Accurate short-term precipitation forecasts are an important basis for improving the capacities for the forecasting of urban rainstorms, urban floods, and small and medium-sized river floods, and disaster risk prevention and control. In order to improve the accuracy of short-term quantitative precipitation forecasts (QPF), a short-term (0-12.h) QPF model based on the optimization fusion method of multi-model forecasts is developed. The developed model system mainly includes four modules: a single model-based quantile mapping correction module, a real-time optimization module based on model forecasts correction, a real-time fusion module of optimized forecasts and quantile mapping correction module of fused forecasts. By taking the Henan Province as a test region, the developed model is applied and tested for the short-term quantitative precipitation forecasting. The results show that the developed model performs well, and the distribution and intensity of the precipitation forecasts using the new approach are much closer to the observations than the traditional methods; the developed model forecast is able to forecast the precipitation intensity and its cover area of the “7·20” rainstorm in Zhengzhou for 6 hours in advance; the forecasting accuracy of the developed model is much higher than the other traditional models through the intercomparing of rainfall predictions from January to September, 2021, and the forecasting accuracy increases gradually as the forecasting lead-time approaches the rainfall event.