Abstract: A structure temperature statistical method was proposed based on the higher-order moment method， which can avoid the uncertainty of the traditional fitting curve method to seeking a distribution as well as can reach a high computational efficiency and accuracy. Based on the long-term monitoring data of a high speed railway bridge in the east China area， by taking box girder and track as an integrated system， the representative values of temperature and its difference in vertical and horizontal direction were calculated. The temperature distribution nephogram and the temperature-difference distribution rules were analyzed. The covering effect of track on the temperature of the box girder surface was studied. Results show that the annual temperature sample of the girder-track system has a middle variability. The distribution is relatively stable with a high temperature standard values of 35.4℃ and 38.15℃ respectively. The largest vertical temperature gradients are 10.33℃ and 12.93℃ respectively. The vertical effective thermal conductivity length of the concrete is approximate 60cm， and the covering effect of the track on the box-girder surface temperature is 4.4℃ declined.

Key words: higher-order moment method, bridge engineering, railway engineering, temperature action representative value, high speed railway