In this paper, for constrained optimization problem, one multi-objective optimization algorithm that ensures both performance robustness and constraint feasibility is proposed when uncertainties are involved in design variables. In the proposed algorithm, the gradient index of objective function assisted by design sensitivity with the help of finite element method is applied to evaluate robustness; the reliability calculated by the sensitivity-assisted Monte Carlo simulation method is used to assess the feasibility of constraint function. As a demonstration, the performance and numerical efficiency of the proposed method is investigated through application to the optimal design of TEAM problem 22--a superconducting magnetic energy storage system.