Enhancing Wind Power Forecasting Accuracy in Canada Using a Solar Data-Enhanced Hybrid Machine Learning Model: Integrating ANN, LSTM, and SVR

As the world moves toward sustainable energy solutions, wind power emerges as a pivotal renewable energy (RE) source due to its accessibility and zero carbon emission. However, its unpredictable nature poses significant forecasting challenges, that impact energy management efficiency. This study tackles this vital challenge by integrating solar power data into advanced machine learning models to enhance the forecast accuracy and quantifying uncertainty of wind power. The MERRA 2 - dataset a comprehensive atmospheric reanalysis from NASA - spanning 2017 to 2019 across three locations of Canadian provinces, British Colombia, Manitoba, and Nova Scotia has been considered for this work. A novel hybrid machine learning framework that combines the strengths of Artificial Neural Networks (ANN), Long Short-Term Memory Networks (LSTM), and Support Vector Machines (SVM), has been used. Utilizing this framework excels in pattern recognition, temporal data processing, and regression analysis, effectively will improve the precision of wind power forecasts. Besides, it provides a robust framework for quantifying forecast uncertainty and enhancing decision making in renewable energy management. The superiority of this model is demonstrated through comparative evaluations against conventional methods using various metrics to establish its efficacy and applicability in real world scenarios.