AI-Based Approach for Behind-the-Meter Solar Photovoltaic Power Forecasting


There are hundreds of thousands of behind-the-meter (BTM) solar photovoltaic (PV) installations across the USA which generate a substantial amount of electricity. These installations can provide an untapped reservoir of dispatchable resources that can be counted on to provide reliable grid services. The BTM load prediction and management can play a pivotal role for effective and reliable operation of the grid system. For these reasons, this research focuses on developing an artificial intelligence (AI)-based approach to estimate the PV capacity based on predicted temperature and solar irradiance. The AI algorithm explored in this research is a subtractive clustering-based adaptive neuro-fuzzy inference system (ANFīS) model. Simulation results show the efficacy of the ANFIS method in predicting the BTM PV power. Also, the performance of the ANFIS system is better than that of the linear regression, regression trees, and regression ensembles models.

Publication Title

Proceedings - 2023 IEEE Conference on Artificial Intelligence, CAI 2023