Transient growth and instability control using iteration-varying filter ILC in flexible dual flywheel system

In the field of precision motion control, many industrial systems have flexible modes. This is especially true for very high performance nanopositioning systems where traditional bearing systems are replaced with flexible structures in order to eliminate sliding friction. The precision motion control is difficult in flexible systems because the rapid phase transition around the resonant frequency must be accurately modeled in order to actively control the resonant motion. In order to experimentally study the effectiveness of various control algorithms on flexible systems, a dual-flywheel flexible system is constructed. The mechatronics system design strategy is presented in this thesis. The control algorithms of interest for this study are Iterative Learning Control (ILC) algorithms, which are very high precision algorithms that learn from repetition of the desired trajectory. One of the challenges in ILC design is the problem of large transient growth. Large transient growth occurs involves many iterations of very large growth of the error before finally converging to a stable control. In many cases the growth is too large to safely employ on physical systems. Recent work indicates that large phase error in system models, such as those that inaccurately model resonant dynamics, can be a significant driver for large transient growth. One new ILC algorithm that was developed for this purpose, but has never been applied in practice, is the Iteration-varying Filter ILC. This thesis presents experimental results from the application of Iteration-varying Filter ILC to the dual flywheel system. Design and tuning of the Iteration-varying Filter ILC are also presented --Abstract, page iii