Model-Free Optimizing Control for Wind Energy Capture and Controls of Floating Wind Turbines




Journal Title

Journal ISSN

Volume Title



Ramification of wind power generation critically depends on the reduction of its levelized cost of energy (LCOE), for which enhancing the energy capture and reducing the fatigue loads are two major pillars. Development of advanced control strategies for enhancing power capture and reducing the structural loads has become a primary aspect for the operation of individual turbines and the whole wind farm, offshore or onshore. This dissertation research aims to pursue investigations on advanced control strategies that cover both energy capture enhancement and load reduction: 1) extremum seeking based model-free control for enhancing wind turbine and wind farm energy capture with improved convergence, and 2) controls of floating offshore wind turbines (FOWT) with novel actuations. For wind turbine operation below the rated wind speed (i.e. Region 2), the primary control objective for variable-speed variable-pitch turbines is to optimize the generator torque for maximizing the energy capture. Extremum seeking control (ESC) has emerged as an appealing model-free Region-2 control method with much less dependency on turbine characteristics and wind speed measurement. However, the previous work of ESC Region-2 control with the rotor power feedback suffers from undesirable convergence due to wind fluctuation. An estimated power coefficient is proposed as the ESC objective function in order to reduce the sensitivity of the optimum seeking process to wind fluctuation. The hub-height free-stream wind speed is estimated with the nacelle anemometer measurement based on the so-called nacelle transfer function (NTF) derived between the nacelle anemometer and met-tower measurement. Also, an ESC integrated inter-region switching scheme is proposed to avoid the load increase associated with the ESC operation. FAST based simulation study shows that the proposed method achieves more robust convergence to wind fluctuation compared to the power feedback based ESC. For Region-2 control of wind farm operation, the Nested-Loop Extremum Seeking Control (NLESC) has demonstrated its effectiveness in enhancing energy capture at the farm level, however, its convergence speed has been highly limited by the delay of wake propagation between the upstream and downstream turbines as the dither signals for optimizing upstream turbines needs to be of much lower frequencies. In this dissertation research, the NLESC is enhanced with a predictor based delay compensation (DCNLESC), based on which the dither frequencies of upstream turbines can be retained at the level for individual turbine operation. The proposed scheme is validated with a three-turbine model in SimWindFarm. The research FOWT controls is conducted for tensioned-leg platform (TLP), and two novel actuation schemes are proposed for motion stabilization and load reduction: dynamic vibrations absorber (DVA) and active mooring line force control (AMFLC) with fishing line artificial muscle (FLAM) actuator. First proposed is to deploy the vertically operated DVAs at the spokes of TLP structure. Via the Lagrange’s equations, a control-oriented model of 11 degrees-offreedom (DOFs) is derived for the TLP-FOWT-DVA system, with which a linear quadratic regulator (LQR) is designed for stabilizing the platform pitch and roll motion. The LQR controller, turbine controller, as well as the DVA model are implemented in Simulink, which is coupled with the wind turbine model in FAST via a dedicated interface. Simulations are performed for 9 m/s and 18 m/s turbulent winds with different wind and wave directions. The simulation results show that the platform motion and tower loads in the lateral direction and mooring line tension load are significantly reduced, while the tower load in the fore-aft direction can be moderately reduced. For achieving the performance in platform motion stabilization and load reduction, the average power consumption of the DVA actuators is less than 0.27% of the wind turbine power generated during the simulated periods. As for the second concept of TLP-FOWT control, the FLAM actuator is proposed to be deployed to the junction between the mooring lines and TLP spokes, realizing AMLFC of TLP-FOWT. The FLAM actuator consists of multiple bundles of twisted nylon fishing lines, with the contracting and stretching forces induced by thermal actuation. A simulation model of the FLAM actuator is developed in Simulink, along with an interface to the mooring line model of TLP-FOWT in FAST. The dynamic model of the FLAM actuator is obtained with ANSYS simulation, and a control-oriented model is obtained for the FOWT platform motion. First, an LQR controller is implemented to validate the proposed framework. Then, based on the development of dynamic hybrid automata (DHA) model for the TLP-FOWT system with FLAM actuators, a hybrid model predictive controller is developed with the inclusion of information on incoming wind and wave. Simulation study shows that, with mild power consumption, the proposed AMLFC strategy can significantly reduce the platform roll motion and the tower-base side-side bending loads with little impact on the rotor speed and power output.



Wind turbines, Offshore wind power plants, Active noise and vibration control, Tension leg platforms, Structural optimization


©2020 Zhongyou Wu. All rights reserved.