Emergency diesel generator control system verification by model checking and compositional minimization

Digital instrumentation and control (I&C) systems containing programmable logic controllers are challenging to verify. They enable complicated control functions and the state spaces (number of distinct values of inputs, outputs and internal memory) of the designs easily become too large for comprehensive manual inspection. Model checking is a formal method that can be used for verifying that systems have been correctly designed. A number of ecient model checking systems are available which provide analysis tools that are able to determine automatically whether a given state machine model satises the desired safety properties. However, model checking of large complex systems is often quite infeasible. In this paper, we present a compositional minimization technique for abstracting large modular function block based systems. We have applied the abstraction technique to the verication of a safety-critical emergency diesel generator control system. The system is so large that the non-abstract model could not be model checked within reasonable resources. Using the abstraction technique we managed to verify several universal properties of the system and were able to discover errors in the system designs. The abstraction technique is intended as a basis for an iterative abstraction renement framework