Evaluation research on asphalt pavement friction performance based on convolutional neural network

Addressing the evaluation of asphalt pavement friction performance, a friction coefficient prediction model was established by collecting friction coefficients from 60 road sections using a pendulum friction tester and analyzing crack quantity statistics. Research indicates a significant positive correlation between pavement crack quantity and friction coefficient. A convolutional neural network was employed for friction coefficient prediction, with genetic algorithm optimizing initial network weights and thresholds, and simulated annealing algorithm's objective function serving as fitness function, effectively improving prediction accuracy. Experimental results show that the error rate between improved predictions and measured values was controlled within 3%, significantly enhancing prediction accuracy. The research findings provide important theoretical basis for asphalt pavement maintenance and intelligent driving systems. The friction performance of asphalt pavement is a crucial indicator of road safety, directly affecting vehicle stability and braking performance. With the rapid development of intelligent driving technology, higher requirements have been placed on the accurate evaluation of pavement friction coefficients. Traditional friction coefficient measurement methods face challenges such as low efficiency and poor real-time performance, making it difficult to meet the immediate demand for road condition information in intelligent driving environments. The development of deep learning technology has provided new approaches to solving this problem, particularly the advantages of convolutional neural networks in image recognition and feature extraction, offering effective tools for pavement condition assessment.