Computation of Optimal Actuator/Sensor Locations
Abstract
Many systems, such as acoustic noise and structural vibrations are distributed in space. The location of control actuators, and also the sensors, is a variable the design of a control system. It has been known for some time that controlled system performance depends on actuator location. However, selection of actuator/sensor locations has not previously been systematically addressed. A mechatronic approach where controller design is integrated with actuator location was used. This effort was complicated by the fact that approximations to the governing equation are required. Different objectives require different cost functions and lead to different algorithms as well as conditions for computation. Algorithms for calculating actuator locations that optimize various cost functions were developed. These algorithms were tested with computer simulations and also an experiment on a beam. The studies demonstrated dramatically improved performance at optimal over non-optimal actuator locations. Even for relatively simple systems, the optimal location is often not intuitively obvious. It was also investigated how the optimal actuator location is affected by the choice of cost function. Controllability is a very popular criterion. It was found to lead to locations that were considerably worse than those selected to optimize the controller design objective.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 26, 2013
- Accession Number
- ADA594613
Entities
People
- Kirsten Morris
Organizations
- University of Waterloo