Linear parameter varying based adaptive model predictive control for power down regulation of floating offshore wind turbines in region–3

Floating offshore wind turbines (FOWTs) enable access to abundant deep-water wind resources and are rapidly gaining traction worldwide as more and more floating wind farms are getting deployed. However, the existing farm-level control strategies often rely on simplified turbine models that overlook the coupled aero-hydro-servo-elastic dynamics of FOWTs. At the wind farm level, power setpoints are allocated using methods such as greedy algorithms, induction control, or wake steering, with the latter two requiring individual turbines to operate in power curtailment mode, i.e. downregulation or derating. This however, can impact platform motions, fatigue loads, and mooring stresses, if plant-wise wind farm control is not optimized. Unlike land-based turbines, FOWTs exhibit unique control challenges such as changes in platform pitch damping sign in region-3, which restrict control bandwidth up to certain wind speeds above rated wind, and impacts power regulation. The systematic design of individual turbine-level controllers for tracking varying power references under these dynamics remains underexplored, particularly with high-fidelity validation. To address this gap, we propose an adaptive model predictive control (AMPC) scheme based on gridded linear parameter-varying (LPV) model, constructed from high-fidelity simulations and scheduled by wind speed and derating level, with gap-metric theory guiding model selection. The control scheme is implemented in MATLAB/Simulink, and tested on the NREL 5 MW OC3 sparbouy in OpenFAST, and results demonstrate improved power tracking and load reduction under varying test conditions compared to down regulation capable gain-scheduled PI control.