Optimization framework for multi-objective energy management strategy in hybrid electric vehicles integrating explainable artificial intelligence

Zhiyuan Lu, Hu Wang, Guanzhang He, Yong Chen, Zihou Li, Zunqing Zheng, Mingfa Yao, Song Zhang and Hao Wang

Applied Energy, 2025, vol. 399, issue C, No S0306261925012140

Abstract: This study presents a multi-objective optimization (MOO) and explainable artificial intelligence (XAI) integrated framework for establishing MOO strategies in hybrid electric vehicle (HEV) and analyzing the decision-making patterns of the system. Specifically, the study first establishes a power-split HEV model and incorporates fuel consumption, electricity consumption, battery degradation, and the number of engine start-stops into the system evaluation metrics. Subsequently, the dynamic programming (DP) algorithm is employed to develop an offline globally optimal multi-objective strategy based on the multi-indicator vehicle model, and the non-dominated sorting genetic algorithm-II (NSGA-II) is used to perform MOO of the strategy's cost function. By combining the cognition-driven analytical hierarchy process (AHP) decision-making method with the data-driven technique for order preference by similarity to ideal solution (TOPSIS) method, the AHP-TOPSIS decision-making method is used to select the optimal solution from the Pareto frontier. Tree-based XAI methods are introduced, employing mean decrease impurity (MDI) and partial dependence plots (PDP) to analyze the interaction mechanisms among the four objectives during the decision-making process. A double-layer random forest (RF) energy management strategy is constructed, combining a five-fold cross-validated RF pattern recognition model with an engine power prediction model based on the optimal solution dataset. The results demonstrate that the multi-objective strategy exhibits better overall performance compared to single-objective strategies. The proposed double-layer RF strategy reduces fuel consumption by 1.8 %, maintains similar electricity consumption, decreases battery degradation by 7.1 %, and reduces the number of engine start-stops by 82.1 % compared to a rule-based (RB) strategy, with minimal deviation from the DP strategy. This validates the superior performance of the strategy in multi-objective control processes.

Keywords: Hybrid power systems; Multi-objective optimization; Machine learning; Explainable artificial intelligence; Energy management strategy (search for similar items in EconPapers)
Date: 2025
References: Add references at CitEc
Citations:

Downloads: (external link)
http://www.sciencedirect.com/science/article/pii/S0306261925012140
Full text for ScienceDirect subscribers only

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:eee:appene:v:399:y:2025:i:c:s0306261925012140

Ordering information: This journal article can be ordered from
http://www.elsevier.com/wps/find/journaldescription.cws_home/405891/bibliographic
http://www.elsevier. ... 405891/bibliographic

DOI: 10.1016/j.apenergy.2025.126484

Access Statistics for this article

Applied Energy is currently edited by J. Yan

More articles in Applied Energy from Elsevier
Bibliographic data for series maintained by Catherine Liu ().