Development Of High-order Harmonic Monitoring And Evaluation System For Traction Power Supply Network

In recent years,with the development of electrified railway,the problem of matching locomotive with traction network is becoming more and more prominent,and the AC-DC type electric locomotive has replaced the original AC-DC type electric locomotive to a large extent and become the main force in the traction network,although the operation of AC-DC type electric locomotive has significantly reduced the harmonic content in the traction network,its pulse width modulation control method will generate high-order harmonics injected into the traction network,resulting in resonance and amplification of high-order harmonics,which will not only cause harm to the electrical equipment and locomotive on the 27.5kV side,but also penetrate into the high voltage 110 kV or 220 kV through the traction transformer.The high-order harmonics will not only cause harm to the electrical equipment and locomotives on the 27.5kV side,but also penetrate into the high-voltage 110 kV or 220 kV network through the transformer and the low-voltage 380 V through the transformer used,causing damage to the equipment in the substation and seriously threatening the safe operation of the traction power supply system.This thesis firstly summarizes the current research status of interaction-locomotive coupling and high-order harmonics in traction interaction at home and abroad,and analyzes the propagation path of high-order harmonics of traction load in traction interaction,and explains the penetration mechanism of high-order harmonics in the whole traction power supply system,parallel conductor,traction transformer and 380 V low voltage side of traction substation;analyzes the hazards and key risks of high-order harmonics,and explains that high-order harmonics in the 27.5kV high-voltage side of the equipment and electric locomotive and the harm caused by the 380 V low-voltage side of the self-powered equipment;and for the problem of high-order harmonics transmission,the arrangement of high harmonic measurement points is proposed.Secondly,the high harmonic monitoring scheme of the self-power system in the traction substation is designed,and a complete simulation model of the traction power supply system and the self-power system is established through Matlab/Simulink simulation software to analyze the impact of high-order harmonics on the self-power system;based on the simulation and actual measurement data,the SVD algorithm is used to identify the harmonic impedance parameters of the self-power load;based on the simulation Based on the simulation results,we design the filtering scheme and embedded high harmonic monitoring device for the self-power system,and verify the filtering performance of the device.And the overvoltage identification method of high-voltage electrical system of rolling stock is studied based on the vehicle-mounted harmonic inspection device.An overvoltage image recognition method based on Shufflenet lightweight convolutional neural network is proposed.The B2 G algorithm is used to map six typical measured overvoltage waveforms in the traction interaction into grayscale images,which are input to the Shufflenet network for model training.The influence of the model parameters on the classification performance is investigated in four aspects: learning rate,sample batch size,network complexity and number of textures,and is also compared with six other shallow machine learning models.The experimental results show that the used method can identify overvoltage types quickly and accurately with a small data set,and the model has a strong generalization ability and reliable identification results.Finally,the hardware and software of the monitoring system are developed and validated.The hardware monitoring terminal is developed and designed based on a 32-bit ARM architecture embedded system,which is connected to the data transmission unit through RS485 communication protocol,and the data transmission unit carries out data transmission with the cloud platform through 2G(4G)/GPRS network,realizing the functions of transient waveform capture and storage,abnormal warning and positioning;the software cloud platform is developed and designed based on The software cloud platform is developed and designed based on ASP.NET environment,and adopts B/S architecture for visualization interface and related algorithm design,realizing the functions of data analysis and processing,abnormal event location capture and overvoltage identification,etc.;and the cloud platform is deployed using the principle of intranet penetration,realizing the functions of data transmission,sharing and multi-user access at the same time.