Boosting Exploratory Testing of Industrial Automation Systems with AI

摘要

Testing of large and complex industrial control systems is challenging as the space of possible input and environmental parameters is large. Searching the entire space for potential failures is practically infeasible. This paper introduces an industrial control system robustness testing problem and evaluates artificial intelligence (AI) based strategies to efficiently explore the space and to identify parameter sets that can cause the system to fail. The proposed solution approach uses regression techniques to speed up the search and clustering methods to identify parameter sets that represent distinct system failures.