An automata based hybrid modeling approach to synthesize sequential diagnostic tests

Although the automata based diagnostic tests have already been developed in the past for differentiating the originally inseparable fault origins in simple chemical processes, their applicability in realistic systems is still questionable. This is primarily due to the facts that automata are basically modeling tools for the discrete event systems (DESs) but most process variables are continuous. To address this practical issue, the dynamic behavior of every processing unit involved in a given operation is modeled in this work by incorporating both the generic engineering knowledge and also the ASPEN-generated dynamic simulation data into a single automaton. The improved test plans can then be synthesized according to the system model obtained by assembling all such automata. The feasibility of this automata based model building strategy is demonstrated with two examples concerning the startup operations of a flash process and also a distillation column.