Multi-stage Robust Implicit Decision Rule for Optimal Control Problem of Energy Storage System

The optimal control problem including random variables is difficult to solve. Existing methods typically rely on defining explicit decision functions to make the problem tractable. However, in practical numerical testing, we observed that some strongly coupled constraints, such as energy storage lev...

Full description

Saved in:
Bibliographic Details
Published in2024 IEEE 20th International Conference on Automation Science and Engineering (CASE) pp. 1120 - 1125
Main Authors Zhao, Jiexing, Zhai, Qiaozhu, Zhou, Yuzhou, Cao, Xiaoyu, Hu, Jianchen, Xue, Fei, Li, Xutao
Format Conference Proceeding
LanguageEnglish
Published IEEE 28.08.2024
Subjects
Aerospace electronics
Costs
Energy storage
Mathematical programming
Microgrids
Optimal control
Optimization
Random variables
Testing
Uncertainty
Online AccessGet full text

Cover

More Information
Summary:The optimal control problem including random variables is difficult to solve. Existing methods typically rely on defining explicit decision functions to make the problem tractable. However, in practical numerical testing, we observed that some strongly coupled constraints, such as energy storage level limits, will impose strict restrictions on these simplified decision functions, potentially leading to significantly suboptimal solutions. Motivated by these challenges, this paper proposes a multi-stage robust implicit decision rule for the scheduling problem of energy storage systems. The main idea is to find an explicitly feasible decision function space to guarantee the multi-stage operating feasibility. When random variables are observed, decisions are adaptively optimized within the feasible decision space by solving a straightforward mathematical programming. Explicit decision functions are not required, ultimately enhancing the feasibility and optimality of the stochastic optimization for energy storage systems. Numerical tests are implemented on a real-world microgrid, verifying the effectiveness of the proposed method.
ISSN:2161-8089
DOI:10.1109/CASE59546.2024.10711743