Identification of frequency response functions of a flexible robot as tool-holder for robotic grinding process

Abstract

The Frequency Response Functions (FRF) are identified for a robot during normal grinding operation using three different identification approaches, namely the spectral method, the Autoregressive with exogenous excitation (ARX) model and the State Space (SS) model. A comparison is made to determine the performance of each approach. The results show that both the ARX and SS models can be used for identification of the FRF but the ARX provides the best performance at the same model order. The spectral estimation method exhibits the worst capacity for the identification of operational frequency response function. It is found from the FRF identification that the acceleration on the feed direction of the robotic grinding process is most sensitive to the excitation force. Further developments are ongoing on the use of these models for evaluating the displacement at the end effector and for controlling the vibration of operational mechanical systems in operation such as grinding.