In order to study the flow field and external characteristics changes within a centrifugal pump during the startup process,a secondary development interface provided by CFD software was utilized.By developing cel functions,the accurate simulation of flow rate and speed variations could be achieved,enabling the simulation of the entire change process during startup.The numerical simulation results demonstrate that there are pulsation phenomena in both the head and efficiency during the initial stage of startup.To address this pulsation phenomenon,changes are made to the structural parameters of the centrifugal pump,such as blade wrap angle and outlet width.Additionally,the simulation results for different structural parameters are used to construct a neural network,which is optimized using a genetic algorithm to determine the optimal solution for reducing the pulsation of head and efficiency in the early stages of pump operation.Ultimately,it is observed that the head pulsation decreases by 3.21 m,while the efficiency pulsation is reduced by 2.87%.