The number of impeller and diffuser blades has a great influence on the head and efficiency of the pump. The 250QJ140 type well submersible centrifugal pump was selected as the research object. By the combination of numerical calculation and experiment, under the variation of blade number of impeller and diffuser, the performance variation and internal flow field distribution of submersible pump were analyzed. Based on the principle of not changing other geometric parameters, 16 submersible pump models with two different combinations of blades were established. ANSYS ICEM software was used to structure each group of models separately, and then the ANSYS CFX commercial software was used to calculate the multi-working constant of each model. For each set of numerical calculation, standard k-ω turbulence model and standard wall function were selected to obtain the performance prediction values of each model under different operating conditions. Compared the predictions of the performance of each group, it can be found that the efficiency of submersible pump model was the highest when the number of impeller and guide vane was 7 at the rated flow rate. Under the low flow conditions and high flow conditions, the pump medium flow angle was changed. Under low flow conditions, increasing the number of blades of the impeller and guide vanes can improve blade rectification of the liquid medium, thereby improving the performance of submersible pumps used in wells. In the case of large flow rates, fewer impellers and guide vane blades were more capable of reducing the effect of the blade on the liquid media crowding. Prototype manufacturing and performance test were carried out under the condition of good performance in large flow conditions. The test results showed that the model performance was better. Under the rated flow conditions, the predicted values of head and shaft power were 2.4% and 1.6% lower than those of the test result, and the predicted efficiency was 1.1% higher than that of the test results. The overall changing trend of numerical prediction results and experimental results with the flow was consistent, which confirmed the numerical accuracy of the research.