(452c) Cybersecurity of Process Control Systems Via a Two-Tier Control Architecture

Since industrial control systems are usually integrated with numerous physical devices, the cybersecurity of control systems plays an important role in safe operation of industrial chemical processes. However, due to the use of a large number of control actuators and measurement sensors and the increasing use of wireless communication, control systems are becoming increasingly vulnerable to cyber-attacks, which is a series of computer actions aiming to compromise the security of control systems (e.g., integrity, stability and safety) and may cause severe industrial incidents [1]. Therefore, the design of advanced detection systems [2] and process control system for nonlinear processes in the presence of cyber-attacks remains an important issue.

In the context of improving process control using a large set of networked sensors and control actuators, a two-tier control architecture was developed for process control systems in [3] to improve plant performance accounting for high-frequency and low-frequency state measurements to handle the increasing number of sensors and actuators. Specifically, a lower-tier control system was developed using PI control based on wired, point-to-point communication with continuous measurements to ensure closed-loop stability, while an upper-tier control system was developed using model predictive control (MPC) based on networked (potentially wireless) communication with asynchronous measurements to improve closed-loop performance. However, in such a two-tier control architecture, the upper-tier control system is more vulnerable to cyber-attacks due to networked measurements. Motivated by this, a secure two-tier control architecture is developed to mitigate the impact of cyber-attacks targeting the upper-tier control system in chemical processes by integrating a machine learning-based detection method [4] with a two-tier control architecture. Specifically, a class of stealthy cyber-attacks that aim to compromise close-loop stability is first designed and applied to the upper-tier control system. Subsequently, a machine learning-based detection method is developed based on continuous state measurements from the lower-tier control system. After the cyber-attack is detected, the upper-tier control system is updated to mitigate the impact of cyber-attacks and re-optimize the closed-loop system. Finally, the effectiveness of the proposed secure two-tier control architecture is demonstrated through a chemical process network example.