Abstract: The service performance of a community life circle depends on understanding how residents' heterogeneous activities interact with facility layouts. Current research has shifted from static spatial analysis toward dynamic spatiotemporal activities, micro-level individual projects, and behavioral decision simulation. Yet it still faces limitations in quantifying individual preferences and systematically representing decision dimensions; the interaction between heterogeneous individuals and space, as well as bottom-up emergence mechanisms, remains insufficiently clarified. Taking the project-activity system of time geography as its theoretical prototype, this study proposes an integrated framework combining the mixed Logit model with agent-based modeling. It explains a three-stage model construction and operating framework: configuring resident and spatial agents with both dynamic and static attributes; designing and estimating decision mechanisms that reflect individual preference heterogeneity; and conducting ABM simulation that integrates planning, execution, and feedback. In this way, the statistical distribution of residents' preference heterogeneity is embedded in an agent simulation model, creating a feedback loop between theoretical modeling and practical validation. Four application scenarios are proposed: multidimensional dynamic evaluation of life-circle service performance; attribution analysis of interaction mechanisms between people and life circles; spatial polysemy shaping and scenario construction based on activity simulation; and scenario prediction and intervention assessment for life- circle development. The framework provides a computable model for analyzing the complexity and heterogeneity of human-space interaction in community life circles and supports human-centered, precise renewal governance.