Urban Wind Energy Assessment Using Machine Learning and Multi-criteria Analysis

The transition toward renewable energy sources in urban environments presents technical and strategic challenges due to the variability of available resources and the inherent constraints of built-up spaces. This study assessed the potential for wind energy generation in Bogotá, D.C., Colombia, applying data science techniques and machine learning models. Clustering analysis, the Analytic Hierarchy Process (AHP), and Weibull distribution modeling used historical meteorological data on wind speed and temperature. The unsupervised analysis identified three representative hourly wind patterns, while the Weibull distribution estimated an operational wind speed of 2.2 m/s. The AHP method facilitated the selection of wind turbines based on technical criteria such as cut-in speed, rated power, and blade number. Among the key results, the 300 X-300 turbine achieved a Capacity Factor (CF) of 25.43%, a Yield Ratio (Yr) of 0.1352 h, and a Performance Ratio (PR) of 43.49%. These indicators reveal that, despite the moderate wind potential during specific periods, the low energy density and high variability limit the technical feasibility of wind systems in the study area. It is concluded that integrating data analysis tools with multi-criteria decision-making methods provides a robust framework for assessing urban wind resources and establishes a solid foundation for the tailored design of sustainable energy solutions.