Automatic Plant-Controller Input/Output Matching using Evolutionary Algorithms

Automation of software development is an actively researched problem. Search-based software engineering aims to apply various search-based techniques to software engineering problems. Recently we proposed the method for automatic generation of function block application using evolutionary algorithms and model checking and applied it to the problem of automatic generation of data connections in distributed control system. The aim of this paper is to further study this method on the problem of matching of input and output connections in a closed-loop plant-controller system. The computed fitness function distribution shows that the evaluated method successfully determines the correct input and output connections between the controller and the plant. Additionally, we evaluate how the composition of specification requirements in the fitness function affects the performance of the (1+1) evolutionary algorithm. We show that additional liveness formulas can improve the performance of the algorithm, while the introduction of safety formulas significantly decreases it.