The 2-DOF dynamic model of these two joint was built by using Euler-Lagrange formulation. It was simplified to a linear-in-parameter mode. The relationship between joint torque and actuator force was deduced by differentiating the function of relationship between the joint angle and cylinder displacement, and friction was introduced into the actuator force. The recursive least square method (RLS) and recursive stochastic Newton algorithm (RSNA) were used to identify the unknown combination parameters respectively. Comparison experiments showed the predictive errors of joint torque of RSNA was reduced by 65% and 63% for the dipper joint and bucket joint respectively. The results demonstrated RSNA was more robust to the system noise. It can identify the dynamic parameters precisely. 