Distance Measures for Strong Observability and a Trisection Algorithm for their Reliable Calculation

Description

In practical applications, systems are often influenced by external disturbances, which can be modeled as unknown inputs. If unknown inputs are considered, strong observability and strong detectability are important system properties for state estimation. Standard criteria for these properties have a binary character and their reliable numerical evaluation therefore is difficult, but important for example in sensor placement. In this talk, continuous distance measures are presented, which provide reliable and interpretable results. The calculation of the distance measures for strong observability and strong detectability involves a non-convex optimization problem and can be challenging. A trisection algorithm is presented that can be used for the reliable and efficient calculation of the distance measures.

Period	2 Feb 2022
Held at	National Autonomous University of Mexico, Mexico