A rapid method for modal parameter prediction in robotic milling

Abstract

Due to the multi-degree-of-freedom structure and low rigidity of robots, they are prone to chatter phenomena when subjected to vibrational milling forces, leading to a decline in milling precision. To mitigate the impact of chatter on machining, the robot's frequency response can be obtained through hammer testing, from which the FRF can be further derived to guide the optimization of the robot's machining posture. However, the extensive range and diverse postures during robotic machining, coupled with the cumbersome and static nature of hammer testing, present significant challenges. To address these issues, this paper proposes a data-driven method for predicting the modal parameters of robots. The hammer test, limited to static conditions, cannot provide real-time modal parameters for arbitrary postures. By analyzing the superposition principles between different frequency response functions, the relationship between robot posture and modal parameters is established, forming the theoretical basis for prediction. A GPR based method for predicting modal parameters is proposed, and in conjunction with the fully discrete method, the stability lobes diagram for any machining posture of the robot is computed in real-time, serving as a stability indicator during machining.