Artificial Intelligence-Based Approach for the Control of PMSG Based Wind Energy System

Assessment of wind energy potential is extremely important before installation of wind turbine (WT) in any location. However, the intermittency of wind speed makes it more difficult to estimate the power potential. Thus, in this paper, two intelligent machine learning techniques namely neural network (NN) and adaptive neuro-fuzzy inference system (ANFIS) were employed to estimate the maximum power from the wind turbine and generates a reference torque to drive the rotor of PMSG. A comparative analysis for both NN based WT and ANFIS based WT is presented, while NN based WT estimated more accurate power than the ANFIS based WT. The NN based WT showed an error as low as 0.039% while ANFIS based WT shows 0.453%. Thus, the NN based WT model is selected and coupled with the PMSG system and is tested using wind data recorded at 100 meters height for Yanbu city, KSA. The PMSG was implemented with speed control mode using a PI controller which makes the NN-WT-PMSG effectively tracks the reference rotor speed and torque and generate proportional electromagnetic torque.