Scholarlink

<p>The energy management strategy (EMS) is critical for the efficiency of fuel cell hybrid electric vehicles. However, fuel cell degradation alters system parameters over the vehicle’s lifetime, rendering conventional static EMS suboptimal and increasing hydrogen consumption. To address this challenge, this paper proposes a degradation-adaptive EMS that dynamically adjusts its power split based on the real-time state of health (SOH) of the fuel cell system (FCS). First, to enable precise online estimation of the SOH, a Long Short-Term Memory neural network model is established, with its hyperparameters optimized using an enhanced ant colony algorithm. Subsequently, incorporating the real-time SOH as a critical input, a rolling-horizon optimization problem is formulated to minimize equivalent hydrogen consumption. A Pontryagin’s minimum principle-based method is employed for a computationally efficient solution. The Hardware-in-the-Loop experiments results demonstrate that, compared to its non-adaptive counterpart, the proposed adaptive EMS reduces hydrogen consumption by 1.46% to 4.02% by actively predicting the degradation state of FCS. Furthermore, it achieves an 8.51% to 12.13% reduction in hydrogen consumption over conventional rule-based strategies with control performance approaching the theoretical optimum.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Adaptive Energy Optimization Based on Fuel Cell Degradation Prediction for Fuel Cell Hybrid Electric Vehicles

  • 期刊论文

  • 发表日期：

    2026年4月7日

查看全文
  • Yilin Wang,
  • Shengyan Hou,
  • Zhihuan Jia,
  • Jinwu Gao,
  • Dong Hao,
  • Hong Chen

摘要

The energy management strategy (EMS) is critical for the efficiency of fuel cell hybrid electric vehicles. However, fuel cell degradation alters system parameters over the vehicle’s lifetime, rendering conventional static EMS suboptimal and increasing hydrogen consumption. To address this challenge, this paper proposes a degradation-adaptive EMS that dynamically adjusts its power split based on the real-time state of health (SOH) of the fuel cell system (FCS). First, to enable precise online estimation of the SOH, a Long Short-Term Memory neural network model is established, with its hyperparameters optimized using an enhanced ant colony algorithm. Subsequently, incorporating the real-time SOH as a critical input, a rolling-horizon optimization problem is formulated to minimize equivalent hydrogen consumption. A Pontryagin’s minimum principle-based method is employed for a computationally efficient solution. The Hardware-in-the-Loop experiments results demonstrate that, compared to its non-adaptive counterpart, the proposed adaptive EMS reduces hydrogen consumption by 1.46% to 4.02% by actively predicting the degradation state of FCS. Furthermore, it achieves an 8.51% to 12.13% reduction in hydrogen consumption over conventional rule-based strategies with control performance approaching the theoretical optimum.