With the technical progress in carbon dioxide （CO ₂） flooding for enhanced oil recovery （CO ₂-EOR），it is of great significance to study the influence mechanism of CO₂ injection on the viscosity of crude oil. This study systematically investigated the synergistic effects of temperature （20-40 ℃），pressure （0.5-3.5 MPa），and water content （0%-60%） on the viscosity of saturated CO ₂   dissolved gas crude oil through laboratory experiments. The results showed that under constant pressure，for every 5℃ increase in temperature，the viscosity decreases by approximately 60%-80%. At a constant temperature，for every 0.5 MPa increase in pressure，the viscosity decreases by approximately 1%-5%. In terms of water content，a 10% increase in water content leads to a rise in viscosity by 15%-50% before the inverted point （water content < 40%）. After the inverted point （water content>40%），for every 10% increase in water content，the viscosity decreases by 50%-80%. Based on the experimental data，the prediction performance of three classic models，namely Standing，Glaso and Vazquez & Beggs （V&B），was evaluated. It was found that the V&B model had the best robustness. The XGBoost algorithm combined with Bayesian optimization was used to dynamically correct the V&B model，and a multi-parameter mapping relationship of temperature，pressure and water content was established. The average relative error of the model was reduced from 20.1% to 6.9%，and the average absolute error was decreased by approximately 66%. This study provides a theoretical basis and intelligent prediction tool for the process design and dynamic optimization of oil and gas gathering and transportation systems under CO₂ flooding conditions.