The injection molding filling process is a complicated non-steady-state, non-isothermal non-Newtonian fluid flow process and heat transfer process. With the development of polymer melt pressure, temperature, shear rate and other physical quantities during the mold filling process will affect the melt rheological properties. This research combines the injection molding visualization device and analysis method to collect real-time online parameters of the co-injection self-reinforced parts in the mold under different molding conditions. The distribution and change data provide an effective means for better characterizing the rheological properties of the polymer in the mold and the theoretical analysis of the mold filling rheology. This article starts with the basic equations of viscous fluid mechanics, draws on the establishment of the mathematical model of the filling process of the conventional injection molding, corrects and resets the basic assumptions and boundary conditions, and establishes a flow that can reflect the melt flow at the slender scale. The basic equation of scale effect is made in the process, and it is substituted into the simulation software for simulation analysis. The results show that there is no deviation in the overall structure of the model, but the parameters in the model should be variable factors that change with the change of the melt molding parameters. Therefore, we made a second correction to the variable parameters of the in-mold co-injection self-reinforced cross-scale viscosity model. Although there is a certain difference (12.62%) between the viscosity results obtained after the simulation and the measured values, it is compared with the conventional model deviation value of the model (101.25%) has been better improved, which can better achieve the purpose of simulation prediction. •A constitutive model of melt-filled in-mold co-injection molding considering the influence of micro-scale effects is established.•Compared with the conventional micro-scale viscosity model, the average prediction error of the size factor model for the IMSR-SPCs melt viscosity is reduced by 14.9%.•We have corrected the variable factors of the in-mold co-injection cross-scale viscosity model, and there is a difference between in comparison with conventional model parameters deviation had a better improved.•Through the visualization experiment, the flow speed of the reinforced melt is faster, the pressure loss is reduced and the temperature decrease is lower than that of the metal flow runners.•The plasticizing temperature has the greatest influence on the enhanced melt viscosity and the injection speed has the least influence.