Taking real-time and effective preventive maintenance measures on in-service roads can reduce traffic congestion, eliminate potential road safety hazards, and greatly reduce road maintenance costs. Based on the Persson friction theory, this study first determines the evaluation indicators for the anti slip ability of asphalt pavement. Based on vehicle dynamics and contact friction between tires and road surface, the anti-skid thresholds of road surfaces for different road conditions and vehicle models are solved through simulation. This study utilizes Python and neural network algorithms to establish a decision-making model for anti-slip recovery technology of asphalt pavement. The experiment shows the model trained by Back Propagation neural network has high accuracy. The training accuracy of the model is stable at around 0.90, and the training loss value is around 0.34, which can be used for decision-making in anti slip recovery technology. When the speed is less than 60 km/h, the increase in the threshold of dynamic friction coefficient is significant. The maximum difference in growth rate is 47.9%. When the speed exceeds 60 km/h, the increase in the threshold of dynamic friction coefficient gradually slows down. At lower speeds, it is more essential to consider the variable value of the dynamic friction coefficient. When the speed is high, more consideration needs to be given to its reference value. This study provides a scientific basis for ensuring that the anti slip ability of the road surface always meets the requirements of driving safety, and has important engineering practical value, economic and social benefits.