Open research questions in Solar Radiation and Photovoltaics

What the literature leaves open

• Limited evidence exists on whether CNN-LSTM hybrids offer tangible benefits over simpler GRUs for short-term horizons. There is a need for accurate forecasting of solar power output to maintain grid stability.

  Comparative Analysis of Deep Learning Architectures for Short-Term Solar Power Forecasting · 2026 · DOI

• Current XAI applications are primarily related to solar and wind energy, leaving other renewable sources, such as biomass, hydropower, and geothermal energy, underexplored. However, the integration of XAI remains in its early stages, facing significant challenges such as the lack of standardized evaluation protocols and limited involvement of end-users in assessing the usability of explanations.

  Renewable energy management using explainable artificial intelligence · 2026 · DOI

• The study highlights the need for innovative machine learning approaches for short-term PV power forecasting.

  Accurate Short-Term Photovoltaic (PV) Power Forecasting: Leveraging the Capabilities of AdaBoost and Decision Trees · 2026 · DOI

• Further studies can investigate the impact of air pollution parameters on solar radiation forecasting in other regions. The development of more advanced machine learning algorithms can improve forecasting accuracy.

  Air pollution and meteorological factors significantly enhance solar radiation forecasting accuracy in the western mediterranean region: a machine learning approach · 2026 · DOI

• The gap is addressed by developing a hybrid model that combines statistical and machine learning approaches.

  Day-Ahead Hourly Forecasting of Solar Radiation using a Physics-based Hybrid Machine Learning Model in Some Selected Locations in Nigeria · 2026 · DOI

• Competing land uses and limited ground-based solar data constrain evidence-based decision-making in Gauteng Province, South Africa. The lack of spatially explicit, transparent, and sustainable approaches for identifying suitable renewable energy sites is a significant gap.

  Explainable geo-informatics for spatial solar suitability analysis in Gauteng Province, South Africa · 2026 · DOI

• Solar cell Modelling The Single Diode Model (SDM) is used to evaluate the performance of photovoltaic (PV) modules under various conditions, as manufacturer data alone is insufficient.

  Adaptive Hybrid Evolutionary Algorithms for High-Accuracy Solar Cell Modelling Under Dynamic Environmental Conditions · 2026 · DOI

• The lack of a comprehensive framework that integrates weather-based clustering, signal decomposition, and hybrid neural networks for PV power forecasting.

  Short-Term PV Power Generation Forecasting Based on Clustering CPO-VMD and Transformer Ensemble Neural Networks · 2026 · DOI

• The system does not integrate deep learning architectures. The system is limited to six Indian cities.

  S.U.R.A.J. (Solar Utility & Radiance Analytical Judgment): Localized ML-Based Forecasting for Diverse Indian Climates · 2026 · DOI

• Further research can be conducted to improve the performance of the U-Net model. The application of deep learning models can be explored for other renewable energy sources.

  Evaluation of U-Net3+ and Attention U-Net for Solar Energy Estimation on Urban Rooftops using LiDAR DSMs · 2026 · DOI

• The study addresses the limitation of traditional siting methodologies in accounting for heterogeneous spatial data.

  Digital Visualization and Siting Optimization of Power Plants Using High-Resolution Meteorological Data and GIS · 2026 · DOI

• There is a need for a forecast-driven economic evaluation framework for grid-export-only PV plants with BESS deployment.

  COST ASSESSMENT FOR GRID-EXPORT PV SYSTEMS USING LONG-TERM FORECASTING: A PRE-INVESTMENT STUDY FOR BESS DEPLOYMENT · 2026 · DOI

• The current literature lacks a unified framework for comparing RNN and LSTM based architectures in PV forecasting via hyperparameter optimization. The canonical Grey Wolf Optimizer (GWO) has limitations, such as reduced population diversity and an increased risk of premature convergence, which can be addressed by introducing randomness into the optimization process.

  Stochastic Grey Wolf Optimization for Hyperparameter Tuning of LSTM and RNN Models in Energy Forecasting · 2026 · DOI

• Traditional statistical methods have limitations in modeling solar radiation. The study aims to improve the forecasts accuracy using a Takagi-Sugeno fuzzy algorithm.

  A TAKAGI-SUGENO fuzzy algorithm for solar irradiation forecasting · 2026 · DOI

• The existing forecasting methods have limitations and do not always provide sufficient accuracy. The use of Big Data technologies can help improve forecasting accuracy and optimize solar energy production.

  Big data technologies in forecasting solar energy generation · 2026 · DOI

• The study identifies the need for accurate performance prediction in solar energy systems. There is a gap in evaluating the performance of different algorithms for solar energy forecasting.

  Comparing The Performance of Algorithms Used for Data Analysis in Solar Energy Systems · 2026 · DOI

• Clear-sky models cannot respond to cloud-driven irradiance fluctuations. Purely data-driven models may lack physical consistency and generalizability.

  AI-Augmented Clear-Sky Algorithms for High-Granularity Solar Irradiance Modelling · 2026 · DOI

• Future research will focus on validating the proposed framework using multi-site and multi-climate datasets, exploring hybrid deep learning architectures, incorporating advanced uncertainty quantification techniques, and integrating forecasting models into real-time predictive control and energy management systems for PV/T applications.

  Comparative Assessment of Temporal Deep Learning Architectures for Photovoltaic–Thermal System Thermal Efficiency Forecasting with Sequence Length Sensitivity Analysis · 2026 · DOI

• The lack of accurate solar radiation prediction is a limitation in the development of solar energy systems. Previous studies have used static hyperparameter optimization approaches, which is a limitation.

  A Multiple Optimizer Based Long Short-Term Memory Models for the Efficient Prediction of Solar Radiation · 2026 · DOI

• Special attention should be paid to hybrid systems which can combine forecasting, energy optimization, uncertainty management and intelligent control and can be scaled for successful deployment in both utility-scale and distributed renewable energy systems.

  A review of machine learning and deep learning models, features, and applications for solar PV forecasting · 2026 · DOI

• One of the limitations of this study is the limited dataset that may partially account for the limited performance of the TCN-EMLP and the transformer.

  Deep learning-based time-series solar power prediction for automatic multisource energy generation systems · 2026 · DOI

• Limited ground-based solar data is available. The deployment of photovoltaic infrastructure is fundamentally a spatial problem.

  Explainable geo-informatics for spatial solar suitability analysis in Gauteng Province, South Africa · 2026 · DOI

• Inaccurate estimation of future PV generation can lead to biased revenue projections, misleading payback calculations, and elevated investment risk. The majority of existing studies evaluate forecasting performance primarily through statistical error metrics.

  COST ASSESSMENT FOR GRID-EXPORT PV SYSTEMS USING LONG-TERM FORECASTING: A PRE-INVESTMENT STUDY FOR BESS DEPLOYMENT · 2026 · DOI

• Integration of real-time meteorological API feeds. Exploration of deep learning architectures such as LSTM and Temporal Fusion Transformer models.

  S.U.R.A.J. (Solar Utility & Radiance Analytical Judgment): Localized ML-Based Forecasting for Diverse Indian Climates · 2026 · DOI

• Evaluating the performance of the proposed SGWO-based system on different PV data sets. Comparing the performance of the proposed SGWO-based system with other optimization algorithms. Evaluating the robustness of the proposed SGWO-based system to different hyperparameter settings.

  Stochastic Grey Wolf Optimization for Hyperparameter Tuning of LSTM and RNN Models in Energy Forecasting · 2026 · DOI

Most-cited papers in Solar Radiation and Photovoltaics

• Prediction of solar energy guided by pearson correlation using machine learning · Energy · 2021 · 509 citations
• A global inventory of photovoltaic solar energy generating units · Nature · 2021 · 472 citations
• Comparison of machine learning methods for photovoltaic power forecasting based on numerical weather prediction · Renewable and Sustainable Energy Reviews · 2022 · 460 citations
• CNN-LSTM: An efficient hybrid deep learning architecture for predicting short-term photovoltaic power production · Electric Power Systems Research · 2022 · 406 citations
• Deep learning models for solar irradiance forecasting: A comprehensive review · Journal of Cleaner Production · 2021 · 405 citations
• Deep learning neural networks for short-term photovoltaic power forecasting · Renewable Energy · 2021 · 391 citations
• Accurate one step and multistep forecasting of very short-term PV power using LSTM-TCN model · Renewable Energy · 2023 · 360 citations
• Improved solar photovoltaic energy generation forecast using deep learning-based ensemble stacking approach · Energy · 2021 · 349 citations
• A review of solar forecasting, its dependence on atmospheric sciences and implications for grid integration: Towards carbon neutrality · Renewable and Sustainable Energy Reviews · 2022 · 311 citations
• Day-ahead hourly photovoltaic power forecasting using attention-based CNN-LSTM neural network embedded with multiple relevant and target variables prediction pattern · Energy · 2021 · 304 citations

Most recent work

• Patch Time Series Transformer−Based Short−Term Photovoltaic Power Prediction Enhanced by Artificial Fish · Energies · 2026
• Explainable and physics-constrained PV power prediction via a hybrid framework Integrating secondary decomposition and improved Transformer-LSTM · Energy · 2026
• Spatio-temporal photovolatic power forecasting via Bayesian-optimized dynamic graph convolutional networks with temporal convolutional networks · Journal of Cleaner Production · 2026
• A comprehensive review of deep learning for solar nowcasting: Enhancing accuracy, reliability, and interpretability · Applied Energy · 2026
• Deep learning prediction models for short-term solar photovoltaic power generation forecasting · Next Energy · 2026
• Mapping Europe’s rooftop photovoltaic potential with a building-level database · Nature Energy · 2026
• Deep hybrid CNN–biLSTM model for accurate solar photovoltaic power forecasting: A comparative study with classical and neural models · Energy Reports · 2026
• Pre-Trained Large Model as Distributed PV Power Forecaster: Integrating Scenario-Consistent Retrieval-Augmented Generation and Cross-Modal Semantic Alignment · IEEE Transactions on Sustainable Energy · 2026
• Short-Term photovoltaic power forecasting based on K-means++ clustering, secondary decomposition and TCN-BiLSTM-Attention model · Electric Power Systems Research · 2026
• Transformer-based deep learning for multi-horizon global power generation forecasting · Energy Conversion and Management · 2026

Related topics in Computer Science