Research On Parameter Identification And Online Correction Methods Of HVDC Transmission System

With the construction of high voltage direct current(HVDC)system and large-scale integration of new energy sources,the proportion of electronic equipment in modern power system increases rapidly.The complex dynamic characteristics of modern power grids promote the development of simulation analysis methods to real-time simulation and online dynamic security analysis(DSA)systems,which require higher accuracy of simulation models and parameters.At present,the power system model parameter identification technology for traditional equipment such as synchronous generator,excitation system,regulator,and power load is mature.However,the parameter identification methods of power electronic equipment,such as HVDC transmission system model,especially online identification methods,needs further study.This paper studies the parameter identification method of the electromechanical transient model of the HVDC transmission system.The main work is presented as follows:1)Offline dominant parameters identification of HVDC model.Through analyzing the structure and parameter sensitivity of the HVDC model,a method for HVDC model dominant parameters searching is proposed to reduce the dimension of the parameters to be identified and improve the identification efficiency.For the equipment model and controller model of HVDC model,the offline identification methods of dominant parameters based on least squares method and particle swarm optimization algorithm are respectively proposed to realize the offline HVDC model parameters identification.2)Online parameters identification of HVDC model based on matching-correction method.Through a two-way classification algorithm,a typical HVDC model database is constructed based on historical and offline simulation data.Based on it,an online HVDC typical parameter matching method is proposed,and the numerical gradients is utilized to further correcting the typical model parameters to improving accuracy.3)Online parameters correction of HVDC model based on deep reinforcement learning.Through analyzing the basic principles of deep reinforcement learning and parameter identification,an online power system model parameter identification framework is first proposed,and then a deep deterministic strategy gradient algorithm based method is proposed to realizing the online HVDC model parameter correction.