Data Analytics–Based Prediction of Solar Photovoltaic Power Generation and System Performance
Year: 2026
Type: conference-paper
Abstract: Accurate prediction of solar photovoltaic (PV) power generation is essential for reliable grid operation, energy management, and performance monitoring in power systems with high utilizes a battery-assisted charging architecture, in which a single solar PV panel charges an intermediate battery bank viand system performance assessment using real operational data obtained from a grid-connected 960 kWp PV plant operating under Mediterranean climatic conditions. The proposed methodology integrates systematic data pre processing, multivariate feature engineering, and supervised learning to model the nonlinear relationship between meteorological parameters and PV output power. Hourly ambient temperature, module temperature, plane-of-aapproach to sizing battery banks tilt angles, and historical PV power are utilized as input features for forecasting. A mathematical model 1 h to 24 h ahead. The developed models are evaluated using Mean Absolute Error, Root Mean Square Error, and Mean Absolute Percentage Error. Results demonstrate that the predicted PV power closely follows the measured generation profile, with low forecasting errors for short-term horizons and stable performance degradation for longer horizons. In addition to accurate power prediction, the analytics framework provides valuable insights into PV system behavior by capturing thermal and irradiance-dependent performance characteristics. The proposed approach is scalable, data-efficient, and suitable for integration into real-time monitoring and decision-support systems for solar PV plants.
Institutions: DES Pune University, Savitribai Phule Pune University
Language: English
Field: Energy
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7204884031