To optimize the performance of the soft actuator, numerical experiments and regression analysis of the multi-structure parameter are presented based on the numerical calculation method. Firstly, numerical experiments and regression analysis were performed on the parameters of different driving channel shapes, including equilateral triangle, square, regular pentagon, regular hexagon, and circle, to determine the driving performance under different channel cross-sections. Secondly, numerical experiments and regression analysis were conducted on the parameters of fiber winding diameter and winding angle of the soft actuator, and qualitative analysis was performed on the failure states of the numerical calculations. Lastly, numerical experiments and regression analysis were carried out on the parameters of wall thickness and driving channel length of the soft actuator to determine the influence of these parameters on the driving performance. Through the numerical experiments and regression analysis, this study provides a necessary theoretical foundation for the optimization design of soft actuator, effectively improving the development efficiency and dynamic performance of soft actuators.