Input Shaping Design for Multi-Input Flexible Systems

This paper investigates a new design technique of input shaping filters for multi-input flexible systems using convex optimization synthesis techniques for finite impulse response filters (FIR filters). The objective of the input shaping filter design is to find the minimum length and the minimum number of nonzero impulses of the FIR filter that forces the system to track the reference command without any residual vibration, while satisfying additional performance and control constraints. This multi-objective optimization is solved using a two-step algorithm that sequentially solves two quasi-convex optimization problems. Compared with previously published nonlinear optimization approaches, this new approach does not require a priori knowledge of the forms of input shaping filters and enables much greater flexibility for including additional performance and robustness objectives. Furthermore, this convex-based approach can be applied to multi-input systems. The multiple input shaping filter has been experimentally verified on the Stanford University Two-Link Flexible Manipulator.