Abstract: To enhance the efficiency of asphalt pavement inspection, the pavement subsidence data from a specific section and station number of an asphalt road were taken as research targets. A fitting and prediction of the development of asphalt road subsidence on the highway was conducted based on the BP neural network. The results showed that BP neural network model can effectively predict road subsidence. The predictive accuracy of the neural network model was steadily improved with an increase in the training data set. Considering engineering efficiency and predictive accuracy, it was recommended to use 32 sets of data as the optimal sample size. The predictive accuracy of the BP neural network model was significantly higher than that of the quadratic curve method, with a relative error reduction of up to 5%. The study confirmed the feasibility and effectiveness of the BP neural network model in predicting the development of pavement subsidence, providing a new method for investigating the development of asphalt pavement subsidence on highways.