Dynamic parameters identification of rolling joints based on the digital twin dynamic model of an assembled ball screw feed system

Abstract

Currently, due to the rare consideration on the coupling of the various rolling joints and their directional dynamic parameters, and the constraints of the traditional modeling methods, the dynamic modeling precision of the ball screw feed system and the dynamic parameters identification accuracy of the rolling joints are difficult to be further improved. In this paper, a novel method to identify the dynamic parameters of rolling joints based on the digital twin dynamic model of the assembled ball screw feed system is proposed. Firstly, synchronizing information of the physical entity, the geometric model is constructed. Then the finite element analysis (FEA) model is constructed which can simultaneously consider multiple rolling joints and their dynamic parameters at multiple directions. Based on the FEA modal data, the deep neural network (DNN) model is constructed to reflect the mapping between the dynamic parameters and the natural frequencies. Thus, the digital twin dynamic model can be established by fusion of these sub-models. Combining the digital twin-driven and experimental natural frequencies, the optimization model is built, and the dynamic parameters can be identified by particle swarm optimization (PSO) algorithm. Finally, the relative error of dynamic parameters identification is less than 3%, which indicates that the proposed method is feasible, effective, and has greater accuracy.