Using the hourly rainfall data in the Data as a Server, based on the North China Numerical Forecasting System (CMA-BJ), the Mesoscale Weather Numerical Forecasting System (CMA-MESO), and the Northeast Numerical Forecasting System (CMA-DB) of the China Meteorological Administration, a multi-model fusion rainfall forecasting method is proposed and tested in Laoning from 2022 to 2023. The results show that without considering the deviation within a radius of 40km, during the Northeast cold vortex rainfall process, the success rate of the fusion forecast in 2022 increased by an average of 13.08%, and the hit rate increased by an average of 0.57%. In 2023,these numbers are 25.7% and 17.5%.During the subtropical high-pressure rainfall process, the success rate of the fusion forecast in 2022 increased by an average of 16.8%, and the hit rate increased by an average of 1%.In 2023,the hit rate increased 289%. During the typhoon storm process, the success rate of the fusion forecast in 2022 increased by an average of 34.21%, and the hit rate increased by an average of 14.9%.In 2023, these numbers are 30.7% and 651%.When short-term heavy rainfall is caused by the influence of systems such as upper troughs, the success rate of the fusion forecast in 2022 increased by an average of 32.2%, and the hit rate increased by an average of 4.07%. Under the influence of systems such as the Northeast cold vortex and upper troughs, the success rate of the fusion forecast shows positive growth, while under the background of subtropical high pressure and typhoons, the fusion forecast is a negative skill compared to the original numerical model.