Bayesian reinforcement learning with mcmc to maximize energy output in hardware-in-the-loop simulations of vertical axis wind turbine

Official URL: http://risc01.sabanciuniv.edu/record=b2583863_(Table of contents)

Vertıcal Axıs Wınd turbınes(VAWT) are ideal for small-scale use in urban areas. The energy produced is optimized by controlling the wind speed to rotor tip speed ratio. The proposed control method which uses a neural network trained by Monte Carlo methods is run on a Hardware-in-the-Loop setup modelled after the VAWT. Once trained, the proposed method is compared to other commonly used control methods such as MPPT and SNC. It is found to produce satisfactory results with energy produced being more than MPPT all the time. It sometimes performs better than SNC.