Human activities critically influence the dynamics of urban functional organization. Understanding these behavioral impacts is essential for effective delineation of urban functional zones (UFZs) in contemporary urban planning and management. Conventional UFZ identification methodologies, predominantly focused on static spatial and structural characteristics, inadequately capture evolving human-land interactions. To address this limitation, we propose an integrative framework combining crowdsourced behavioral data with agent-based modeling to simulate functional transformations. Our approach adopts a bottom-up perspective, employing self-organizing maps (SOM) and K-medoids clustering with morphological tessellation units, thereby better aligning with the self-organizing characteristics inherent to urban systems. The effectiveness of this method in modeling the impact of dynamic human activities on urban function is demonstrated through the Wuhan case. The results show that human activity modifies urban functions; however, its effect is not decisive and exhibits a trend of initial fluctuations followed by stabilization, with mix-functional zones demonstrating greater stability to behavioral impacts. Although UFZ delineation changes substantially after simulation, the proposed approach identifies unique UFZs that remain sensitive to human activity variations. This human-land interaction modeling method offers urban planners a novel analytical tool to understand and plan urban functions supporting sustainable urban development.