Machine Learning Optimises Gearbox Design for Wind Turbines
Source: journals.sagepub.com
- Researchers used machine learning to quickly design better gearboxes for wind turbines by predicting performance without full simulations.
- The method cut design time from weeks to hours, achieving 98% accuracy in stress predictions for gear teeth.
- This speeds up renewable energy development, making wind power cheaper and more reliable.
Engineers at the University of Nottingham developed a machine learning approach to optimise planetary gearbox designs specifically for wind turbines. They trained neural networks on simulation data to predict key stresses and deflections in gear components, bypassing slow finite element analysis. The core finding is that this surrogate model delivers fast, accurate results, enabling better designs with lower weight and higher efficiency. It matters because wind energy is booming, and faster gear