A wind speed estimation method of a wind power generator is provided to estimate the wind speed continuously increased and decreased, exactly and to secure excellent tracking performance of an SVR(Support Vector Regression) algorism for an actual value regardless of a constant of the power generator or a turbine. A wind speed estimation method of a wind power generator estimates wind speed in a controller of the wind power generator by using an SVR algorism and uses a system comprising an input unit, an output unit, a memory, and a calculation unit. During a training process, turbine speed, a set of turbine output and wind speed, and a constant parameter are input through the input unit and a Kernel function k is set by the input unit. A Kernel function, Lagrange multipliers, and bias b are computed in the calculation unit. The computed input/output relation is displayed through the output unit. An allowable error parameter and a parameter for controlling complexity of a model are controlled through the input unit. During a wind speed estimating process, an SVR calculation unit(10) is defined by installing the SVR algorism in the memory for the controller of the wind power generator. The real turbine speed and a Kernel function k are computed in the SVR calculation unit.
展开▼