A hybrid fuzzy and convolutional neural network framework for urban road traffic risk and sustainability assessment

一种用于城市道路交通风险和可持续性评估的混合模糊卷积神经网络框架

阅读:1

Abstract

Urban redevelopment is essential to improving sustainability and livability, but traffic congestion is still a problem, not only during rush hour but also during construction, accidents, and other interruptions. Existing research frequently concentrates only on optimizing traffic flow, with little consideration given to stakeholder-driven viewpoints or environmental and anthropogenic risk concerns. Although Geographic Information Systems (GIS) have great promise for gathering and evaluating vast amounts of spatial data, little is known about how they might be used in comprehensive frameworks for traffic risk assessment. To fill these shortcomings, this study suggests a brand-new hybrid framework for evaluating traffic sustainability that combines GIS with a Convolutional Neural Network (CNN) model and a Multi-Criteria Decision-Making (MCDM) technique. By (i) using a fuzzy Delphi method to systematically weight various environmental and anthropogenic criteria, (ii) utilizing oversampling methods to address data imbalances in risk prediction, and (iii) embedding CNN-based modeling in an interactive GIS platform to produce fine-grained, stakeholder-relevant risk maps, the framework significantly improves on previous research. The usefulness of the framework is demonstrated by a case study conducted in Foshan City, Guangdong Province, where road collapse segments were divided into five risk levels. Of these, 7% were categorized as high risk and 4% as extremely high risk, with the majority of these segments being in the eastern and southeastern regions. The training and testing accuracies of the Fuzzy-CNN model were 0.989 and 0.982, respectively, indicating high predictive performance. For municipal governments and urban planners, the generated traffic risk map offers a scalable and data-driven decision-support tool. This research tackles major shortcomings of current models in balancing sustainability, risk management, and urban resilience and advances the state of the art in GIS-based traffic risk analysis by explicitly integrating biodiversity and environmental elements into traffic planning.

特别声明

1、本页面内容包含部分的内容是基于公开信息的合理引用;引用内容仅为补充信息,不代表本站立场。

2、若认为本页面引用内容涉及侵权,请及时与本站联系,我们将第一时间处理。

3、其他媒体/个人如需使用本页面原创内容,需注明“来源:[生知库]”并获得授权;使用引用内容的,需自行联系原作者获得许可。

4、投稿及合作请联系:info@biocloudy.com。