Analysis And Prediction Of Power Quality In Urban Rail Transit Power Supply System

With the rapid development of urban rail transit, its interaction with the distribution network is increasingly strengthing, need to assess the power quality of subway lines having been put into operation, analyze the mechanism of power quality problems and establish the power quality prediction system of a new subway line. This paper focuses on current harmonics, voltage fluctuation detection algorithm in the condition of small time granularity and power quality prediction method under the background of large data to start a more in-depth research, in order to provide a theoretical basis for the making of power quality control measures and the design of traction power supply scheme.Build a transient simulation model of two way feeding traction power supply system based on MATLAB/Simulink, construct the traction substations and traction network model, and use a variable nonlinear impedance to accurately simulate the change of traction network impedance caused by vehicle operation. Analyze the different operating characteristics of vehicle in traction conditions and brake operating conditions, build a vehicle drive control system. Based on the above model, achieve a whole process dynamic simulation of single driving-car between stations, and then analyze the mechanism of urban rail power quality problems.Based on the particularity of urban rail transit harmonic current, consider of the advantage of wavelet packet transform (WPT) in band division and capability of Hilbert-Huang transform in extracting instantaneous parameters, propose an improved Hilbert-Huang transform (WPT-HHT) based on the preprocess of WPT in order to analyze harmonic current’s time-varying characteristics. Study the selection principle of WPT function and decomposition levels, analyze the improvement of mode mixing by WPT pretreatment compared with the traditional HHT method. In addition, introduce support vector machine regression based on particle swarm optimization (PSO-SVR) to extend signal endpoints, which avoiding the traditional HHT end effect problems to some extent. Lastly, apply the WPT-HHT method to analyze harmonic current’s time-varying characteristics during the traction progress of mass transit vehicles based on24-hour recorded data of one traction substation of Beijing subway line1.Considering of the urban rail vehicle traction process takes short time, in the analysis of voltage fluctuations, flicker index based on statistical calculations is no longer applicable, and propose ideas of using instantaneous d-q transformation to extract the voltage flicker envelope for the precise description of voltage fluctuation details. Analyze the influence of signal frequency fluctuation on the calculation accuracy of instantaneous d-q transformation method, and draw up the error formula in case of frequency fluctuation. Propose error control method based on filtering ud2+uq2, and verify with simulation based on MATLAB/Simulink. The results show that:improved algorithm has high precision, meet the needs of engineering calculations.Propose power quality prediction method of new urban rail transit line based on data mining technology, use gray correlation analysis to determine the power quality factors, use agglomerative hierarchical clustering algorithm to classify the subway lines having been put into operation, then use classification and regression tree (CART) for new line type matching. Finally, complete sample training and prediction by SVR.