Lunch colloquium Jacques Noom (PhD at N4CI)

01 February 2023 12:30 till 13:30 - Location: instruction room g, 3me | Add to my calendar

"Model-free data-driven fault diagnosis"

Data-driven fault diagnosis for dynamical systems typically is done in two steps. First the candidate models of the system are identified followed by a fault diagnosis step based on the identified candidate models. The first step requires large amounts of historical data to identify the candidate models.  In our contribution we introduce the problem of model-free data-driven fault diagnosis, in which both identification of the linear time-invariant system and fault diagnosis are carried out simultaneously. This waives the requirement for (large amounts of) historical data. The problem is formulated within a blind system identification context resulting in computationally efficient solutions based on convex optimization. Online performance is achieved by recursive implementation of a proximal algorithm.