Abstract
Maintaining both safety and efficiency in dynamic process systems is a persistent challenge, especially in safety-critical energy applications. This work introduces a real-time, risk-based control framework that embeds safety directly into control decision-making through a structured two-stage approach.
In the first stage, a deterministic control layer is developed by integrating a dynamic Risk Indicator (RI) into a model predictive control (MPC) scheme. This RI continuously monitors thermal deviations and enables the controller to respond proactively to emerging risk conditions. In the second stage, probabilistic safety constraints are incorporated into an explicit MPC (eMPC) formulation using multi-parametric programming. This allows the system to translate uncertainty into actionable control laws and embed chance-constrained safety limits—enabling real-time control law evaluation without the need for online optimization.
The framework is validated on a cyber-physical platform of a proton exchange membrane water electrolyzer (PEMWE), used here as a representative case study. A linear state-space model derived from experimental data is used to implement the two-stage controller, which regulates stack current and water flow rate. The deterministic layer ensures thermal stability, while the probabilistic extension enhances flexibility and robustness under uncertainty.
This work presents a unified, computationally efficient strategy for embedding real-time risk assessment into advanced control design—supporting resilient, high-performance operation across a wide range of energy systems.
Keywords: Risk assessment, Model Predictive Control, Safety, Multi-Parametric Programming, Optimization, Probabilistic constraints, chance constrained programming