Using entropy to maximize the usefulness of data collection
This paper presents a generic methodology for measurement system configuration when the goal is to identify behaviour models that reasonably explain observations. For such tasks, the best measurement system provides maximum separation between candidate models. In this work, the degree of separation between models is measured using Shannon’s Entropy Function. The location and type of measurement devices are chosen such that the entropy of candidate models is greatest. This methodology is tested on a laboratory structure and, to demonstrate generality, an existing fresh water supply network in a city in Switzerland. In both cases, the methodology suggests an appropriate set of sensors for identifying the state of the system.
Nutzung und Vervielfältigung:
Alle Rechte vorbehalten