Abstract: This article proposed driving curve algorithm for heavy haul train (HHT) based on improved BP Neural Network (NN), used the actual train driving data for NN learning, described the nonlinear characteristics when the train braked, set the NN model of train braking operation process. The decompression target value of train braking and release time were used for the calculation of motion equation to obtain the driving curve for HHT. By comparing the curve with the real data from the Shuo-Huang Heavy Haul Line, the results showed that the method was effective.