Control of wind turbine power and vibration with a data-driven approach

An anticipatory control scheme for optimizing power and vibration of wind turbines is introduced. Two models optimizing the power generation and mitigating vibration of a wind turbine are developed using data collected from a large wind farm. To model the wind turbine vibration, two parameters, drive-train and tower acceleration, are introduced. The two parameters are measured with accelerometers. Data-mining algorithms are applied to establish models for estimating drive-train and tower acceleration parameters. The prediction accuracy of the data-driven models is examined in order to address their feasibility for an anticipatory control scheme. An optimization control model is established by integrating the data-driven models in the presence of constraints. A particle swarm optimization algorithm is applied to optimize the model.