LIDAR Assisted Model Predictive Control of a Next Generation Wind Turbine for Tower Fatigue Load Reduction and Improved Speed Control

摘要

The introduction of turbine-mountable, LIDAR-based, wind-field sensing technology has prompted investigation into methods of leveraging this new data to improve turbine controller performance. Model Predictive Control (MPC) is a candidate method to utilise future wind speed data whilst controlling for multiple control objectives. The optimisation can include system constraints, system nonlinearities, simultaneously control multiple actuators to achieve multiple control objectives and incorporate disturbance information such as that supplied by LIDAR; making this control framework extremely relevant to the wind turbine application. We present simulation comparisons of MPC performance against that of a classically designed state-of-the-art baseline controller on a next-generation offshore wind turbine. The comparison demonstrates that with a range of sources of model uncertainty and noisy/delayed measurements, MPC is able to perform speed control and tower damping on par with the optimised baseline; however model uncertainty for the in-plane turbine dynamics has led to increased generator torque actuation which may increase inplane loading. The MPC controller sees significant speed control and tower damping performance improvements when provided with realistic LIDAR feedback. This additional performance can allow designers to redistribute control focus depending on the optimal tuning level for the lowest cost-of-energy.