Open Access

ARTICLE

Robust Model Predictive Control Optimization of Hybrid System with Power-Heat-Hydrogen Storage

Geng Zhang, Fangzheng He, Jiayi Hu, Jiangjiang Wang^*
Yanzhao Electric Power Laboratory, North China Electric Power University, Baoding, China
* Corresponding Author: Jiangjiang Wang. Email:
(This article belongs to the Special Issue: Clean Energy and Low-Grade Energy Utilization: Material, Component, and System Innovation)

Energy Engineering https://doi.org/10.32604/ee.2026.081393

Received 01 March 2026; Accepted 28 April 2026; Published online 21 May 2026

Abstract

Hybrid systems face significant challenges in maintaining stable operation due to inherent uncertainties in renewable sources and demand loads. To address this, this study proposes a robust model predictive control framework for hybrid system with multi-energy storage (electrical/thermal/hydrogen storage). The system integrates photovoltaic, gas turbine, electrolyzer, and heat pump, leveraging multi-vector storage to buffer uncertainties through dynamic energy conversion and storage coordination. A hybrid forecasting model generates time-varying prediction intervals for source-load uncertainties, quantifying probabilistic bounds via kernel density estimation. The robust model employs these intervals to formulate a min-max optimization problem, proactively coordinating multi-energy storage dispatch under uncertainty. The framework is reformulated into a tractable robust counterpart through constraint tightening, ensuring feasibility across all uncertainty realizations. A case study validates the proposed method. The impacts of confidence levels in balancing robustness against conservatism within uncertainty sets and prediction horizon length on scheduling adaptability are analyzed. The simulation results demonstrate that a 50.6% reduction in grid dependency vs. deterministic model predictive control, despite a 29.7% higher nominal cost, ensuring feasibility under worst-case scenarios. Shrinking prediction horizons from 6 to 2 steps increases costs by 43.6% due to conservative electricity procurement. The approach explicitly co-designs uncertainty-aware predictions with multi-storage dynamics, advancing real-time scheduling robustness for complex hybrid system.

---

Keywords

Hybrid system; robust model predictive control (RMPC); multi-energy storage; uncertainties; interval prediction

---