Explainable Machine Learning for Battery Side Power Estimation from Archived Photovoltaic Data

Giovanni Dimas Prenata; Ahmad  Ridho’i

doi:10.31258/ijeepse.9.3.165-183

Giovanni Dimas Prenata Universitas 17 Agustus 1945 Surabaya, Surabaya, Indonesia.
Ahmad Ridho’i Universitas 17 Agustus 1945 Surabaya, Surabaya, Indonesia.

DOI: https://doi.org/10.31258/ijeepse.9.3.165-183

Abstract viewed: 0 times

pdf downloaded: 0 times

Keywords: Battery side power, Explainable artificial intelligence, Incremental Conductance MPPT, Photovoltaic, Random Forest.

Abstract

This retrospective exploratory study applies explainable machine learning to an archived photovoltaic dataset containing 81 observations from three nominally 100 Wp modules over five measurement days. Battery side power was calculated as Pbat=Vbat×Ibat, and Vbat and Ibat were excluded from the predictors. The original measurement sequence, instrument specifications, and calibration records were unavailable; therefore, the variables labelled Voc and Isc are treated as archived electrical descriptors rather than independently verified simultaneous open-circuit and short-circuit measurements. A five-feature base model used irradiance, temperature, time, Voc, and Isc. An augmented model added Vmp,est and Imp,est, which were deterministic transformations using module-specific datasheet ratios. Under nested leave-one-day-out validation, the augmented Random Forest achieved RMSE=6.9779 W, MAE=5.4078 W, and R²=0.8823, compared with RMSE=7.2847 W for the base model. This 0.3068 W difference is reported descriptively because measurement uncertainty is unavailable. Leave-one-module-out RMSE ranged from 6.4670 to 10.1599 W. Attribution results indicate reliance on irradiance and current-related feature families. The findings quantify predictive patterns within this archived dataset and do not validate the original measurement system or dynamic MPPT performance.

References

W. De Soto, S. A. Klein, and W. A. Beckman, “Improvement and validation of a model for photovoltaic array performance,” Solar Energy, vol. 80, no. 1, pp. 78–88, 2006, doi: 10.1016/j.solener.2005.06.010.

T. Huld and A. M. Gracia Amillo, “Estimating PV module performance over large geographical regions: The role of irradiance, air temperature, wind speed and solar spectrum,” Energies, vol. 8, no. 6, pp. 5159–5181, 2015, doi: 10.3390/en8065159.

S. Dubey, J. N. Sarvaiya, and B. Seshadri, “Temperature dependent photovoltaic (PV) efficiency and its effect on PV production in the world—A review,” Energy Procedia, vol. 33, pp. 311–321, 2013, doi: 10.1016/j.egypro.2013.05.072.

L. Breiman, “Random forests,” Machine Learning, vol. 45, pp. 5–32, 2001, doi: 10.1023/A:1010933404324.

A. E. Hoerl and R. W. Kennard, “Ridge regression: Biased estimation for nonorthogonal problems,” Technometrics, vol. 12, no. 1, pp. 55–67, 1970, doi: 10.1080/00401706.1970.10488634.

J. H. Friedman, “Greedy function approximation: A gradient boosting machine,” The Annals of Statistics, vol. 29, no. 5, pp. 1189–1232, 2001, doi: 10.1214/aos/1013203451.

M. Li, H. Sun, Y. Huang, and H. Chen, “Shapley value: from cooperative game to explainable artificial intelligence,” Autonomous Intelligent Systems, vol. 4, no. 1, Art. no. 2, 2024, doi: 10.1007/s43684-023-00060-8.

S. M. Lundberg and S.-I. Lee, “A unified approach to interpreting model predictions,” in Advances in Neural Information Processing Systems 30, pp. 4768–4777, 2017.

S. Varma and R. Simon, “Bias in error estimation when using cross-validation for model selection,” BMC Bioinformatics, vol. 7, Art. no. 91, 2006, doi: 10.1186/1471-2105-7-91.

M. Kuzlu, U. Cali, V. Sharma, and Ö. Güler, “Gaining insight into solar photovoltaic power generation forecasting utilizing explainable artificial intelligence tools,” IEEE Access, vol. 8, pp. 187814–187823, 2020, doi: 10.1109/ACCESS.2020.3031477.

E. Gomes, A. Esteves, H. Morais, and L. Pereira, “Leveraging Explainable Artificial Intelligence in Solar Photovoltaic Mappings: Model Explanations and Feature Selection,” Energies, vol. 18, no. 5, Art. no. 1282, 2025, doi: 10.3390/en18051282.

R. M. Rizk-Allah, L. M. Abouelmagd, A. Darwish, V. Snasel, and A. E. Hassanien, “Explainable AI and optimized solar power generation forecasting model based on environmental conditions,” PLoS ONE, vol. 19, no. 10, Art. no. e0308002, 2024, doi: 10.1371/journal.pone.0308002.

O. Petrosian and Y. Zhang, “Solar power generation forecasting in smart cities and explanation based on explainable AI,” Smart Cities, vol. 7, no. 6, pp. 3388–3411, 2024, doi: 10.3390/smartcities7060132.

S. Nematzadeh and V. Esen, “Explainable machine learning and predictive statistics for sustainable photovoltaic power prediction on areal meteorological variables,” Applied Sciences, vol. 15, no. 14, Art. no. 8005, 2025, doi: 10.3390/app15148005.

C. Strobl, A.-L. Boulesteix, T. Kneib, T. Augustin, and A. Zeileis, “Conditional variable importance for random forests,” BMC Bioinformatics, vol. 9, Art. no. 307, 2008, doi: 10.1186/1471-2105-9-307.

C. Sun, J. Ling, and J. Wang, “Research on a novel and improved incremental conductance method,” Scientific Reports, vol. 12, Art. no. 15700, 2022, doi: 10.1038/s41598-022-20133-7.

I. Owusu-Nyarko, M. A. Elgenedy, I. Abdelsalam, and K. H. Ahmed, “Modified Variable Step-Size Incremental Conductance MPPT Technique for Photovoltaic Systems,” Electronics, vol. 10, no. 19, Art. no. 2331, 2021, doi: 10.3390/electronics10192331.

M. Derbeli, C. Napole, O. Barambones, J. Sanchez, I. Calvo, and P. Fernández-Bustamante, “Maximum Power Point Tracking Techniques for Photovoltaic Panel: A Review and Experimental Applications,” Energies, vol. 14, no. 22, Art. no. 7806, 2021, doi: 10.3390/en14227806.

S. André, F. Silva, S. Pinto, and P. Miguens, “Novel Incremental Conductance Feedback Method with Integral Compensator for Maximum Power Point Tracking: A Comparison Using Hardware in the Loop,” Applied Sciences, vol. 13, no. 7, Art. no. 4082, 2023, doi: 10.3390/app13074082.