To deeply analyze the natural and social relationships in the complex flood system of Jincheng City, this study adopted aagent-based modeling approach to model heterogeneous social agents from both micro and macro perspectives, capturing the macro-level emergence of the system.Using system dynamics methods, it performed an evolutionary analysis of the flood resilience characteristics at the societal level, enabling a quantitative assessment of the flood’s impact on the social sector. The results show that as the rainfall return period increases, the pressure on the drainage system continuously rises, along with an increase in overflow and surface inundation areas. Under a 10-year rainfall return period, the flooded areas at the water depth of ＞0-0.15 m, >0.15-0.4 m, and ＞0.4 m are 2.03 km²,0.29 km², and 0.21 km², respectively. Under the global evacuation mode, the number of evacuated agents increases at a decreasing growth rate. Under the zonal evacuation mode, the growth rate of evacuated agents first increases and then decreases. The unit distance travel task indicator under different rainfall return periods shows local fluctuations, with an overall decreasing growth rate over time. The global evacuation mode exhibits a better balance between evacuation time and distance. The increment of urban flood resilience under different rainfall return periods shows an increasing trend followed by a decrease. The adaptive capacity increment for urban flood in heavy rain events shows an increasing trend, while the resistance capacity increment reaches its minimum value at the peak time and then gradually increases.