Maintenance of process control algorithms based on dynamic program slicing

Today´s industrial control systems gradually lose performance after installation and must be regularly maintained by means of adjusting parameters and modifying the control algorithm, in order to regain high performance. Industrial control algorithms are complex software systems, and it is particularly difficult to locate causes of performance loss, while readjusting the algorithm once the cause of performance loss is actually realized and found is relatively simple. In this paper we present a software-engineering approach to the maintenance problem, which provides tools for exploring the behavior of a control algorithm, enables maintenance personnel to focus on only relevant parts of the algorithm and semi-automatically locate the part of the algorithm that is responsible for the reduced performance. The solution is tuning-free and can be applied to installed and running systems without modifying the existing control algorithm, which makes the solution well suited for industrial applications.