Network Identification with Partial Excitation and Measurement
(2019) 58th IEEE conference on Decision and Control (CDC19). — Location: Nice, France (11.December.2019)
Files
BGH19CDC1.pdf
Open Access - Adobe PDF
- 201.85 KB
Details
- Authors
- Abstract
- The question addressed in this paper is to determine what is the input/output information that must be acquired in order to identify a network of dynamical systems, or part of such network. More specifically, which nodes must be excited and which nodes must be measured in order to render a network, or a given piece of a network, identifiable. We show that identification of a given set of edges requires that all nodes connected to these edges must be either excited or measured or both. We study in detail the identifiability of four types of edge sets: edges arriving at a given node, edges leaving a given node, edges forming a tree and edges forming a loop. For each case, we characterise excitation and measurement patterns that provide identifiability.
- Affiliations
Citations
Bazanella, A. S., Gevers, M., & Hendrickx, J. (2019). Network Identification with Partial Excitation and Measurement. IEEE Conference on Decision and Control, Including the Symposium on Adaptive Processes. Proceedings, 5500-5506. https://hdl.handle.net/2078.5/254664 (Original work published 2019)