Voltage Regulation in Polymer Electrolyte Fuel Cell Systems Using Gaussian Process Model Predictive Control

TL;DR

The paper addresses reliable voltage regulation in a PEFC under workload disturbances while enforcing safety limits on hydrogen pressure and input-rate changes. It proposes a GP-MPC framework that uses two Gaussian process models to predict $V_{FC}$ and $P_{H2}$ and linearizes them to a discrete-state space form for iterative quadratic programming, incorporating a variance-based constraint to hedge against model error. The key contributions are the development of a GP-MPC with a variance-informed hydrogen pressure constraint and a comparative evaluation against physics-based MPC, showing successful tracking at $V_{FC}=48$ V and safety adherence under $I=110$–$120$ A, with a trade-off of higher overshoot and slower response. This approach reduces reliance on detailed physical models and sensor requirements, offering a model-lean alternative with robust constraint handling for PEFC systems.

Abstract

This study introduces a novel approach utilizing Gaussian process model predictive control (MPC) to stabilize the output voltage of a polymer electrolyte fuel cell (PEFC) system by simultaneously regulating hydrogen and airflow rates. Two Gaussian process models are developed to capture PEFC dynamics, taking into account constraints including hydrogen pressure and input change rates, thereby aiding in mitigating errors inherent to PEFC predictive control. The dynamic performance of the physical model and Gaussian process MPC in constraint handling and system inputs is compared and analyzed. Simulation outcomes demonstrate that the proposed Gaussian process MPC effectively maintains the voltage at the target 48 V while adhering to safety constraints, even amidst workload disturbances ranging from 110-120 A. In comparison to traditional MPC using detailed system models, Gaussian process MPC exhibits a 43\% higher overshoot and 25\% slower response time. Nonetheless, it offers the advantage of not requiring the underlying true system model and needing less system information.

Figures (7)

• Figure 1: Gaussian process training results. The orange shaded area is the one $\sigma$ confidence interval, the prediction value plus/minus one prediction standard deviation (one $\sigma$).
• Figure 2: MPC control process
• Figure 3: GP-MPC control process
• Figure 4: Output voltage under the current step disturbance.
• Figure 5: Constraint handling and system inputs of MPC and GP-MPC.