41 344 violation records collected over two months by electronic enforcement cameras in the study area were analyzed in order to uncover the spatiotemporal mechanism of illegal parking and propose targeted management strategy. Area of interest (AOI), point of interest (POI) and road network data were integrated to construct a built environment indicator system, encompassing land use, transportation supply and road design. A particle swarm optimization–extreme gradient boosting (PSO-XGBoost) model was applied to examine the relationship between parking violation and built environment factor across different time period. Model interpretation was conducted using Shapley additive explanation (SHAP). The PSO-XGBoost model achieved R² exceeding 0.90 for four key time intervals on weekday (8:00—9:00, 12:00—13:00, 18:00—19:00 and 22:00—23:00), outperforming traditional machine learning model. The impact of built environment factor varies across time period. The number of public transit route is most influential during morning and evening peak. Restaurant POI and school AOI are more influential during midday. Accessibility of off-street parking facility is most critical at night. SHAP interaction effect reveals nonlinear and threshold characteristic, indicating that some countermeasures lose effectiveness once specific range is exceeded. These findings highlight the need to develop time-specific and area-specific parking management strategy that account for built environment heterogeneity, thereby supporting refined governance of illegal parking.