Estimation of wind turbine parameters with piecewise trends identification

Parameters of wind turbine and its control system have a major impact on power grid dynamic characteristic and the stability with more and more wind farms interconnected into the grid. It is feasible for genetic algorithms (GA) applied in parameter identification of wind turbine and its control systems using the fault recorded data as the raw signal. In this paper, the raw signal curve is divided into three sections to improve the identification accuracy. The first is the fault section, followed by the power recovery section and the state recovery section. The last two sections have the same beginning point, and the power recovery section is included in the state recovery section. GA is used not only in the distinguish of the three sections but also the parameters identification in each section. In the fault section GA loop, the fault parameters are distinguished. In the power recovery section GA loop, turbine's protection system parameters are mainly identified. In the state recovery section GA loop, wind turbine and its ordinary control system parameters are identified. In one step iteration of each GA loop, parameters represented by the individual genes are used in the power grid time domain simulation, in which the evaluation function is to compute the total deviation of the raw signal and the simulation result. One group of wind turbine parameters are identified with the method and the simulation result is consistent with the raw measured data.