PLC-based Implementation of Process Observation and Fault Detection for Discrete Event Systems

This paper demonstrates how PLC-code for process observation can be automatically generated from a discrete event process model expressed as either PLC-code or as extended finite automata (EFA). The generated code will also detect faults due to unexplainable sensor signal changes. Extended finite automata are automata augmented with variables and transition conditions. These features make EFA suitable for modeling systems that interact with a PLC through binary signals. In order for the generated observer code to work, the EFA model must fulfill some requirements concerning determinism and observability. These requirements are here formalized. Using the PLC-languages of the IEC 61131 standard for process modeling narrows the gap between the pragmatic industry and the formal academia. The cyclic execution model of a PLC is here replaced by execution at stochastic intervals.