Solar Photovoltaic Power Potential and Generation Forecasting in Nigeria using Machine Learning
Fuente:
Zenodo
Gespeichert in:
| 1. Verfasser: | |
|---|---|
| Format: | Recurso digital |
| Veröffentlicht: |
Zenodo
2026
|
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866902270300389376 |
|---|---|
| author | IJMSRT |
| author_facet | IJMSRT |
| contents | <p>Abstract <br>Nigeria possesses significant solar energy <br>potential, yet uncertainty in generation <br>forecasts limits large-scale adoption. This <br>study applies ML models to forecast solar PV <br>output using satellite- derived meteorological <br>data. <br>The increasing trend of using solar <br>photovoltaic (PV) energy as an alternative to <br>fossil fuels has increased the demand for <br>precise forecasting tools, particularly in <br>developing nations such as Nigeria, where grid <br>instability and load mismatch are prevalent. <br>This paper examines the capability of <br>Artificial Neural Network (ANN) models for <br>forecasting solar PV energy in grid-connected <br>systems in the Nigerian energy sector. Based <br>on the historical meteorological and load <br>demand data from 2020 to 2025, an ANN <br>model was designed, trained, and simulated <br>using MATLAB R2022a software. The model <br>included essential parameters such as solar <br>irradiance, temperature, and time variables to <br>forecast solar power generation. <br>The simulated values were validated against <br>the actual output to determine the accuracy of <br>the model using parameters such as Mean <br>Absolute Percentage Error (MAPE), Root <br>Mean Square Error (RMSE), and the <br>coefficient of determination (R²). The ANN <br>model yielded a MAPE of 6.83%, an RMSE <br>of 12.47 kW, and an R² of 0.95, indicating <br>excellent forecasting accuracy and <br>adaptability to the non-linear solar output <br>variations. In addition, the study presents <br>graphical results, such as predicted vs. actual <br>output graphs, error distribution histograms, <br>and regression plots, which verify the <br>robustness of the model. <br>These findings affirm the viability of using <br>ANN for the forecasting of solar PV and <br>underscore its promise to improve the energy <br>planning, stability, and dispatch of energy in <br>Nigeria. Moreover, the findings of this study <br>encourage the adoption of AI-based <br>forecasting tools to improve the </p> <p>energy management of Nigeria to maximize <br>the benefits of renewable energy. This paper <br>contributes to the existing knowledge on <br>intelligent forecasting for smart grid <br>applications and presents a model that can <br>be replicated in other developing countries. </p> |
| format | Recurso digital |
| id | zenodo_https___doi_org_10_5281_zenodo_19096423 |
| institution | Zenodo |
| language | |
| publishDate | 2026 |
| publisher | Zenodo |
| record_format | zenodo |
| spellingShingle | Solar Photovoltaic Power Potential and Generation Forecasting in Nigeria using Machine Learning IJMSRT <p>Abstract <br>Nigeria possesses significant solar energy <br>potential, yet uncertainty in generation <br>forecasts limits large-scale adoption. This <br>study applies ML models to forecast solar PV <br>output using satellite- derived meteorological <br>data. <br>The increasing trend of using solar <br>photovoltaic (PV) energy as an alternative to <br>fossil fuels has increased the demand for <br>precise forecasting tools, particularly in <br>developing nations such as Nigeria, where grid <br>instability and load mismatch are prevalent. <br>This paper examines the capability of <br>Artificial Neural Network (ANN) models for <br>forecasting solar PV energy in grid-connected <br>systems in the Nigerian energy sector. Based <br>on the historical meteorological and load <br>demand data from 2020 to 2025, an ANN <br>model was designed, trained, and simulated <br>using MATLAB R2022a software. The model <br>included essential parameters such as solar <br>irradiance, temperature, and time variables to <br>forecast solar power generation. <br>The simulated values were validated against <br>the actual output to determine the accuracy of <br>the model using parameters such as Mean <br>Absolute Percentage Error (MAPE), Root <br>Mean Square Error (RMSE), and the <br>coefficient of determination (R²). The ANN <br>model yielded a MAPE of 6.83%, an RMSE <br>of 12.47 kW, and an R² of 0.95, indicating <br>excellent forecasting accuracy and <br>adaptability to the non-linear solar output <br>variations. In addition, the study presents <br>graphical results, such as predicted vs. actual <br>output graphs, error distribution histograms, <br>and regression plots, which verify the <br>robustness of the model. <br>These findings affirm the viability of using <br>ANN for the forecasting of solar PV and <br>underscore its promise to improve the energy <br>planning, stability, and dispatch of energy in <br>Nigeria. Moreover, the findings of this study <br>encourage the adoption of AI-based <br>forecasting tools to improve the </p> <p>energy management of Nigeria to maximize <br>the benefits of renewable energy. This paper <br>contributes to the existing knowledge on <br>intelligent forecasting for smart grid <br>applications and presents a model that can <br>be replicated in other developing countries. </p> |
| title | Solar Photovoltaic Power Potential and Generation Forecasting in Nigeria using Machine Learning |
| url | https://doi.org/10.5281/zenodo.19096423 |