Artificial Intelligence-Based Prediction of Flexibility Requirements in Power Systems

In the evolving landscape of power systems, the integration of renewable energy sources introduces a significant layer of variability and uncertainty demanding strategic finesse to address flexibility requirements of the power system. This paper proposes an innovative artificial intelligence (AI)-based approach to account for the flexibility requirements of the power systems in order to reflect the variability and uncertainty of renewable energy sources and demands. One of the main aspects of the proposed methodology is the precise prediction of power system net load which is a dynamic quantity representing the real-time difference between system load and renewable generations. To achieve highly accurate net load forecasts, an AI-driven method is employed by utilizing a deep learning model in the form of a convolutional neural network-long short-term memory (CNN-LSTM) hybrid. To address the inherent uncertainty in predicted net load, a quantile regression model is adopted which provides the net load range at a certain confidence level. This dual-pronged methodology combines precise prediction and uncertainty quantification to precisely characterize the ramp-up and ramp-down flexibility requirements of the power system. A real-data case study as well as a comparative case study are investigated to demonstrate the model effectiveness.