Evaluation of Measurement Noise and Sampling Frequency for Online Identification of End Effector Stiffness and Damping in a Cartesian Robot

An online parameter identification method using pseudo-random binary sequence (PRBS) excitation is proposed to estimate the flexibility and damping of an experimental end effector setup in a belt-driven Cartesian robot. The proposed identification method is first evaluated on a simulation model for choosing an appropriate measurement method, saving potential instrumentation costs and validating the feasibility. A detailed simulation model of the real system is developed, incorporating nonlinear position-based Coulomb friction, viscous friction, end effector flexibility, and measurement noise replicating real measurement data. The model is simulated with the same conditions as the real system, including the identical control architecture deployed in the robot’s PLC. The flexibility of the end effector is modeled by a mass–spring mechanism consistent with the real setup. A PRBS torque excitation is applied to the controller output to induce vibrations in the end effector, and a linear least squares (LLS) regression is employed to identify the spring stiffness and damping parameters. The results demonstrate accurate identification of stiffness and damping in the end effector over increasing measurement noise and sampling frequencies. This information enables the selection of suitable sensors for such identification in the end effector of a Cartesian robot system, validating the effectiveness of the proposed approach.