Demand Forecasting and Resource Scheduling of Independent Energy Storage Market in Power Grid with Deep Learning

Abstract

The power grid presents several obstacles for demand forecasting and resource scheduling, such as a substantial amount of data, a growing number of factors influencing the demand profile, uncertainty in the generation profile of distributed energy from renewable sources, and a shortage of historical data. Here, we provide a unique market-oriented energy storage method based on artificial intelligence (AI) that aims to optimize operational profit in the electricity market between consumers, energy storage, and grid service providers. The approach is divided into two parts; the first is the Residual Network-50 (GR-ResNet-50) algorithm, which is used to overcome future uncertainties related to load needs and power pricing. The second uses deep learning based on the Radial Heap Basis Algorithm (RHBA) to determine the most effective charging or discharging operation considering the grid peak states, load demand, and the state of the batteries. The performance of the proposed method in this study is compared with the conventional method in terms of 0.007% of low MAPE and 20 seconds of less execution time. The research shows that the proposed strategy significantly lowers on-peak power, improves operational profit, and improves effective performance.