State space identification of max-plus-linear discrete event systems from input-output data

We present a method to identify the parameters of a state space model for a max-plus-linear discrete event system from input-output sequences. The approach is based on recasting the identification problem as an optimization problem over the solution set of an extended linear complementarity problem. Recently, we have shown that such a problem can be solved much more efficiently than previously by using a mixed integer programming approach. The resulting algorithm allows us to identify a state space model of a max-plus-linear discrete event system from input-output data. This method works for both structured and fully parameterized state space identification. In addition, we also obtain an estimate of the state sequence.