Stochastic estimation and control of wind turbines for enhancement of efficiency

This dissertation is concerned with the problems associated with control system design of variable speed wind turbines. Two of the major technical difficulties in the design of wind turbine control systems are addressed: (1) The uncertainty in the system model, especially the power coefficient which characterizes the maximum aerodynamic efficiency and the optimal operation point of a particular turbine. (2) The inability to monitor the wind speed experienced by the turbine in practice to the accuracy required.;The major result of this dissertation is an algorithm to estimate the power coefficient of variable speed wind turbines from commonly available observation data, namely the rotor angular velocity as derived by angular displacement and the generator output power. We formulate the identification of the power coefficient of variable speed wind turbines as a parameter estimation problem for a continuous-time nonlinear system with discrete time measurement. A generic Newton-Raphson algorithm is applied to the estimation problem to identify the power coefficient in the presence of observation noise.;As a specialization of the power coefficient estimation algorithm, a stochastic wind speed estimation algorithm which requires only the rotor angular velocity measurement (as derived by angle data) is proposed. The algorithm is designed to work under any arbitrarily specified control law.;The theory and algorithms are verified by Matlab simulation based on simulated data. The effects of the initial guess, the noise variance, and the time history are studied in detail. Experiments were also conducted on a 600kW full-scale research wind turbine to evaluate the performance of the algorithms on real data.