Stochastic and Deterministic Disturbance Cancellation for Nano-Precision Systems.

摘要

This thesis develops signal processing and control algorithms as well as specialized hardware implementation required to achieve nano-precision using long range motion mechatronic positioning systems. Implementations focus on the Multi-Scale Alignment Positioning System (MAPS) designed for nano-imprint lithography, and specifically focus on challenges pertaining to a single MAPS Halbach brush-less linear motor.;In order for both high speed long range motion and nano precision, novel analog/digital quadrature signal processing algorithms were developed. Nanometer precision is achieved via the analog signal while high speed motion is captured via the digital signal. In ideal sensors this data-fusion produces the desired high speed high precision sensing desired. Algorithms were also developed to compensate for non-linearities in the sensors. Finally, dedicated electronics were developed and integrated to attain required board level high speed sampling rates.;Standard control algorithms were deemed insufficient to achieve nano-resolution positioning in the presence of both deterministic and stochastic disturbances which are present in all long range motion mechatronic positioning systems. Therefore three methods for asymptotic deterministic disturbance rejection are proposed. In the first methodology, a peak filter is used to implement an internal model principle (IMP) type control that enables flexible on-the-fly multi-frequency disturbance rejection. Due to this flexible nature, an additional frequency disturbance estimator can be used to adjust the multiple frequencies online. In addition to the frequency adjustment, online adaptive control (AC) is used to reduce out stochastic colored noise.;In the second methodology a finite impulse response (FIR) filter is designed using convex optimization algorithms was developed. This architecture allows for the inclusion of constraints such as transient response, algorithm computation complexity, robust stability, memory size, and nominal performance during control algorithm synthesis, some of which are not provided with the other control methods.;In the final design methodology, a time-varying approach is developed to take advantage of the known onset of deterministic disturbances. By using this additional information, this approach exploits the structure of Kalman Filter (KF) disturbance estimation to develop a time varying filter which is able to quickly compensate for any new disturbance, faster than any other method. Again, AC, is implemented to minimize stochastic processes for further performance enhancement.