Machine learning and deep learning models based grid search cross validation for short-term solar irradiance forecasting

In plain language

Photovoltaic power generation is prone to interruptions due to natural fluctuations in solar radiation, making accurate forecasting essential for reliable energy supply and battery storage planning. A comparative evaluation examined standard machine learning and deep learning algorithms for short-term solar irradiance prediction using a five-year dataset of hourly meteorological observations gathered in Islamabad. Hyperparameters across the models were tuned using five-fold grid search cross-validation, with performance assessed across metrics including Adjusted R-squared, normalised root mean square error, and mean absolute error. Among nine deep learning architectures, a hybrid CNN-LSTM model performed best with an Adjusted R-squared score of 0.984. Gradient boosting regression proved superior among six machine learning methods, reaching an Adjusted R-squared score of 0.962. Explainable artificial intelligence frameworks, including SHAP and LIME, were incorporated to interpret model decisions.

Key takeaways

A hybrid CNN-LSTM architecture achieved the highest predictive accuracy among nine deep learning models with an Adjusted R-squared of 0.984.
Gradient boosting regression led machine learning approaches with an Adjusted R-squared score of 0.962 across six evaluated models.
Five-fold grid search cross-validation was successfully used to optimise hyperparameters for both machine learning and deep learning algorithms.
Evaluation utilised five years of hourly meteorological sensor measurements collected between 2015 and 2019 in Islamabad.
Model outcomes were interpreted using explainable artificial intelligence tools, specifically SHAP and LIME.

Why it matters

Solar electricity production can drop sharply when sunlight fluctuates, complicating grid stability and energy storage. Highly accurate short-term irradiance forecasts help operators anticipate power deficits and dispatch stored battery reserves reliably. Demonstrating transparent, high-performing predictive models supports the broader shift toward renewable energy by reducing the operational risks caused by intermittent solar availability.

Commercialisation angle

The algorithms could be incorporated into solar plant monitoring software, grid management systems, and battery storage controllers to balance power loads. Targeted users include photovoltaic operators and energy utilities seeking dependable short-term generation projections. The work is at an applied and tested stage using historical meteorological data, though the abstract does not indicate live operational deployment or integration into commercial software.

AI-generated from the published abstract. Always read the original work before citing.

Abstract

Abstract In late 2023, the United Nations conference on climate change (COP28), which was held in Dubai, encouraged a quick move from fossil fuels to renewable energy. Solar energy is one of the most promising forms of energy that is both sustainable and renewable. Generally, photovoltaic systems transform solar irradiance into electricity. Unfortunately, instability and intermittency in solar radiation can lead to interruptions in electricity production. The accurate forecasting of solar irradiance guarantees sustainable power production even when solar irradiance is not present. Batteries can store solar energy to be used during periods of solar absence. Additionally, deterministic models take into account the specification of technical PV systems and may be not accurate for low solar irradiance. This paper presents a comparative study for the most common Deep Learning (DL) and Machine Learning (ML) algorithms employed for short-term solar irradiance forecasting. The dataset was gathered in Islamabad during a five-year period, from 2015 to 2019, at hourly intervals with accurate meteorological sensors. Furthermore, the Grid Search Cross Validation (GSCV) with five folds is introduced to ML and DL models for optimizing the hyperparameters of these models. Several performance metrics are used to assess the algorithms, such as the Adjusted R 2 score , Normalized Root Mean Square Error (NRMSE), Mean Absolute Deviation (MAD), Mean Absolute Error (MAE) and Mean Square Error (MSE). The statistical analysis shows that CNN-LSTM outperforms its counterparts of nine well-known DL models with Adjusted R 2 score value of 0.984. For ML algorithms, gradient boosting regression is an effective forecasting method with Adjusted R 2 score value of 0.962, beating its rivals of six ML models. Furthermore, SHAP and LIME are examples of explainable Artificial Intelligence (XAI) utilized for understanding the reasons behind the obtained results.

Research topics

Solar Radiation and Photovoltaics
Energy Load and Power Forecasting
Photovoltaic System Optimization Techniques

Discussion

Discuss this research

Have you built on this work, tried to replicate it, or seen it applied in practice? Share what you know. Verified researchers and MARATTO™ domain experts can open a discussion, and any member can reply. Contributions are reviewed before they appear.

No discussion yet. Open the first thread.

New to MARATTO™? Create a free account.