Abstract: In order to determine friction welding technological parameters correctly and quickly, an optimization model for friction welding technological parameters based on least square support vector machine (LSSVM)and artificial fish-swarm algorithm (AFSA)was presented.With three friction welding technological parameters such as fiction pressure, upset pressure and fiction temperature as optimization parameters, welding joint tensile strength as optimization object, the nonlinear mapping relation between optimization parameters and optimization object was fitted by LSSVM.Firstly, experiments were taken to get data samples, and LSSVM model was established through data samples above.Then, the model was optimized by AFSA to get welding technological parameters.The results show that the construction model is easy, the optimization solution is quick, and the parameters optimized by this method make welding joint tensile strength increase by 2.1% comparing with that parameters optimized by orthogonal regression method.