Abstract: In order to solve the problems of insufficient short-term load forecasting accuracy and low matching between energy storage scheduling and load demand in medium voltage distribution networks under the background of distributed new energy grid connection, a short-term load forecasting and energy storage collaborative scheduling optimization model was studied and constructed. An efficient solving algorithm was designed to dynamically optimize the scheduling strategy.The experimental part is based on the case simulation of the improved IEEE33 node system. Compared with the traditional model, the constructed prediction model reduces the average absolute percentage error by 3.2% to 5.8%, effectively reducing the load prediction error, fully utilizing the regulating effect of energy storage, and significantly improving the operational stability and economic operation level of the medium voltage distribution network.