Study On Statistical Prediction Of Traction Load In Newly Built Electrified Railway

One of the important symbols of modernization is the high-speed railway,which is also the general trend of the development of the passenger traffic.It is the primary goal to ensure its safe and reliable operation in the construction.The construction of high speed railway also brings a lot of influence,especially in the power grid will produce negative effects such as negative sequence.Therefore,it is very important to study the influence of the new high-speed railway power quality,which can provide some reference for the design of traction substation and power network planning.This paper from the the large number of measured data of traction load,using the method of probability theory and mathematical statistics to process,summarize and analyze the distribution of feeder current,and further the sample data will be divided into two kinds of traction and regeneration conditions.The probability density fitted by beta distribution respectively on the two conditions of the current.Through the analysis of the fitting coefficients shows that when k?g(the charged effective coefficient)is less than 1.18.the fitting effect is good.Then,the power factor which can distinguish between traction and regenerative current is studied,and the law of power factor distribution under two conditions is summarized.Then,using the Monte Carlo sampling,sample data acquisition considering diurnal feeder current no-load,regeneration and the traction condition,and the measured value by comparing the sample values.The error between them is in the acceptable range,And it proves the feasibility of fitting distribution traction and regenerative current.According to the boundary conditions of design and use,the load forecasting model of the newly built railway line can be established according to the boundary conditions.The fitting method based on the distribution of the feeder current probability density function of the traction substation is convenient,simple and effective,but the statistical analysis shows that the fitting effect is not good when k?g(the charged effective coefficient)is more than 1.18.Considering the limitation of the beta distribution,according to the forecast theory,the traction load based on the massive measured data,analyzing its distribution in statistics,select the charged effective coefficient,maximum value,standard deviation,skewness as features,using fuzzy C means clustering algorithm,a plurality of traction substation classification by size,in order to establish the characteristics of probability model of feeder current of traction substation of electrified railway,according to the Bayes discriminant analysis method,in both looking for traction probability model and matching model of load probability library,so as to obtain the prediction model of the traction load the new electric railway,for the new high-speed railway traction load on the power grid can provide the basic data quality impact prediction...