Abstract
The paper deals with a methodology for identifying switching sequences and switching instants of switched linear system (SLS). The proposed methodology is based on the Discrete Particle Swarm Optimization (DPSO) technique. The identification problem is formulated as an optimization problem involving noisy data (system inputs and outputs).
•Hybrid systems.•Identification of a switched linear systems.•Identification of active mode.•Discrete Particle Swarm Optimization (DPSO).•Application to: buck–boost converter.
In this paper, a methodology for identifying switching sequences and switching instants of switched linear systems (SLS) is derived. The identification problem of a SLS is a challenging and non-trivial problem. In fact, it involves interaction between binary, discrete and real-valued variables. A SLS switches many times over a finite time horizon and thus estimating the sequence of activated modes and the switches locations is a crucial problem for both control and Fault Detection and Isolation (FDI). The proposed methodology is based on the Discrete Particle Swarm Optimization (DPSO) technique. The identification problem is formulated as an optimization problem involving noisy data (system inputs and outputs). Both a set of binary variables corresponding to each sub-model before and after each switch, and the corresponding switching instants are iteratively adjusted by the DPSO algorithm. Thus, the DPSO algorithm has to classify which sub-system has generated which data. The efficiency of the proposed approach is illustrated through a numerical example and a physical one. The numerical example is a Switched Auto-Regressive eXogenous (SARX) system and the physical one is a buck–boost DC/DC converter.