Abstract: In order to investigate the variation of residual stress along the depth after grinding, the grinding process of spiral bevel gear was analyzed. a randomly distributed multi-abrasive particle grinding model for spiral bevel gears was established using the virtual grid method, and a single factor simulation scheme was designedd. Firstly, a three-factor four-level orthogonal experiment was devised. Secondly, the grey relational analysis method was utilized to delve into the effects of grinding parameters on residual stress and obtain the optimal combination of grinding parameters. Finally, a prediction model between the degree of grey relational analysis and grinding parameters was established, and its accuracy was verified. The research findings indicate that the ground gear surface exhibits a residual compressive stress state, with the maximum compressive stress located in the subsurface layer. As the depth increases, the residual compressive stress gradually transitions to tensile stress and finally tends to zero at deeper levels. The effect of grinding parameters on residual stress is ranked as a_\mathrmp (grinding depth) > v_\mathrms (grinding speed) > v_\mathrmw (feed rate). With the increase of these parameters, the residual stress on the gear surface increases to varying degrees.