Artificial Intelligence and ChatGPT in Optimal Management of Battery Energy Storage Systems: Advantages and Challenges

This is the authors' abstract. We don't add key points for this paper.

In the presence of renewable resources, such as solar and wind, which are inherently intermittent, battery energy storage systems (BESS) are among the most popular energy storage solutions for increasing grid reliability and stability. Despite all the advantages, energy storage operations can be challenging, from monitoring the output of renewable energy sources to scheduling charging and discharging. The primary aim of this study is to predict renewable energy generation and energy consumption with the highest possible accuracy using artificial intelligence (AI) techniques for the optimal management of BESS. Specifically, the long-term short-term memory model (LSTM) is used. Then, OpenAI ChatGPT will be applied to support the prediction results and determine the optimal charging and discharging strategies. The advantages and challenges of using ChatGPT and other AI prediction techniques are also discussed. The results showed that among the different generative AI, GPT-3.5 achieved a mean absolute error (MAE) of 3.225, mean squared error (MSE) of 46.196, and root mean squared error (RMSE) of 6.797, demonstrating significant improvements over the basic hyperparametric tuning of the LSTM technique.