A multi-disciplinary optimization software Isight and CFD simulation software NUMECA were used to integrate the optimization strategy, numerical experimental design, approximation model and global optimization algorithm into one design platform. The user interface was created by Python. The automatic optimization design platform suitable for a hydraulic model of nuclear reactor coolant pump has been established. Based on the model pump obtained by our research group in preliminary studies, the automatic optimization design platform was used to establish the approximation model, and it can reflect the relationship between design variables and the objective function. Then the impeller and impeller-guide vane interaction was studied. Multi-island genetic algorithm was used. A pump model with improved hydraulic performance was obtained. The results show that the initially built integrated optimization platform is of applicability. The feasibility of the optimization strategy has also been verified.