Research On Parameter Identification Of Doubly Fed Wind Turbine And Dynamic Equivalent Method Of Wind Farm

With the continuous and in-depth development of energy revolution in the world,renewable energy power generation has become the strategic demand of world energy.Wind power generation,as a typical representative,has an steadily increasing of installed capacity and its penetration rate in power system is also enhancing.At present,among all kinds of wind turbines in commercial operation,doubly fed induction generator(DFIG)has become one of the most widely utilized wind turbines in the world as a consequence of its low mechanical stress,high energy efficiency,and comparatively low rated power of converters.Therefore,this paper focuses on the DFIG-based wind power generation system,and carries out a series of research as follows:(1)The structure and basic principle of DFIG are introduced,the dynamic models of its components are established,and the numerical solution process of DFIG-based wind power system is deduced and analyzed so as to be the foundation of the follow-up research.(2)A parameter identification method of DFIG’s converter control system is proposed.Based on its mathematical model,the identifiability of the converter control system parameters is analyzed,and the difficulty-eazy degree of identification is obtained utilizing trajectory sensitivity,taking the output power is as observation,.Using the measurable electrical time series,the discrete extended Kalman filter iterative equation is obtained by strict derivation.Thus the control parameter identification model is established.The parameter identification results are obtained through repeated simulation,and the feasibility and robustness of the method are further verified.(3)Aiming at the nonlinear state equation of DFIG,a dynamic state estimation model is established by using adaptive cubature Kalman filter.Based on the principle of volume numerical integration,equal weight volume points are constructed to approximate the unit state variables.At the same time,the adaptive technology is introduced to estimate the process noise covariance in real time through the Sage-Husa estimator and the dynamic state estimation curves of DFIG are obtained.Analogue simulation on the test system are conducted,and the state estimation results demonstrate that the proposed approach has higher estimation accuracy.(4)A multi-machine dynamic equivalent approach of DFIG-based wind farm considering the parameters of converter control system is put forward.Based on the measured data,starting from the structural similarity of the wind turbine output characteristic time series,the clustering index is established by combining the morphological similarity distance and cosine similarity between the output trajectories.The cluster division is developed with K-means clustering algorithm,and the robustness of the clustering method in different scenarios is analyzed by combining with the contour coefficient.Aiming at the dynamic equivalence of the wind turbines in each cluster,a trajectory sensitivity based key parameters selection is introduced to reduce the control parameters to be identified.A multi-objective adaptive function is established utilizing the power output characteristics of the wind farm’s point of common coupling.The adaptive evolutionary multi-objective particle swarm optimization algorithm is applied to identify the converter control system parameters of the equivalent units in the cluster.Thus,there comes a multi-machine equivalent model of wind farm.Finally,the simulation in the test system verifies the accuracy of the equivalence method in multi-disturbance scenarios.