Deep optimization of model predictive control performance for wind turbine yaw system based on intelligent fuzzy deduction

Abstract

Because wind direction has the trait of time varying, yaw is a common state of wind turbine (WT). To improve the energy capture and reduce the yaw actuator usage time, a Model Predictive Control (MPC) method based on the Fuzzy Deduction Weight Coefficient Evaluator (FDWE) is proposed. Specifically, in view of the two contradictory control objectives of energy capture loss ratio and yaw actuator usage ratio in MPC, FDWE is designed to dynamically adjusts the weight coefficient connecting the two objectives according to the predicted wind direction. On this basis, to fully exert the advantage of FDWE, the fuzzy rule and membership function (MF) are tried to be simultaneously optimized. For this complex optimization problem involving different types of multiple design variables, three different solving strategies are proposed: fuzzy rule-MF ordered optimization, mixed integer optimization, and association optimization. Finally, an improved Adaptive Grid Algorithm Multi-Objective Particle Swarm Optimization is presented to solve the formulated optimization problems. The results indicate that the optimized FDWE-MPC using the three strategies not only improves the energy capture of the WT, but also reduces the use of yaw actuator compared to the baseline MPC. Consequently, the proposed method is promising in reducing the production cost for the WTs.