Multi-objective optimization design of flexspline structure based on NSGA II for harmonic drive

For harmonic reducer single performance optimization, it will sacrifice other performance. The adverse effect on other properties is difficult to estimate. In this paper, a novel approach is proposed for harmonic reducer performance optimization. Considering the transmission efficiency, stiffness and stress in thin-walled cylinder, the nondominated sorting genetic algorithm II is used to solve the Pareto-optimal front. The thickness of the flexspline meshing point, the width of gear, the thickness and length of the flexspline are used as variables. Within the room temperature, rated load and speed, the transmission efficiency, stiffness and stress are used as objective functions. In order to keep the assembly dimensions unchanged, the dimension of the circular spline does not change. The optimal combination of structure parameters is explored at different weights. Under different weights, the efficiency can be improved by 18.047 %, or the stiffness can be improved by 23.778 %, or the stress can be reduced by 10.682 %. The thickness and length of flexspline are not significantly connected to the magnitude of the load. The length and the width are more affected by the load. Different working conditions have different requirements on the performance. The research can provide different design solutions for the optimization of harmonic drive under different working conditions.