Abstract
In this paper, we demonstrate a Bayesian Network (BN) model that returns indicators in terms of range of values and likelihoods of occurrence from the inputs of operating conditions, i.e. inflow wind speed, rotating speed, tilt and pitch angles. The BN could also be used to derive operating conditions (i.e. rotation speed or pitch angle of blade) that maximize the possibility of achieving a desired power output while subject to certain turbine structural conditions.
The BN model is built from a large database of high-fidelity CFD/FEA simulations of a 5MW NREL wind turbine. The abnormality is assumed to be in the form of abnormal deformation and/or stress that are numerically generated by introducing corresponding faults in the blade structure. The database generation, construction and demonstration of the BN model will be presented in the paper.
