A one-dimensional multi-population biofilm mathematical model was developed to evaluate the mechanisms by which chemical oxygen demand (COD) influences nitrogen removal performance and microbial community in partial nitrification/anammox (PN/A) process. Simulation results demonstrated that influent COD concentration significantly affected the spatial distribution of functional microorganisms: under low COD conditions (<50 mg/L), ammonia oxidizing bacteria (AOB) predominantly colonized the outer layer of granules, while high COD levels (>150 mg/L) promoted heterotrophic bacteria (HET) as the dominant population. The study further revealed that moderately increasing the influent C/N ratio expanded the optimal dissolved oxygen (DO) operational window and enhanced the nitrogen removal contribution of denitrification by 20–30%, enabling stable total nitrogen removal efficiency exceeding 85%. By refining the stoichiometric matrix of the model, precise tracking of nitrogen oxide transformations across different metabolic pathways was achieved. Quantitative analysis confirmed strong spatial correlation between high-activity zones of anaerobic ammonium oxidation bacteria (AnAOB) and peak system performance (total nitrogen removal >80%), validating anammox as the core nitrogen removal pathway. This study provides critical theoretical guidance for optimizing PN/A process parameters, particularly elucidating the synergistic regulatory mechanism of influent COD and DO concentrations for achieving stable and efficient operation.