Output Power Smoothing of Doubly Fed Induction Generator Wind Turbine Using Very Short Term Wind Speed Prediction Based on Levenberg–Marquardt Neural Network

Abstract

Wind energy is a renewable energy with great potential. Unlike fossil fuels, wind energy is clean, pollution-free from CO2 emissions, and inexhaustible. However, wind speed is not constant, it fluctuates rapidly, and is uncontrollable. Fluctuating wind speed causes fluctuating output power at the wind turbine. Fluctuating wind power causes the grid frequency to fluctuate, which in turn reduces the quality of the transmitted power and generates instability in the power system. To reduce wind power fluctuations, the output power smoothing method can be used. This paper proposes the smoothing power output method without using energy storage devices to produce a constant output power of the doubly fed induction generator. The fluctuating wind speed generates constant output power based on wind speed predictions. Predicted average wind speed using neural networks with the Levenberg–Marquardt learning algorithm is based on the measurement data of wind speed in Indonesia, Nganjuk prefecture. Simulations are performed using Matlab Simulink. Simulation results show that the output power can be kept constant for a certain period of time. The speed of the rotor with this proposed method has an average above optimal rotor speed.