Research On Identification Algorithm Of Atypical Load State Of Metro Traction Power Supply

As the best business card to show the comprehensive strength of the city and an important tool to solve the problem of urban traffic pressure,the metro must guarantee the operation safety of its power supply system through technological innovation.The DC traction network is the only power component in the metro traction power supply system that is not equipped with backup,and has higher reliability requirements for its feeder protection.However,the atypical load current caused by unknown causes frequently causes the metro traction network feeder protection to malfunction,and even causes major traffic accidents such as prolonged outages.In order to make the existing feeder protection have the function of avoiding atypical load current,this paper takes the metro traction power supply system as the engineering background,and uses the system model and current waveform analysis to analyze the atypical load generation mechanism and feature differences.This paper proposes a method for identifying atypical loads of metro traction power supply based on fractal correlation dimension,and combines it with recorded data for experimental verification.The main research is as follows:First,the power supply method and main structure of the metro DC traction power supply system are described and analyzed in detail.Based on this,the traction power supply system is simulated and modeled to realize the simulation of start-up conditions and fault conditions under bilateral power supply conditions.Complete the cause analysis of the atypical load oscillation current.Secondly,first distinguish the atypical load oscillation current and fault current from an intuitive perspective,and then use EMD analysis and full-wave Fu algorithm to analyze the waveforms of atypical load current and fault current,and find the differences in waveform characteristics.Thirdly,the conditional entropy method is used to jointly determine the two important parameters of the associative dimension method,the embedding dimension and the delay time,and the best combination value is selected.The fractal dimension of the starting state,atypical load state and fault state is calculated by the correlation dimension method.The fractal dimension range of different states is analyzed through a large number of experiments.The setting value of the fault criterion is determined to realize the state recognition.Then use the GUI to display the algorithm visually.Finally,the real-time digital simulation system RTplus and the DC protection measurement and control device KF1305 were used to perform semi-physical verification of the associated dimension state recognition algorithm.The experimental verification proves the feasibility and effectiveness of the correlation dimension state recognition algorithm in state recognition.Thirdly,based on the completed simulation model of the traction power supply system,the cause analysis of the frequently occurring oscillating current in the system is performed.The conditional entropy method is used to jointly determine the embedding dimension and delay time of two important parameters in the correlation dimension method,and the optimal combination value is selected.The fractal dimension of atypical load state and fault state is calculated by correlation dimension method,and the range of fractal dimension of different states is analyzed through a large number of experiments.After the setting value of the fault criterion is determined,the algorithm is displayed using the GUI.Finally,the real-time digital simulation system RTplus and the DC protection measurement and control device KF1305 were used to perform semi-physical verification of the associated dimension state recognition algorithm.The experimental verification proves the feasibility and effectiveness of the correlation dimension state recognition algorithm in state recognition.