Machine Learning in Modern SCADA Systems: Opportunities and Challenges
Ivana Šenk, Srdjan Tegeltija, Laslo Tarjan
Abstract
One of the key automation levels in every industrial system is Supervisory Control and Data Acquisition (SCADA), a system used to collect data from the processes, visualize them to the users in an adequate form, and provide monitoring and control capabilities. As the industrial world transitions towards more advanced and interconnected infrastructures, the potential benefits of leveraging machine learning algorithms for enhanced monitoring, control, and decision-making in SCADA systems become evident. This paper explores the integration possibilities of machine learning models within modern SCADA systems. It discusses the key opportunities, including improved anomaly detection, predictive maintenance, and optimized system performance. Simultaneously, it addresses challenges such as availability of quality data, data security, and model interpretability, as well as practical implementation challenges.