A new output error method for a decoupled identification of electrical and mechanical dynamic parameters of DC motor-driven robots

Abstract

Abstract The identification methods based on the inverse dynamic model allow to use the linear least-squares technique to estimate the dynamic parameters (IDIL-LS method) of mechanical and electrical systems. Assuming a good data filtering to calculate derivatives and reference trajectories that excite the system dynamics, this method has been validated through the experimental identification. Recently, an output error (OE) avoids calculating the derivatives of actual data. This is possible by using the same reference trajectory and the same control law for the actual system and the simulated one. The optimal parameters minimize the 2-norm of the error between the actual output y and the simulated one ys. This nonlinear least-squares problem is dramatically simplified using the inverse dynamic model which is linear to the parameters. This method, called DIDIM (Direct and Inverse Dynamic Identification models), has been validated on robot's dynamic parameters identification. In this paper, DIDIM is applied to the decoupled identification of the electrical and mechanical dynamic parameters of DC motor-driven robots using only the reference control voltage of the PWM amplifier and the current reference of the current loop.