A supervised machine-learning method for optimizing the automatic transmission system of wind turbines

Abstract

Large-scale wind turbines mostly use Continuously Variable Transmission (CVT) as the transmission system, which is highly efficient. However, it comes with high complexity and cost too. In contrast, the small-scale wind turbines that are available in the market offer a one-speed gearing system only where no gear ratios are varied, resulting in low efficiency of harvesting energy and leading to gears failure. In this research, an unsupervised machine-learning algorithm is proposed to address the energy efficiency of the automatic transmission system in vertical axis wind turbines (VAWT), to increase its efficiency in harvesting energy. The aim is to find the best adjustment for VAWT while the automatic transmission system is taken into account. For this purpose, the system is simulated and tested under various gear ratios conditions while a centrifugal clutch is applied to automatic gear shifting. The outcomes indicated that the automatic transmission system could successfully adjust the spinning in line with the wind speed. As a result, the obtained level of harvested voltage and power by VAWT with the automatic transmission system are improved significantly. Consequently, it is concluded that automatic VAWTs, equipped with the machine-learning capability can readjust themselves with the wind speed more efficiently.

Savonius wind turbine blade design and performance evaluation using ANN-based virtual clone: A new approach

“… Ref. & PY Turbine Type Dataset Algorithm Feature [ 52 ] 2019 SWT CFD Simulation ANN, GA Improving the efficiency by designing a deflector plate [ 69 ] 2014 VAWT Experimental ANN Turbine rotor's performance coefficient and torque coefficient estimation [ 70 ] 2021 VAWT Experimental ANN Investigating the influence of aerodynamic parameters of the turbine [ 71 ] 2021 Not specified Simulation Long Short-Term Memory (LSTM) Predicting the parameters [ 72 ] 2022 VAWT Simulation Random Forest (RF) Automatic transmission system optimization [ 73 ] 2017 Not specified CFD Simulation Mean, LR, M5, RF Wind speed prediction [ 74 ] 2018 SWT CFD Simulation ANN, GA Parameter optimization Overlap ratio, Number of stages, Blade rotation [ 75 ] 2016 SWT Simulation ANN Prediction of aerodynamic characteristics [ 76 ] 2019 Not specified …”