Research On Load Modeling Method Based On PMU Measured Small Disturbance Data

Power system operations depend on a series of simulation calculation,such as power flow calculation,transient stability analysis,small disturbance stability analysis,etc.The accuracy of calculation results directly affects the safety and stability of power systems.And the basis of power system simulations is the establishment of simulation models.Among the models of power system components,the accurate load models are difficult to establish due to the time-varying characteristic,randomness and complexity of power loads.The load modeling methods in previous literature generally use the response data of specialized power system tests under the condition that the voltage amplitude fluctuations are greater than 10%.In contrast,the small disturbance data widely exits in the actual power systems and is easy to be obtained from the measurement units.However,the application of the small disturbance data for load modeling has not been widely reported yet.Moreover,the existed researches have not discussed whether the identification of load models is open-loop system identification or closed-loop system identification.Therefore,this dissertation proposes closed-loop and open-loop identification methods of load model based on Phasor Measurement Unit(PMU)measured small disturbance data.The main contributions of this dissertation are as follows.1.A series of problems about load modeling based on PMU measured small disturbance data are analyzed,including the characteristics of PMU data,the structure of the load model used in this dissertation,parameter identification methods,etc.First,the working principles of WAMS and PMU are described and the characteristics of PMU data are summarized.The PMU data is measured synchronously with a high accuracy,which leads to a huge amount of data.Next,the composited load model used in this dissertation is established and the relationship between coefficients of the load model in transfer function form and the transient parameters of the load model is derived analytically.Finally,system identification is discussed and different identification methods are compared,which are beneficial to the appropriate selection of load model identification methods according to different conditions.2.An approach is proposed for the closed-loop identification of load model based on PMU measured small disturbance data.The small disturbance data excited by ambient signal is used to identify the load models.And the load model order is chosen by the sum of squared errors between the response output data of the load model and the PMU data.Moreover,the noise model is identified,which helps to obtain a consistent estimation of the load model parameters.The new approach mainly improves some shortcomings of the previous load modeling methods,such as needs of specialized tests to obtain identification data and the lack of methods for closed-loop system identification.In addition,the condition when the load models are identifiable is discussed in this dissertation based on the input data,output data and the load model structure.Finally,the proposed method is directly applied to the Zhejiang grid for load model identification,and the response output data of the identified load models is verified by the PMU data of the actual power systems.3.The condition when the open-loop system identification methods are applicable for load model identification is given in this dissertation.More specifically,when the disturbance of other load nodes is much larger than that of the load node to be tested,the open-loop system identification methods are applicable.And this dissertation proposes an open-loop system identification method for load modeling based on PMU measured small disturbance data.The data set which has relatively small fluctuation of the current amplitude is used to identify the load model.And an algorithm is proposed to correct the wild values of the measured data.Compared with the existed open-loop system identification methods for load modeling,the proposed method can not only realize the online identification of load models,but also obtain the consistency estimation of load model parameters,which meets the needs of the time-varying characteristic and the accuracy of load models.