This study focuses on Tongdu Town in Dongchuan District, Kunming City, and constructs a model for agricultural land use change based on human-environment interactions using an agent-based model. By utilizing census data, geographical environmental data, and field survey data, the study simulates the spatially distinct processes of agricultural labor, annual income of farming households, and agricultural land use changes in Tongdu Town from 2010 to 2015. The result shows that changes in agricultural land use are strongly correlated with labor, household income, and slope, but less so with soil fertility. Approximately 63.35% of agricultural land use changes occurred among low-income farming households, particularly the conversion of farmland to grassland — 87.60% of this specific conversion took place in low-income households. The transition from farmland to forest predominantly occurred in areas with relatively higher population densities and steeper slopes. The simulation model demonstrated excellent performance, with a spatial accuracy of 92.23%. The method developed for simulating agricultural land use changes in this study can offer valuable references for other regions, and thus provide scientific support for the implementation of ecological conservation, rural revitalization, and other related projects in China.