Renewable energy sources integration via machine learning modelling: A systematic literature review

Area of Science:

• Electrical Engineering
• Computer Science
• Environmental Science

Background:

• Renewable energy sources (RESs) integration presents grid stability challenges due to their stochastic nature.
• Traditional forecasting models (physical, statistical) have limitations in accuracy and computational efficiency.
• Machine learning (ML) offers powerful data-driven tools for analyzing complex RES data.

Purpose of the Study:

• To conduct a systematic literature review on ML-based approaches for RES power output forecasting.
• To identify the most effective ML techniques for managing RES uncertainty.
• To discuss the integration of RES forecasts into grid management and future research directions.

Main Methods:

• Systematic literature review of ML-based forecasting methods for RES.
• Analysis of deep artificial neural networks (e.g., LSTMs) and ensemble strategies.
• Evaluation of ML performance against traditional forecasting models.

Main Results:

• Deep artificial neural networks, especially Long-Short Term Memory (LSTM) networks, excel at modeling RES power output's autoregressive nature.
• Ensemble strategies effectively handle large, fluctuating RES datasets.
• ML-based forecasting outperforms traditional methods in accuracy and efficiency.

Conclusions:

• LSTM networks and ensemble methods are highly suitable for accurate RES power output forecasting.
• Effective RES uncertainty management via ML is vital for grid integration.
• Future research should focus on integrating ML forecasts into operational grid decision-making.