Enhancing Fault Detection in Wind Turbines Using Ensemble Algorithms

Arias Velásquez R.M.; Castro Romero A.M.; Conde Chumpitaz P.R.; Mejía Lara J.V. · Lecture Notes in Networks and Systems (2025) · DOI: 10.1007/978-3-032-00712-4_2

Resumen

This paper addresses the critical issue of wind turbine failure, focusing on the detection and mitigation of anomalies in gearbox components, specifically oil and bearing temperatures. Historical data analysis reveals the gradual deterioration of main bearings during operation, necessitating proactive maintenance measures. Two predictive models are developed using ensemble algorithms, achieving high accuracy of 94.7154% for gearbox oil temperature and 95.758% for gearbox bearing temperature. Comparative analysis shows slight differences in accuracy and computational time between the models. Practical implications include the identification of values exceeding operational limits, aiding in maintenance scheduling and efficiency improvement. These models offer valuable insights for enhancing wind turbine reliability and reducing energy losses associated with mechanical failures. Further refinement and integration of these models could significantly benefit wind turbine operations. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

Instituciones: Universidad Tecnológica del Perú