Bearing Power Loss Predictions in Wind Turbine Gearbox: An Approach Based on LLMs

Janice Anta Zebaze, Azanzi Jiomekong, Innocent Souopgui, Germaine Djuidje Kenmoe

Abstract: A constant and consistent supply in electrical energy in a location is a reflection of a good economy. Developing countries nevertheless don't have access to this quality of energy, which slows down their economy and consequently development. Wind is a clean, sustainable and renewable resource which can be used to meet the energy needs in such countries. However, the intermittent nature of wind yields fluctuations on the amount of energy produced by a wind turbine. Coupled with frictional power losses in the wind turbine gearbox bearings, one can't be sure on the exact amount of energy that will be produced. This leads to the distribution and management issues. To tackle this issue, we propose here the use of Large Language models. These are tools which have been proving their potential in various domains till date and whose potential are still to be seen in the field to our knowledge. Taking advantage of their flexibility and adaptability to any model and dataset, we intend to explore its abilities in the fields of wind energy and tribology. Making use of available data, predictions on the wind energy potential and power losses will be carried out using Large Language models such as BERT. The results of this work intends to promote the use of wind energy by lifting barriers in thee management and knowledge of the resource.