Scholarlink

<p>To enhance the net power output of fuel cell systems under varying conditions, this study addresses this challenge by developing an adaptive neural network (NN) control strategy enhanced with parameter optimization. The focus is on the hydrogen supply subsystem and its performance is critical to the system’s overall net power. First, the detailed model of the hydrogen supply subsystem is established, and the detailed analysis about anode pressure, the hydrogen excess ratio (HER) and the net power output is given in this paper. Subsequently, a two-dimensional genetic algorithm (TDGA) is introduced to optimize the reference signals for anode pressure and HER, thereby ensuring maximum net power. By utilizing input–output linearization techniques, the originally coupled nonlinear multi-input multi-output (MIMO) system is decoupled and transformed into a canonical structure. Based on this transformation, an adaptive NN controller is designed to regulate the flow valve and hydrogen circulation pump. Through a series of simulations and hardware-in-loop (HIL) tests, the proposed control strategy is demonstrated to effectively optimize net power across diverse operating scenarios. Quantitative comparisons further highlight the advantages of the proposed optimization algorithm and adaptive control techniques in enhancing both power output and control system performance. Compared with prior control studies, the main contributions of this work are summarized as follows: (1) A TDGA strategy is introduced to determine the steady-state reference values of HER and anode pressure for net power optimization. (2) A feedback-linearization decoupling controller is designed to independently regulate HER and anode pressure. (3) To improve dynamic performance and robustness, a NN based on the decoupled model is developed to compensate for system uncertainties and unknown disturbances.</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

Neural network control and optimization for fuel cell hydrogen supply system

  • 期刊论文

  • 发表日期：

    2026年8月4日

查看全文
  • Cunliang Ye,
  • Yunlong Wang,
  • Yu Gao,
  • Hexu Yang

摘要

To enhance the net power output of fuel cell systems under varying conditions, this study addresses this challenge by developing an adaptive neural network (NN) control strategy enhanced with parameter optimization. The focus is on the hydrogen supply subsystem and its performance is critical to the system’s overall net power. First, the detailed model of the hydrogen supply subsystem is established, and the detailed analysis about anode pressure, the hydrogen excess ratio (HER) and the net power output is given in this paper. Subsequently, a two-dimensional genetic algorithm (TDGA) is introduced to optimize the reference signals for anode pressure and HER, thereby ensuring maximum net power. By utilizing input–output linearization techniques, the originally coupled nonlinear multi-input multi-output (MIMO) system is decoupled and transformed into a canonical structure. Based on this transformation, an adaptive NN controller is designed to regulate the flow valve and hydrogen circulation pump. Through a series of simulations and hardware-in-loop (HIL) tests, the proposed control strategy is demonstrated to effectively optimize net power across diverse operating scenarios. Quantitative comparisons further highlight the advantages of the proposed optimization algorithm and adaptive control techniques in enhancing both power output and control system performance. Compared with prior control studies, the main contributions of this work are summarized as follows: (1) A TDGA strategy is introduced to determine the steady-state reference values of HER and anode pressure for net power optimization. (2) A feedback-linearization decoupling controller is designed to independently regulate HER and anode pressure. (3) To improve dynamic performance and robustness, a NN based on the decoupled model is developed to compensate for system uncertainties and unknown disturbances.