Research On Fault Classification And Location Of Dc Line In MMC-HVDC System Based On VWAA-SVM Algorithm

In recent years,with the transformation of the global energy structure,China has taken the lead in putting forward the strategic goal of "double carbon" energy,which has been appreciated and supported by many countries around the world.With the continuous development of science and technology,power electronics technology has also entered a stage of rapid development.Modular Multilevel Converter based High-voltage Direct Current(MMC-HVDC)can improve the efficiency and reliability of transmission.MMC-HVDC is widely used in overhead transmission lines with long distance and high voltage,which can realize the interconnection of multiple AC / DC systems across regions.However,the failure of MMC-HVDC system will affect the reliability of power supply.At present,the research on DC transmission lines is mainly focused on post-fault protection and control,but there is little research on fault location and fault classification on DC transmission lines.the ability to quickly locate fault distance and identify fault types is very important for the safe and stable operation of power system.Therefore,this thesis mainly studies the line fault location and fault classification of MMC-HVDC system.First of all,this thesis summarizes the topology and operation principle of MMC converter,establishes the mathematical model of MMC,and then studies the capacitor voltage sharing strategy of sub-module SM and the modulation mode of converter station.The simulation model of double-ended MMC-HVDC system is built on the PSCAD/EMTDC simulation platform.The steady-state operation is studied and analyzed,which provides a research basis for follow-up fault location and fault classification.Secondly,a DC line fault classification method based on line fault traveling wave signal feature recognition is proposed in this thesis.All kinds of line faults are set at different locations of the line,and the vector weighted average algorithm is used to optimize the support vector machine model in the MATLAB simulation platform.Then the support vector machine is used to analyze the line fault voltage traveling wave signal,and the experimental results show that the recognition accuracy of all kinds of faults on the line is high and the classification effect is obvious.Finally,various fault types and fault characteristics of HVDC transmission lines in MMC-HVDC system are analyzed and studied.The traditional traveling wave fault location accuracy and reliability are affected by many factors,and the transition resistance will also affect the accurate calibration of the wavefront.Based on the time-frequency domain characteristics of fault traveling waves,a combined fault location method for two-terminal faults on MMC-HVDC transmission lines is proposed,which combines the natural frequency method and the traveling wave method.Firstly,the natural frequency method is used to measure the distance roughly in the first step,and then the rough location range of the fault point is calculated.In the second step,Ensemble Empirical Mode Decomposition(EEMD)algorithm is introduced to decompose the signal and calibrate the wave head.A more accurate location of the fault point is calculated from the double-ended fault location formula.Through the comparative experiments of various fault types,the results show that the fault combination location method eliminates the influence of wave velocity on accurate location and effectively identifies the traveling wave head,which further improves the accuracy of fault location.