Abstract: This paper discussed the deficiencies existed in the current pavement performance prediction methods and establishes a new performance prediction model based on the neural network theory with back propagation (BP) algorithm. The model uses the temperature, humidity and traffic volume as the influence factors and rate of pavement damage, deflection as the pavement performance prediction indices. This model was applied to five existing roads which have extensive data collected. The results are satisfactory.

Key words: Neural Network, Back Propagation Algorithm, Pavement Performance