Assessment of wind energy potential in northwestern Iraq: A hybrid optimization method based on Weibull distribution and intelligent algorithms

Abstract

This study presents a comprehensive comparative analysis of conventional and artificial intelligence (AI)-based methods for estimating Weibull distribution parameters using 10 years (2013-2023) of hourly wind speed data from six sites in northwestern Iraq: AlKaam, AlShirgat, Baiji, Nenwa, Sinjar, and Haweja. Three traditional methods, the method of moments (MOM), the maximum likelihood method (MLM), and the least squares method (LSM), are compared against three AI algorithms: particle swarm optimization (PSO), differential evolution (DE), and genetic algorithm (GA). Model performance is evaluated using statistical indicators, including RMSE, MAE, R², and chi². Results show that PSO consistently outperforms all other methods, achieving the highest coefficient of determination (R²) of 0.998 and the lowest error values across all locations. Among conventional techniques, MLM achieved superior accuracy, with an average RMSE of 0.073 and an R² of 0.57. Wind resource assessment revealed Haweja as the most promising site, with a mean wind speed of 5.69 m s^–1, followed by AlKaam (5.59 m s^–1) and Baiji (5.57 m s^–1). Economic evaluation showed that using PSO-estimated parameters significantly improves energy forecasting accuracy and reduces the levelized cost of energy (LCOE) to as low as $0.03/kWh at Haweja. The 2.5 MW WT5 turbine achieved optimal performance, generating over 8000 MWh year^–1 with capacity factors exceeding 39% at high-potential sites. This work provides a robust framework for accurate wind energy potential assessment in Iraq and similar arid regions, demonstrating the critical advantage of AI-driven parameter estimation in enhancing both technical and economic feasibility of wind power projects.