Study on Medium-Term Optimal Dispatching Method of Power System with High Proportion of New Energy Based on Load Forecasting

With the rapid development of various new energy sources, the load change of the power system becomes more complex. In view of the complexity of load changes, this paper proposes a medium-term optimal scheduling method based on load forecasting for a high percentage of new energy power systems. First, a load prediction method based on deep learning is proposed. Then, a recurrent scheduling model based on load forecasting is constructed, combining the time correlation among thermal power, wind power and photovoltaic units to establish a recurrent scheduling model with internal and external double cycles. Finally, the accuracy of the load forecast was verified by comparing the load profiles, and the loop scheduling model was analyzed using the improved IEEE 24-node system. The results show that the predicted loads obtained using the deep learning model can be well used for optimal scheduling, and the scheduling results of the system are more in line with the actual changes.