Day-ahead profit forecasting of microgrid using LSTM algorithm

Monitoring and analyzing power generation and consumer behavior in the power market will be quite difficult. In the context of a competitive market purchasing and selling structure, it is more difficult to forecast profit under different uncertainties. Microgrid planning and operation are going to be complicated. In this manuscript, we proposed a day ahead profit (DAP) forecasting model for micro grid consisting of a wind energy system (WES), battery energy storage system (BESS), and photovoltaic system (PV) participating as an aggregator in the pool-co market under uncertainties such as step load variations from +50 % to - 50 % and line contingencies. Long Short-Term Memory (LSTM) model has been used to obtain DAP due to its high accuracy for the power market. The proposed non-integer linear programming model and DAP forecasting model is solved by modifying Matpower files in Matlab and Python in Google colab respectively. The effectiveness of the proposed methodology has been verified by using a modified IEEE 33 bus system.