Research On Validity Evaluation And Paramter Calibration Of Generator Model Based On Hybrid Dynamic Simulation

China's power grid becomes more and more large and complex. Only the staff of learning about its dynamic characteristics accurately that the power system can run safely, this must resort to the dynamic simulation tools. So the credibility of simulation results is critical to almost every aspects in power systems, such as planning, design and operation, etc. The generator is the key equipment for the dynamic behavior of power system. The validity of its simulation model will greatly influence the dynamic simulation results. Thus, in this paper the problems of how to evaluate the validity of generator model and how to improve its validity level are analyzed following the order that realization of hybrid dynamic simulation, quantitative method to evaluate the validity of generator model, parameters identification and calibration of model.First, hybrid dynamic simulation based on the data from wide area measurement system (WAMS) is used to achieve simulation model validation of power system with the discrepancy between dynamic simulation results and measured values. Based on the research feature of generator model, this paper proposes a hybrid dynamic simulation validation strategy by using voltage source method. It replaces the external system with voltage source, then incorporates field data of voltage and angle into the boundary buses of simulation subsystem simply and effectively.Then, the differences between simulation results and measured values are quantified to achieve validity evaluation of model. To overcome the shortcomings of the current simulation error algorithm, a new validity algorithm of model based on shape similarity and numerical value similarity is proposed in this paper, which applies related coefficient and residual analysis with weighted improvement. A validity indicator of generator model is established through this methord. The indicator can correctly reflect the accuracy of generator model, and has good anti-noise performance. In addition, dynamic evaluation thoughts about model validity are also given with disturbance depth.Finally, if the validity indicator of generator model is low, the parameters will be calibrated. But too many parameters bring difficulty to calibration. Hence this paper proposes a simplified identification strategy for parameters of generator model based on the sensitivity analysis of perturbation theory. In the strategy only the parameters with great influence on dynamic response are identified to reduce the number of parameters for calibration. What's more, a genetic algorithm is applied to identify simplified parameters of generator and excitation system. The model validity can be improved through parameters calibration, in which the identification results and the original parameters of model are compared in search of error parameters.The proposed algorithms are implemented based on Matlab program and simulation software Simulink/SPS. Case studies on test system with simulation data and generator model with field data demonstrate the efficiency of the method in this paper.