Wind Power Forecast using Machine Learning Approach at Al Duqum Area

As part of Oman’s Vision 2040 to increase renewable energy contribution to the total energy consumption, a 200 MW wind power project is proposed in the Al Duqum area. Accurate wind power forecasting is essential to ensure grid stability, manage variability, and optimize grid management strategies for integrating this renewable energy project into the power grid. This research focuses on developing and evaluating various AI-based models tailored for the Al Duqum area. It explores existing forecasting machine learning methodologies. Furthermore, the steps involved in data collection, preprocessing, and feature selection for choosing the input variables -such as wind speed, wind direction, maximum gust, and gust direction- that significantly influence forecasted values are provided in detail. This paper also examines the configuration and design of different AI-based models, including Extreme Learning Machine (ELM), Long Short-Term Memory (LSTM), Random Forest (RF), and models integrating ensemble techniques such as Adaptive Boosting (AdaBoost) and optimization algorithms like Particle Swarm Optimization (PSO). These models are used to provide a long-term wind power forecasting model based on two historical datasets from the Al Duqum area and a datasheet for the proposed wind turbine. The datasets include daily meteorological data, such as wind speed and wind direction.