Identification of 5-Axis Machine Tools Feed Drive Systems for Contouring Simulation

Abstract

An identification technique is introduced for identifying closed loop transfer function of machine tool’s feed drive systems to be used in simulation of the tracking and contouring performance of Computer Numerical Controlled (CNC) machine tools. The identification is performed from air-cutting tests utilizing only standard G-codes containing linear motion commands. A general transfer function model is derived for representing the closed loop tracking response of the feed drive system. The model considers the drive to be controlled by commonly used controller schemes such as P-PI Cascade, PID or the Sliding Mode Controller (SMC) with feed-forward dynamic and friction compensation. The parameters of the model transfer function are fitted tominimize the discrepancy between the actual and predicted axis position on the axis. In order to guarantee the stability of the identified model transfer function, bounds on the pole locations are imposed. The resultant constrained non-linear optimization problem is solved efficiently using the Particle Swarm Optimization (PSO) method. For achieving reliable convergence of the stochastic PSO algorithm, a parameter tuning strategy is presented. Simulation and experimental studies show that the identified feed drive model captures the fundamental dynamics of the drives system accurately for simulating their closed loop response. Combined with the kinematics of the machine, contouring errors of 5-axis CNC machine tools during simultaneous multi-axis motion are predicted.