基于道路设施空间耦合场景图的农村公路速度感知差异OA

To reveal the formation mechanism of driver speed perception differences under the multi-factor coupling of"human-vehicle-road-environment"in road-facility spaces,this paper constructs a road-facility spatial coupling scenario graph and integrates a graph attention network with a frequent subgraph mining algorithm to achieve interpretable prediction and mechanism analysis of driver speed perception differences.The proposed scenario graph consists of four modules:visual road geometry,visual facility environment,driver awareness state,and vehicle motion state,which jointly represent multi-source subjective and objective data and their topological relationships.Based on the scenario graph,a graph attention network is trained to predict driver speed perception differences and output the attention scores of different nodes,thereby identifying key factors affecting these differences.The results show that the proposed model achieves a prediction accuracy of 90.71％,outperforming several commonly used comparison models.This indicates that the road-facility spatial coupling scenario graph can effectively characterize the formation process of driver speed perception differences.Furthermore,key nodes are selected based on attention scores,and a frequent subgraph mining algorithm is employed to identify typical subgraph patterns with favorable coverage and support in scenarios where drivers underestimate or overestimate their actual driving speed.The results reveal that the nodes and attribute combinations across the different modules differ significantly between speed overestimation and underestimation scenarios.Driver speed perception differences are not determined by a single factor,but are jointly shaped by the interaction between external features,such as road geometry and facility environment,and dynamic factors,such as driver awareness state and vehicle motion state.Specifically,external road environment features primarily shape drivers'initial judgments of perceived speed,while driver awareness state and vehicle motion state further regulate the accuracy of speed perception,ultimately leading to individualized speed perception outcomes.This paper provides a theoretical basis for improving the accuracy of driver speed perception and optimizing rural road facility design and safety management.