Research On Distributed Dynamic State Estimator For Power System With Large-scale Wind Power Integration

Dynamic state estimation(DSE)can be used for measurement filtering and state trajectory tracking using the constraints of the power system dynamic models.It provides more complete and accurate state information for power system scheduling and control.Large scale integration of wind power results in more obvious fluctuation characteristics of system state and measurement data,which puts forward higher requirements for DSE.Existing DSE research,however,mainly focuses on the traditional power systems dominated by synchronous generators and there is a lack of in-depth discussion on the DSE of power systems with large-scale wind power integration.Therefore,this thesis systematically studies the distributed dynamic state estimation method for power system with large-scale wind power integration.Firstly,DSE is performed separately at power plant and wind farm,and then their estimation results are sent to the system-side dispatch center to implement the system-wide distributed DSE which takes into account the state constraints of dynamic components.The main innovations of the thesis are as follows:(1)Aiming at the problems that the DSE of synchronous generator is more susceptible to bad data and noise parameter uncertainty due to increased measurement fluctuation,the limitations of the conventional cubature Kalman filter are revealed and the bad data is classified as innovation bad data and measurement bad data.Based on the characteristics of large inertia and non-abrupt state of synchronous generator,a bad data suppression method based on exponential smoothing prediction and innovation anomaly detection is proposed.Further,the influence of noise statistical parameter uncertainty is analyzed and the adaptive H∞ filtering method based on the innovation norm is used to suppress the influence of the uncertainty of the noise parameters,which effectively improves the robustness of the DSE method of synchronous generator.(2)For the problems that DSE results are easy to diverge due to high order of doublyfed induction generator(DFIG)and fast dynamic of its state,the converter outputs are regarded as the unknown inputs of DSE,and then the simplified DSE model of DFIG is established.The augmented state cubature Kalman filter is used to estimate the state and unknown input vector simultaneously,which eliminates the dependence of DSE on the DFIG control strategies.Further,the influencing factors of the discretization accuracy of the continuous system are analyzed,and an adaptive measurement interpolation method based on the local truncation error estimation is proposed to overcome the divergence problem of the existing DSE algorithms and improve the computational efficiency.(3)In view of the problems that wind farm DSE has high measurement requirements and the measurement data quality problems are more prominent in renewable energy power system,based on the characteristics that a wind farm can be considered as a small microgrid,a method for constructing redundant measurement sets considering the spatiotemporal correlation of wind farm measurements is proposed,which effectively reduce the requirement of measurement configuration.The robust weighted least absolute value method is used to estimate the DFIG terminal electrical quantities,which reduces the measurement noise and filters out the bad measurement data before DSE.Then a process noise variance correction method based on the process noise ratio coefficient is proposed to suppress the influence of the innovation bad data.(4)In order to solve the problem that the state transition equation of bus voltage is difficult to establish accurately and the centralized estimation mode is difficult to meet the real-time requirements of DSE,the constraint relationship between dynamic state and algebraic state of power system is analyzed.Based on the DSE results of the dynamic components,the state transition equation of bus voltage considering system dynamic state constraints is derived.Furthermore,based on the principles of hierarchical coordination and distributed estimation,the method of estimation-coordination-correction is adopted to realize the fast and accurate acquisition of system state information.