A systematic literature review (SLR) on the integration of renewable energy sources (RESs) via machine learning (ML) modelling is conducted to identify the most widely used ML-based approaches for forecasting RES power outputs. The review highlights that deep artificial neural networks (ANNs), particularly long-short term memory (LSTM) networks, and ensemble strategies are the most effective for RES power output forecasting due to their ability to model autoregressive patterns and handle fluctuating data. These methods outperform traditional statistical and physical models in terms of accuracy, robustness, and generalisation. The review also discusses the integration of forecasted RES outputs into decision-making problems such as unit commitment and economic dispatch to address grid challenges. Key findings include the popularity of ML-based techniques like ANNs, support vector machines (SVMs), and ensembles in RES forecasting. The review emphasizes the importance of accurate RES power output prediction for grid stability and efficient coordination between transmission and distribution system operators. The results show that most SLR articles focus on short-term forecasting, with fewer addressing very short-term and long-term horizons. Data collection primarily relies on historical on-site measurements, though benchmark datasets are also used for comparison. Weather parameters, historical power data, and satellite imagery are commonly used features in forecasting models. The review identifies trends in ML-based forecasting, including the use of deep neural networks, SVMs, and ensembles, and highlights their potential for improving grid operations and addressing economic, operational, and managerial challenges. The study concludes that further research is needed to optimise ML-based forecasting models for different RES types and to incorporate network-related parameters for more accurate predictions.A systematic literature review (SLR) on the integration of renewable energy sources (RESs) via machine learning (ML) modelling is conducted to identify the most widely used ML-based approaches for forecasting RES power outputs. The review highlights that deep artificial neural networks (ANNs), particularly long-short term memory (LSTM) networks, and ensemble strategies are the most effective for RES power output forecasting due to their ability to model autoregressive patterns and handle fluctuating data. These methods outperform traditional statistical and physical models in terms of accuracy, robustness, and generalisation. The review also discusses the integration of forecasted RES outputs into decision-making problems such as unit commitment and economic dispatch to address grid challenges. Key findings include the popularity of ML-based techniques like ANNs, support vector machines (SVMs), and ensembles in RES forecasting. The review emphasizes the importance of accurate RES power output prediction for grid stability and efficient coordination between transmission and distribution system operators. The results show that most SLR articles focus on short-term forecasting, with fewer addressing very short-term and long-term horizons. Data collection primarily relies on historical on-site measurements, though benchmark datasets are also used for comparison. Weather parameters, historical power data, and satellite imagery are commonly used features in forecasting models. The review identifies trends in ML-based forecasting, including the use of deep neural networks, SVMs, and ensembles, and highlights their potential for improving grid operations and addressing economic, operational, and managerial challenges. The study concludes that further research is needed to optimise ML-based forecasting models for different RES types and to incorporate network-related parameters for more accurate predictions.