Selective catalytic reduction (SCR) technology is the most widely used denitrification method in industry. However, the calibration of its kinetic model is very problematic due to the large number of parameters involved and the nonlinear characteristics, especially for the undesired reactions such as formation-decomposition of ammonium nitrate (AN), ammonium sulfate (AS), ammonium bisulfate (ABS), etc. In this paper, the synthetic gas test bench and the SCR system of WD10 marine diesel engine burning high-sulfur fuel oil (HSFO) and low-sulfur fuel oil (LSFO) are taken as the research objects. Firstly, based on the experimental data obtained from the synthetic gas test bench, the pre-exponential multiplier and activation energy of each reaction were identified using the Discrete-Grid algorithm. In this work, considering that the production of N2O, SO3, AN, AS, and ABS in the undesired reactions is very low, not on the same order of magnitude as that of NO and NH3, it is easy to overlook their reaction characteristics in the optimization process. We have adopted an approach that amplifies the error between the predicted concentration and the observed concentration of the species to solve this problem. Secondly, a detailed and complete SCR kinetic model was constructed. Finally, with the parameter identification results obtained from the synthesis gas bench as initial values, all the kinetic parameters of the WD10 marine engine SCR model was calibrated simultaneously using the Discrete-Grid algorithm. The results showed that the SCR kinetic model derived from the synthetic gas test bench was suitable for the real marine engine SCR system when the range of temperature and space velocity (SV) of the synthetic gas bench test is similar to that of the marine SCR system.