Day-ahead Energy Sharing Schedule for the P2P Prosumer Community Using LSTM and Swarm Intelligence

Prosumer community forms by prosumer who is not only consuming energy but also generating renewable energy (e.g., solar) and capable of selling surplus energy to other consumers. Peer-to-peer (P2P) energy sharing behavior of the smart grid is evolving to reducing the usage of non-renewable energy. However, non-renewable energy is still used in some time intervals due to the unbalance between energy load and generation. Therefore, in this paper, we study an energy scheduling problem that includes the energy amount for battery charge/discharge along with energy sharing scheduling among the prosumer community. First, we formulate an optimization problem and the objective is to minimize the non-renewable energy usage of the entire community. This problem includes the day-ahead energy demand prediction stage and battery charge/discharge, and energy sharing scheduling stage. Second, to solve the formulated problem, a long-short-term memory (LSTM) and particle swarm optimization (PSO) joint approach is proposed, in which the LSTM based model is used to forecast day-ahead energy demand, while PSO is utilized in the second scheduling stage by considering P2P behavior. Finally, the evaluation result shows our proposed LSTM prediction model outperforms the autoregressive integrated moving average (ARIMA) model by comparing the mean squared error, root-mean-square error and total training time. PSO improves the overall usage of non-renewable energy.