- 城市智慧更新:前沿理论与方法
- 文章编号:1009-6000(2025)08-0019-08
- 中图分类号:TU984 文献标识码:B
- Doi:10.3969/j.issn.1009-6000.2025.08.003
- 项目基金:国家自然科学基金面上项目“基于要素流模拟的城镇低效用地智能识别与优化”(52478060);自然资源部2023年度部省合作试点项目“国土空间规划实施监测网络(CSPON)系统关键技术研发及示范”(2023ZRBSHZ061)。
- 作者简介:孙佳炜,南京大学建筑与城市规划学院,硕士研究生;
秦萧,通信作者,南京大学建筑与城市规划学院,副教授/特聘研究员、博士生导师,城市A I与绿色人居环境营造省高校重点实验室副主任,研究领域为人工智能与城镇低效用地优化;
张姗琪,南京大学建筑与城市规划学院,助理教授/特聘研究员、博士生导师,研究领域为城市社区感知与设施布局优化;
孟石,南京大学建筑与城市规划学院,博士研究生。
- 基于CNN的城市低效居住用地智能识别与协同更新路径研究——以南京花园路片区为例
- CNN-Enabled Identification and Collaborative Renewal Pathways for Inefficient Residential Land: A Case Study of Huayuan Road District, Nanjing
- 孙佳炜 秦萧 张姗琪 孟石
- SUN Jiawei QIN Xiao ZHANG Shanqi MENG Shi
- 摘要:
随着低效用地再开发与老旧小区改造的深入推进,低效居住用地的精准识别成为实践中亟待解决的问题。文章以南京市花园路片区为研究对象,通过构建区域—地块双尺度指标体系,融合空间聚类与深度学习技术,突破传统静态评价方法的局限性,提出“本体修复—邻域联动”的协同治理路径。结果表明,相较于传统方法,AI 技术对邻域功能组合特征的自动化提取与动态化识别显著提升了低效居住用地识别的科学性与系统性,该方法可为老旧小区更新提供技术与实践路径层面的支撑。 - 关键词:
低效居住用地;智能识别;深度学习;协同更新 - Abstract: Amid the advancement of low-efficiency land redevelopment and old residential neighborhood renewal, the precise identification of underutilized residential land has emerged as a critical practical challenge. This study focuses on the Huayuan Road area in Nanjing, establishing a dual-scale (regional and parcel-level) indicator system. By integrating spatial clustering and deep learning techniques, we transcend the limitations of conventional static evaluation methods and propose a collaborative governance pathway of “on-site restoration and neighborhood linkage”. Results indicate that AI-driven automated extraction and dynamic identification of neighborhood functional combinations significantly enhance the scientific rigor and systematicity of inefficient residential land identification compared to traditional approaches. This method provides methodological and practical support for the regeneration of aging neighborhoods.
- Key words: inefficient residential land; precision identification; deep learning; collaborative renewal