Sensor selection for observability in Interpreted Petri Nets: a genetic approach

This paper addresses the minimal cost sensor selection problem for observability in Interpreted Petri Nets (IPN) models of Discrete Event Systems (DES). A simple genetic algorithm to solve this problem is presented. This procedure takes advantage of a characterization of the observability property for IPN models presented in a previous work. Such characterization is based on the event-detectability and marking-detectability properties, which can be tested in a polynomial time. The presented genetic algorithm is quite simple and helps to reduce the design effort and time of DES.