A dynamic identification method for joint robot based on better student psychology based optimization(BSPBO) algorithm was proposed.The dynamics model of an articulated robot was established using Newton-Euler method,and a fifth-order Fourier series conforming to the motion constraints was designed as the excitation trajectory；the SPBO algorithm was introduced and improved as follows:increasing the proportion of Good Student classification in the student population,improving the treatment of solving variables out of bounds in the iterative process,in order to improve the exploitation ability and global exploration ability of the algorithm,and overcoming the defect that SPBO tended to fall into local optimum.Finally,the robot platform with joint torque measurement function was used to conduct experiments on the identification of dynamics parameters.The results show that the BSPBO algorithm has higher convergence accuracy and faster convergence speed,and can finish the identification of the kinetic parameters of each joint stably and efficiently.