumps.
维普
万方数据