The residual stress resulting from welding was commonly disregarded in traditional fatigue life evaluation. This reduced the accuracy of life prediction and increased the challenges associated with optimizing the fatigue life of the shear wave seismic vibrator baseplate. This work introduced a modified fatigue life prediction model that considered the interaction between welding residual stress and load stress, based on the S-N curve. The study revealed the impact of welding process parameters on residual stress. A mathematical correlation was established between welding residual stress, welding speed and interlayer temperature. We developed a fatigue life optimization model by leveraging the local sensitivity method. The NSGA-II algorithm was ultimately employed to obtain the Pareto solution set. We determined the optimal fatigue life parameters using a combination of entropy weighting and TOPSIS. The optimized results showed a remarkable 17.72% improvement of fatigue life, the optimization method was proved to be accurate and efficient.