In the context of carbon neutrality， brine water viscosity directly influences the CO₂-brine water two-phase flow process in the reservoir when the technology of CO₂ storage in brine water reservoirs is used to achieve carbon reduction targets. Currently，the viscosity prediction method based on the pressure effect still needs to be improved. In this study， the existing calculation methods were optimized by using the least squares method， BP neural network， and BP neural network based on a genetic algorithm.The brine water viscosity was taken as a binary function of temperature and molar concentration， as well as a ternary function of temperature， molar concentration， and pressure， respectively， and an optimized model of viscosity prediction considering the effect of pressure was established. Then， the optimal prediction was obtained， and the effect of viscosity on flow was systematically analyzed based on the level-set method of COMSOL software. The results show that the existing empirical formula could be optimized using the least squares method， but the effect is not apparent. After considering the pressure， the prediction accuracy can be improved by 45.2% using the binary BP neural network and by 57.32% using the binary BP neural network. Therefore， in the case of sufficient experimental data， reliable brine water viscosity values can be obtained in a wide range of pressures based on the BP neural network model. In the absence of the corresponding experimental data， the brine water viscosity values can be obtained by the empirical formula method because the empirical formula method can predict the trend of viscosity change. In addition， it can be found that the viscosity affects the distribution of the fluid in the flow channel through the simulation results， which in turn affects the flow velocity. A larger viscosity ratio indicates a minor fluctuation of the average velocity at the outlet and faster stability， and a larger viscosity ratio indicates smaller residual water saturation， which is more favorable for the replacement process， and the two are in a logarithmic function of the relationship.