面向精准城市设计方案的时空活力数智推演方法OACHSSCD

As China's urban development shifts toward enhancing the quality of the existing built environment,fostering urban vitality has become a key objective of ur-ban design and renewal.However,existing urban vitality assessments mainly rely on post-development observations,limiting their ability to support the optimization of ur-ban design proposals during the planning stage.This study integrates multi-source ur-ban data with machine learning algorithms and proposes a computational framework for spatiotemporal urban vitality modeling.The framework combines fine-grained built environment data and Location-Based Service(LBS)data and employs a Long Short-Term Memory(LSTM)neural network architecture to capture spatiotemporal vi-tality patterns.Based on the design indicators of urban design proposals,this ap-proach enables the generation of block-scale vitality curves at an hourly temporal resolution.Furthermore,transfer experiments and practical applications are conducted to validate the method's cross-city adaptability and practical applicability.By introduc-ing a spatiotemporal vitality perspective into urban design evaluation,this study pro-vides a quantitative approach to support urban design optimization and vitality en-hancement.