Machine Learning Based Maximum Power Prediction for Photovoltaic System

Indian Journal of Pure & Applied Physics

article

Anshul Agarwala¹  Nitish Kumar¹  Pawan Dubey² 

[1] National Institute of Technology;Madhav Institute of Technology & Science

关键词: Supervised machine learning;    Data driven modeling;    Boost converter;    MPPT (Maximum power point tracking);    Article;   

DOI  :  10.56042/ijpap.v60i10.62197

来源: National Institute of Science Communication and Information Resources

【 摘 要 】

This manuscript proposes a data-driven machine learning algorithm to track maximum power for PV (photovoltaic) panel systems. Data from the PV panel system connected to a boost converter has been collected. PV Voltage, current, temperature, irradiance, PI and power value have been collected for the supervised machine learning-based modeling. Where PV Voltage, PV current, temperature, and irradiance are the predictors, and PI (proportional integral) is the response of the machine learning-based model. The proposed system becomes more efficient with time while existing MPPT (maximum power point tracking) work on a specific logic for whole life. The model efficacy has been analyzed based on accuracy, scattering plot, and ROC (receiver operating characteristics) curve.

【 授权许可】

Unknown