Research On Traction Load Statistics Method And Application Based On Data Mining

The development of electrified railway in China is accelerating.The load of electrified railway is widely distributed in the power grid,which is supplied by the power system.Traction load is a kind of special power load.It is very necessary for railway development and construction to ensure traction load stable and reliable.Especially for the power quality problems such as negative sequence current which is generated in the power grid.And the railway department needs to adopt the corresponding power management scheme.Therefore,it is necessary to study the statistical methods of new electrified railway load distribution with predictive effect.The result can provide a certain data foundation for the design of new traction substation and grid planning.In this paper,based on the existing measured data of a large number of traction loads,the data mining technology and probability theory and mathematical statistics method were used to obtain the traction load probability density distribution model.Combined with the incomplete ? function,the calculation method of 95% probability value was obtained.The calculation results were compared with the measured data and the error analysis was made.At the same time,the idea of Monte Carlo sampling was introduced,and the sample of traction load was generated according to the probability model,which provided theoretical data for the calculation of power quality when the electrified railway was connected to the power grid.According to the limitations of the ? distribution fitting statistical method,this paper selected the typical digital characteristics of traction load and the sample moments of each order as the boundary conditions.A plurality of existing traction loads were classified by selecting a fuzzy C clustering method based on index weights.From the similarity of each type of load probability histogram distribution and the cluster effective index value,the relative superiority of the clustering result based on index weights was determined.Thereby,a characteristic sample library of the traction substation load distribution model with a certain sample size which was matching with the new line traction load was formed.According to the application background of the energy storage device which can achieve “cut the peak and fill the valley” in the electrified railway,the power quality prediction and evaluation problem was transformed into the constraint problem between the traction transformer capacity and the short circuit capacity of the connected power system PCC.From the power quality index of three-phase voltage unbalance,the prediction idea of the capacity adaptation relationship between the electrified railway and the power system and the design of energy storage device capacity were discussed and summarized.Finally,taking a new line design scheme as an example,after obtaining the data of the full-day load sample from the load statistical forecast,the evaluation scheme of the power quality index when the substation is connected to the external power grid was given.Then the prediction and evaluation of the capacity of the energy storage device when the three-phase voltage unbalance exceeded the standard was carried out.The application of the statistical methods of the traction load in the power quality assessment and the design of the energy storage device capacity were realized.