Virtual model building for 5-axis laser drilling machine from field data

This paper presents an approach of feed drive modelling and identification for a 5-axis laser drilling machine. The identification directly exploits in-process data during the production without causing additional downtime. A multibody dynamic model as a multi-input multi-output (MIMO) system is adopted for considering the dynamic coupling effect which is inherent for a 5-axis machine with rotary axes. The model also includes the torque ripple, one of the common disturbances for direct drive motors. The above dynamic model together with Stribeck friction and Coulomb friction are identified by combining Least Squares (LS) and Particle Swarm Optimization (PSO). For the purpose of process simulation, which may contain arbitrary motion including reversal and near-zero velocity, the LuGre friction model is later applied to identify the dynamic characteristics of friction. The result shows that these different dynamic characteristics can be successfully identified and decoupled. This virtual model becomes a good foundation for the process simulation and other applications.