Research On Measurement-Based Load Modelling Of Power System

With the concept of the global energy Internet,it is put forward that the interconnection of the power gird is more extensive and the system capacity is greater. A large number of renewable energy that accessed into network has caused system simulation with the traditional model may not reflect the real operation of the system. The traditional simulation precision and system control need to be improved due to the influence of load accuracy. In the aspects of system simulation, planning, control, load forecasting, and so on, more and more accurate load models are needed.In the system planning and design,the ZIP static model or a motor integrated model are used in the system.using the traditional load model makes the system simulation results have a big gap between the actual results.And the simulation of the system is limited by the traditional load model.In this paper,firstly, comparison and analysis of the shortcomings of traditional and interpolation class static models in terms of fitting accuracy. Then, a static load model based on the combination of the traditional class model and the three interpolation model is studied, and the characteristics of the model in the fitting accuracy of the model are researched.And then the process of model simplification and general model solving process of three order induction motor is analyzed, and the processing method of the measured data is obtained according to the simplified condition of the model. Parameter identification uses improved particle swarm optimization, the dynamic mechanism model of the three order induction motor parallel ZIP is modeled by using the measured DFR data.Finally, the characteristics of the measured data are analyzed to simplify the two order differential dynamic non mechanism model for the purpose of reducing the model parameters. And the validity of the simplified model and the applicability of the model in different fault conditions are verified by the identification of the measured data and MATLAB/Simulink simulation data. Finally, the time and space complexity of the identification algorithm is analyzed between the dynamic mechanism and the non mechanism model. The strategy of online analysis and offline analysis selection model is given.