交通运输工程与信息学报

【背景】步行环境作为城市交通系统的重要组成部分，其质量直接影响居民的出行体验和常规公交、轨道交通等方式的选择。然而现有相关研究未充分考虑步行环境对居民出行意向的定量影响，步行环境的量化也缺乏主客观指标的共同表征。【目标】针对现有研究的不足，本文旨在深入分析步行环境对城市居民出行方式选择的影响机制。【方法】通过引入主观规范、态度、步行环境等9个潜变量以更准确地刻画不同出行方式选择行为的心理决策过程，创新性地应用图像识别技术与个人主观感知指标共同表征步行环境变量，并采用结构方程模型刻画各心理潜变量间的相互关系，以及9个潜变量与其对应的测量变量之间的关系，进而构建融合步行环境影响的出行方式选择行为混合模型，以提升模型对出行者决策行为的预测精度和解释力。【结果】以重庆中心城区的出行者为实例分析对象，结果表明，步行环境对步行方式选择的影响最大（0.665），其次是对常规公交（0.663）、轨道交通（0.384）和小汽车（0.304）的影响。此外，考虑步行环境等潜变量的SEM-MNL模型的拟合优度由0.275 2提升至0.454 4，表明引入步行环境的模型解释性更强。【应用】研究结果可加深对城市居民出行方式选择行为的认识，为改善城市步行环境、提高居民绿色出行意愿奠定良好的理论基础。

[Background] As an important component of urban transportation system, the quality of pedestrian environments directly affects residents' travel experiences and choices of conventional public transportation, rail transit and other modes. However, existing research has not fully considered the quantitative impact of pedestrian environments on travel intentions, and the quantification of pedestrian environments lacks a common representation of subjective and objective indicators.[Objective] In response to the shortcomings of existing research, this study aims to deeply analyze the impact mechanism of the pedestrian environment on urban residents' travel mode choices.[Methods] By introducing nine latent variables such as subjective norms, attitudes, and walking environment to more accurately depict the psychological decision-making process of different travel mode choice behaviors, innovative image recognition technology and personal subjective perception indicators were applied to jointly characterize the walking environment variables. Structural equation modeling was used to characterize the interrelationships between various psychological latent variables, as well as the relationships between the nine latent variables and their corresponding measurement variables. A mixed model of travel mode choice behavior considering the influence of the walking environment was constructed to improve the predictive accuracy and explanatory power of the model regarding the decision-making behavior of travelers. [Results] Considering travelers in the central urban area of Chongqing as an example, the results showed that the walking environment had the greatest impact on the choice of walking mode(0.665), followed by the impact on conventional public transportation(0.663), rail transit(0.384), and cars(0.304). In addition, the goodness of fit of the SEM-MNL model considering latent variables such as the walking environment increased from 0.2752 to 0.4544, indicating that the model incorporating the walking environment had stronger interpretability. [Application] The research results can deepen our understanding of urban residents' travel mode choice behavior, improve the urban walking environment, and lay a good theoretical foundation to encourage eco-friendly travel.