System Identification and Adaptive Control of the Multi-Axis Bending and Twisting Process

Abstract

An adaptive control scheme for the multi-axis bending and twisting process based on system identification of process models is presented. Three system identification methods, the frequency-response, the least-squares, and the singular value decomposition methods, have successfully identified the process models. Control based on these identified process models performed better than control based on the theoretical process models proposed by Luo et al. (1996). The shape errors between the intrinsic geometric quantities of the actual and desired parts are applied to the inverse identified process models to calculate incremental changes in axis commands for subsequent iterations. Comparison with experimental results demonstrates that adaptive control based on the identified process models achieves more rapid convergence of the iterations than the previous approach.