Research On Transient Stability Assessment Method Based On Bata-Model Hybrid Drive

With the rapid development of large-scale renewable energy and AC/DC hybrid power systems,the proportion of components with power electronic characteristics in the power system has increased rapidly.The problem of "dimension disaster" faced by the transient stability analysis of new energy power systems has become more prominent,and the traditional model-driven modeling and solving methods are difficult to continue.In order to solve the problem of "dimension disaster",combined with the increasing frequency and density of data acquisition in new energy power system,this paper proposes a new transient stability assessment method based on datamodel hybrid drive for new energy power system.The main framework of the research content mainly includes two parts:wind farm equivalent model for transient stability analysis and time-domain simulation analysis of transient stability based on datamodel hybrid drive.The main idea is that,first of all,one wind farm is equivalent to one wind turbine,so that the grid connected to multiple wind farms is converted into the grid connected to multiple wind turbines.Then,on this basis,combined with the time domain simulation method,the data-driven model is used to replace the differential equations in the power electronic components to realize the rapid simulation of the transient process of the new energy power system.The research on wind farm equivalent model for transient stability analysis carries out physical characteristics aggregation and equivalence and data-driven equivalence of the entire wind farm in turn,and finally the wind farm with multiple wind turbines is equivalent to a single wind turbine model.In the process of physical characteristics aggregation and equivalence,based on the idea of physical model drive,the active transient response characteristics and output characteristics of wind turbine connection points at different wind speeds are obtained through simulation,and all wind turbines are aggregated into four equivalent wind turbines using K-means method.Based on the data-driven strategy,the data-driven equivalence establishes a multi-period optimization model with the minimum active power deviation at the wind farm connection point,and combines the LSTM neural network training to aggregate four equivalent wind turbines into a single wind turbine model,realizing the equivalent aggregation of transient process of the entire wind farm.Finally,an example with 32 wind turbines is used to verify that the proposed equivalent method can effectively reflect the transient process characteristics of the wind farm.This part of research provides a practical solution for the dimensionality reduction of the power system security and stability analysis model.The research part of transient stability time domain simulation analysis based on data-model hybrid drive starts with the calculation process of time domain simulation method,and establishes a model framework driven by the hybrid of physical-driven model and data-driven model.To solve the problem of complex calculation of differential equations in power electronic components,this paper proposes a solution to build a data-driven model using PSO-BP neural network to replace the computationally complex modules.Aiming at the problem of small simulation step size,the Stride Euler method is proposed to make the data-driven model realize stride prediction.While ensuring the simulation accuracy,the differential equations of power electronic components with small time scales are integrated into a data-driven model with large time scales to further improve the computational efficiency.Finally,the improved IEEE 14 node model is used to validate the proposed algorithm and model,and the results show that the proposed method can improve the simulation speed while ensuring high accuracy.