智能交通管理、气候响应型城市规划以及具备灾害韧性的数字孪生等应用，体现了面向复杂动态系统的“感知–预测–适应/学习”循环在实践中的运行。本文考察了高性能计算（HPC）、深度学习和地理大模型（geographic large models）在构建反馈回路中的使能作用，并探讨了数据集成、模型可解释性与治理等长期存在的挑战。最后，我们提出智能地理学（IG）应被视作一个持续演化的社会技术生态系统，通过适应与自主学习，将空间数据转化为知识与行动力。

Applications such as smart traffic management, climate-responsive urban planning, and disaster-resilient digital twins illustrate the sensing–prediction–adaptation/learning cycle in practice for complex changing systems. We examine the enabling roles of HPC, deep learning, and geographic large models in implementing feedback loops, and address persistent challenges in data integration, interpretability, and governance. We conclude with a vision of IG as an evolving socio-technical ecosystem that through adaptation and self-learning turns spatial data int