Abstract: In order to obtain the optimal parameters of a new elliptic-gear & crank-rocker weft insertion mechanism,optimization mathematical models of the mechanism are established,its objective function is chosen as the minimum fluctuating acceleration of gripper head andi″sosceles trapezoid shaped″acceleration of gripper head in a mation loop,and its constraint equations are the formulas which can meet the requirements of inserting weft and the kinematic behaviors of the mechanism.The models are high-dimensional nonlinear multi-objective models with constraints.Combining adaptive infeasibility degree and elitist non-dominate sorting genetic algorithm(NSGA-Π),an improved algorithm is proposed.And an aided-optimization software was compiled.A group of optimal parameters is obtained,and the kinematics performances of the mechanism with the optimal parameters are analyzed and compared with the original.The results show the algorithm is perfectible,which can solve nonlinear multi-objective optimization question with constraints effectively.