Abstract. Due to the difficult of parameters calibration in existing three metro traction energy consumption models, this research first develops a gray related hierarchy analysis model to determine the main factors mainly considering the operational data. Furthermore, a traction energy consumption model based on neural network model is accordingly proposed to calculate the traction energy consumption of metro of one line due to statistics data which are gained by gray related hierarchy analysis model. It is found that the relative error of predicted values and actual values is a maximum of 8.61%, a minimum of 0.01% and the average relative error is 3.12% by using the operation data from one of Beijing subway lines. Results indicate that the model can predict traction energy consumption of a single metro line with high accuracy.
IntroductionThere are several factors can impact on power consume of train operation, such as driving tactics of the train, alignments of the track, attributes of the train, operation management of the rail transport system, natural environment and others [1]. The need for reasonably find out primary ones among them, and then correctly calculate the traction energy consumption becomes urging. As far, there are three existing energy calculation models of urban transit system which are: the regression analysis model based on data (MBD) [2-4], the calculation model based on electric power (MBE) [5][6][7] and the calculation model based on kinematic methods (MBK) [8][9][10][11]. Most of them aim at traction consume of traditional or high speed train but urban railway. The characteristics of urban railway traffic system are short distance between neighboring stations; frequently start/stop and relative low speed, and all these should be taken account of for its measure model of traction consumption.According to the characteristics of existing three kinds of metro traction energy consumption models, it is necessary to search a new model for calculating metro traction energy consumption. In this paper, based on factors analysis traction power consumes, mainly factors are ranked according to the gray relation. Moreover, measure model of urban railway system power consume based nerve net are set up, combining data from railway company. For later research, the precision of these models are well evaluated. This paper is organized as follows. Section 2 describes energy consumption factor and the mathematical expression of the traction energy consumption model based on neural network. In section 3, a numerical example is presented for the model. Section 4 summarizes the paper.