This research topic focuses on data-driven modeling of evolving systems with the ultimate goal of aiding control and maintenance of these systems. Our objective is to propose a set of tools for estimating models of complex physical processes from measured data. These models are then to be used for diagnosis, detection and prognosis of these processes. These tools are to rely on system identification and dynamical classification algorithms, and they should support both gray and black-box approaches. These methods are expected to be used for detecting faults and new operating modes in general, and slow changes of behavior (drift) in particular. In terms of applications, we concentrate on energy management, environmental and transportation systems. For the future, we would like to work on :
- methods for modeling the behavior of evolving systems in multi-mode environment,
- data-driven methods for detecting and predicting drifts,
- adaptive supervision of complex processes.