ARRTOC: Adversarially Robust Real-Time Optimization and Control

Real-Time Optimization (RTO) plays a crucial role in process operation by determining optimal set-points for lower-level controllers. However, tracking these set-points can be challenging at the control layer due to disturbances, measurement noise, and actuator limitations, leading to a mismatch between expected and achieved RTO benefits. To address this, we present the Adversarially Robust Real-Time Optimization and Control (ARRTOC) algorithm. ARRTOC addresses this issue by finding set-points which are both optimal and inherently robust to implementation errors at the control layers. ARRTOC draws inspiration from adversarial machine learning, offering a novel constrained Adversarially Robust Optimization (ARO) solution applied to the RTO layer. We present several case studies to validate our approach, including a bioreactor, a multi-loop evaporator process, and scenarios involving plant-model mismatch. These studies demonstrate that ARRTOC can improve realized RTO benefits by as much as 50% compared to traditional RTO formulations that do not account for control layer performance. © 2024 The Authors