Impact-type penetrators are devices that apply the impact generated by their internal components to penetrate soil. The penetration effect of impact-type penetrators is affected by the physical parameters (e.g., mass and stiffness) of their internal constituent elements. Therefore, optimal parameters must be obtained by using a dynamic impact penetrator model to maximize the dive distance of each impact. However, the dynamic impact penetrator models are nonlinear and difficult to describe. Thus, in this paper, this work proposes a segmentation method for modeling the penetrator motion to establish an accurate dynamic model that can be divided into four states. Buffer spring pre-compression, which is introduced as a new influencing parameter to improve the performance of the penetrator, and the genetic algorithm is used for optimization in accordance with the characteristics of the required optimization parameter set. Parameter stability is then analyzed by considering the actual project application. Then, the control variable method is employed to explore the influence of changing the obtained parameters on the penetration effect. Finally, a processing prototype designed on the basis of the acquired parameters is used for experimental verification. This work addresses the complexity of the dynamics model of penetration and the difficulty encountered in determining parameter values.