The performance of servo actuators is significantly affected by factors such as friction and backlash. Simplified analytical models have difficulty meeting the increasing demands of control systems. Therefore, this study proposes a new data-driven method for modeling nonlinear factors. Through experimental verification, the newly proposed modeling method significantly improves the accuracy of nonlinear models. This achievement provides strong support for the precise control of servo actuators.