Deep Learning and Snow Geese Optimization for Plug-in Hybrid Electric Vehicle Energy Management: A Review

Abstract

The increasing adoption of plug-in hybrid electric vehicles (PHEVs) has intensified the need for intelligent energy management strategies that can efficiently optimize power distribution while maintaining battery health and system performance. Traditional rule-based and model-based approaches often fail to address the nonlinear and dynamic nature of real-world driving conditions. In recent years, deep learning and optimization techniques have emerged as promising solutions for enhancing energy management systems (EMS). This review focuses on hybrid frameworks that integrate advanced optimization algorithms, such as Snow Geese Optimization (SGO), with Relational Bi-level Aggregation Graph Convolutional Networks (RBAGCN) for efficient energy management in PHEVs. Graph-based deep learning models are particularly effective in capturing complex relationships between system components, while physics-informed approaches improve model reliability by embedding physical constraints. Recent studies demonstrate that combining optimization algorithms with neural networks enhances convergence, prediction accuracy, and adaptability. Physics-informed neural networks (PINNs) further improve generalization by incorporating physical laws into learning processes, reducing dependency on large datasets. This paper provides a comprehensive review of recent advancements highlighting the strengths and limitations of various techniques and identifying future research directions for developing efficient, scalable, and intelligent EMS for next-generation PHEVs.