Depth of weld penetration and heat affected zone width are very important to laser welding quality.Laser welding is a complicated process,and quantization analysis of this process is quite difficult.In this work,a set of TC4 titanium alloy thin plate specimens were used as laboratory samples.Two radial basis function neural network (RBFNN) models were used to predict weld penetration depth and heat affected zone width.In order to minimize the heat affected zone width and maximize the depth of weld penetration,the above two neural networks were used as objective functions of multi-objective optimization algorithm.A simulated annealing algorithm is used to find the optimal solution within non-inferior solutions of the multi-objective optimization algorithm.The results show that the heat affected zone width and the depth of weld penetration are well balanced by this method.