Improving Wind Turbine Pitch Control by Effective Wind Neuro-Estimators

Data about wind are usually available from different databases, for locations. In general, this information is the average of speed over time. The reports crucial designing turbine controllers. But when working with floating offshore turbines (FOWT), two problems arise regarding measurement. On one hand, there no buoys at deep sea, but near coast where not so strong neither stable; measurements do fully correspond to reality. other these devices subjected extreme environmental conditions (waves, currents, ...) that produce disturbances and thus may distort measurements. To address problem, work presents a novel pitch neuro-control architecture based on neuro-estimators effective wind. control system composed proportional-integral-derivative (PID) controller, lookup table, neuro-estimator, virtual sensor. neuro-estimator used estimate in FOWT forecast its future value. Both current signals combined power controller. sensor also provides measure related turbine, such as angle angular velocity generator. Neural networks trained online adapt changes environment. Intensive simulations carried out validate effectiveness neuro approach. Controller performance compared PID, obtaining better results. Indeed, an improvement 16% sinusoidal 8% observed.