In the hydrogen supply system for PEMFC (proton exchange membrane fuel cell), previous research has primarily focused on developing models for fixed-geometry ejectors, which face significant challenges such as narrow operating ranges and limited controllability. To address these limitations, we establish a comprehensive hydrogen supply system model based on a novel composite adjustable ejector. The simulation model verified by experiment realized the linear control of the load current on the displacement of the needle valve, and an accurate entrainment ratio prediction model is obtained. Using a proportional-integral-derivative and pressure control algorithm (PID-PCA) composite strategy, the dynamic response characteristics of key performance parameters in a 200 kW PEMFC hydrogen supply system under variable loads are assessed. Results show that PID-PCA control reduces anode pressure response time by 45.50% (to approximately 3 s) compared to traditional PID control. Under PID-PCA control, the hydrogen utilization ratio, entrainment ratio, hydrogen supply excess ratio and flow rate can quickly and accurately track load demand and swiftly reach a stable state. The volume fraction of nitrogen in the anode gas supply pipe is always below 4.00%, and the volume fraction of hydrogen is stable above 91.56%. Both the hydrogen utilization ratio, entrainment ratio, and hydrogen supply excess ratio meet the system's design requirements, and the ejector's power coverage is extended to 93.25%. This research establishes a theoretical foundation for fuel cell hydrogen supply technologies in high-dynamic-load application scenarios.

The practical application of fuel cell vehicles is hindered by poor performance and durability of proton exchange membrane fuel cells (PEMFCs), especially at low Pt loading and low relative humidity conditions, owing to limited proton transport properties and electrochemical reaction surface areas of conventional electrode design. This study proposes a patterned design with ordered ionomer micropillars as an alternative electrode structure to enhance the performance and durability of PEMFCs with low Pt loading (0.1  mg cm-2) across a wide range of humidities. Compared to conventional electrodes, the patterned electrodes achieved up to 29% peak power density improvement under low-humidity (50% relative humidity) conditions. Pore-scale multiphysics modeling demonstrated that the ionomer micropillars inside the patterned electrodes transform the distorted transport pathways of conventional, randomly distributed catalyst layers into ordered, short-range channels for rapid oxygen and proton delivery, which significantly enhances the oxygen reduction reaction rate. Moreover, the patterned electrodes exhibited improved durability, showing lower performance loss compared to conventional electrodes after accelerated stress testing across a wide range of humidities. Strategic redesign of the ionomer micropillars demonstrated that the H1.8W0.38 electrode (with a micropillar height of 1.8  μm and width of 0.38  μm), with its increased penetration depth and appropriately adjusted width, achieves optimal performance by minimizing oxygen and proton transport resistances by 6% and 14%, respectively, compared to the pre-optimization (H1.5W0.38) electrode, under low-humidity (50% relative humidity) conditions. The proposed patterned electrode architecture offers a promising approach for advancing the development of high-performance, cost-effective, and durable fuel cells.

Gas diffusion layer (GDL) plays an important role in proton exchange membrane fuel cells (PEMFCs), where its microstructure directly determines the transport properties and overall performance. Traditional methods lack the ability to efficiently reconstruct the microstructure of the GDL and to finely characterize its transport properties, thus failing to effectively guide structural design in engineering applications. To address this issue, an AI-driven Reconstruction and Structure-resolved Transmission Modeling (ADR-STM) framework is developed. This closed-loop framework integrates real microstructure reconstruction, automated segmentation, multiphysics numerical simulation, and quantitative analysis. The ADR-STM framework can efficiently reconstruct real microstructure and achieve phase segmentation, which reduces the consuming time by 96% compared with traditional method. A reconstruction algorithm is developed to alter porosity without changing the fiber orientation distribution. Moreover, the lattice Boltzmann method in this framework can simulate electrical and thermal conduction and gas transport in the GDL, respectively, enabling direct computation of bulk electrical resistance, effective thermal conductivity, and gas permeability. The ADR-STM framework provides an alternative approach to bridge the real microstructures and transport properties, thus offering guidance for GDL designs.

Porous carbon is often used to support platinum (Pt) catalysts in proton exchange membrane fuel cells (PEMFCs). Nanoscale pores in the carbon support improve the performance of PEMFCs by inhibiting perfluorosulfonic acid (PFSA) film coverage on the Pt particles. However, the nanopores parameters significantly affect morphologies of PFSA and water molecules inside pores, thereby affecting oxygen transport. Hence, understanding molecular behaviors in pores is essential for optimizing pore structures. In this study, the morphologies of PFSA and water molecules and the oxygen transport resistances inside the pores of porous carbon were analyzed for different pore diameters and depths via molecular dynamics (MD) simulations. The results demonstrated that the looser arrangement of PFSA molecules in the entire mesoporous carbon and the expanded accessible volume enhancing oxygen transport pathways are the main causes of decreased oxygen transport resistance for larger diameter mesopores. The effect of pore depth on oxygen transport resistance results from the competition between the elongation of oxygen transport pathways and the reduced dense PFSA layer density around the Pt surface. Moreover, oxygen transport is dominated by the PFSA density near the Pt surface rather than the overall PFSA density within mesopores. In particular, the formation of a dense PFSA layer leads to a significant increase in oxygen transport resistance. Therefore, increasing the pore diameter and reducing the pore depth while avoiding the formation of a dense film on the Pt surface is preferred to enhance oxygen transport.

Hydrogen ejectors in proton exchange membrane fuel cell (PEMFC) systems can reliably recover hydrogen without parasitic power, making them an ideal alternative to pumps. However, improving fuel utilization, enhancing ejector efficiency, and extend the ejector’s operating range are important issues currently faced by PEMFC systems. In this study, a comprehensive evaluation system including four indicators (hydrogen entrainment ratio (ωH 2 ), ejector outlet relative humidity ( RH c ), ejector outlet temperature ( T c ), and ejector efficiency ( η ejector )) are established based on the operating conditions and structural parameters characteristics of the ejector. The influence of secondary flow relative humidity ( RH s ) and secondary flow temperature ( T s ) on ejector performance are analyzed. The results show that the hydrogen entrainment ratio decreases as the secondary flow relative humidity or temperature increases. RH c increases and T c decreases as the gas mixture reaches the outlet of the ejector. Ts should be controlled below 353K and RH s below 57% to prevent condensation of water vapor at the ejector outlet, at which point the ejector efficiency is 55.82%. To improve the ejector efficiency, the secondary flow temperature can be further increased, but the secondary flow relative humidity must be controlled accordingly. A theoretical basis for the development of high-temperature PEMFC is also provided.

Accurate economic assessment of proton exchange membrane fuel cell (PEMFC) vehicles is essential for optimizing control strategies in the PEMFC industry, which is largely driven by the need to reduce costs. Traditional data-driven approaches have focused on reconstructing typical driving cycles from real-world speed data, often overlooking the intensity and acceleration of these cycles. These factors are crucial for water and heat management in PEMFCs and can lead to inaccurate estimates of hydrogen consumption. This paper introduces a novel algorithm for typical driving cycles reconstruction based on real-world data, named the improved twodimensional Markov Chain Monte Carlo (2D MCMC) approach using Metropolis-Hastings (M- H) sampling. The approach innovatively encodes the integration of real-time vehicle speed and acceleration sequences into a hierarchical 2D state transition probability matrix. To optimise both accuracy and computation time, the M- H based sampler is newly introduced to generate typical driving cycle without the computational burden of multiplying large matrices. Moreover, by integrating the agglomerative nesting (AGNES) alongside a comprehensive evaluation system that incorporates simulation and bench testing, the proposed approach effectively weights real-world route conditions in the economic assessment. Case studies involving 10 PEMFC hybrid buses in Shanghai, China, validate the effectiveness and robustness of the proposed method. Comparative analyses show that the relative errors in hydrogen consumption per 100 km between the reconstructed and real-world driving cycles are within 1.20-3.01% for all ten buses in Shanghai, with computation times reduced by up to 12.60% compared to the existing methods.

Mesoporous carbon supports mitigate platinum(Pt)sulfonic poisoning through nanopore-confined Pt deposition,yet their morphological impacts on oxygen transport remain unclear.This study integrates carbon support morphology sim-ulation with an enhanced agglomerate model to establish a mathematical framework elucidating pore evolution,Pt utili-zation,and oxygen transport in catalyst layers.Results demonstrate dominant local mass transport resistance governed by three factors:(1)active site density dictating oxygen flux;(2)ionomer film thickness defining shortest transport path;(3)ionomer-to-Pt surface area ratio modulating practical pathway length.At low ionomer-to-carbon(I/C)ratios,limited active sites elevate resistance(Factor 1 dominant).Higher I/C ratios improve the ionomer coverage but eventually thicken ionomer films,degrading transport(Factors 2-3 dominant).The results indicate that larger carbon particles result in a net increase in local transport resistance by reducing external surface area and increasing ionomer thickness.As the proportion of Pt situated in nanopores or the Pt mass fraction increases,elevated Pt density inside the nanopores exacerbates pore blockage.This leads to the increased transport resistance by reducing active sites,and increasing ionomer thickness and surface area.Lower Pt loading linearly intensifies oxygen flux resistance.The model underscores the necessity to optimize support morphology,Pt distribution,and ionomer content to prevent pore blockage while balancing catalytic activity and transport efficiency.These insights provide a systematic approach for designing high-performance mesoporous carbon catalysts.

Pt-Co alloy catalysts are widely used in proton exchange membrane fuel cells (PEMFCs) due to their high mass activity; however, Co dissolution remains a major challenge. Leached Co ions exchange with the sulfonic acid sites of the ionomer, reducing proton conductivity and O2 permeability, ultimately leading to catalyst degradation and performance loss. In this study, on-line inductively coupled plasma mass spectrometry (ICP-MS) was employed to systematically investigate the Co leaching behavior of ordered intermetallic Pt-Co alloy catalysts under dynamic electrochemical conditions. Our findings reveal that Co dissolution occurs over a broad potential range, with pronounced leaching during potential transitions. Notably, the upper potential limit plays a crucial role in governing the extent of Co leaching under cycling conditions. Accelerated stress tests demonstrate a two-stage leaching process, consisting of an initial rapid dissolution phase followed by gradual stabilization. Failure analysis indicates that Co atoms at the catalyst surface preferentially undergo dealloying and dissolution, contributing to structural degradation. Moreover, at higher upper potential limits, Co dissolution closely resembles that of Pt, with significant leaching occurring during the reduction potential scan. These findings underscore the electrochemical instability of Pt-Co alloy catalysts and highlight the need for advanced structural stabilization strategies to enhance their long-term durability in PEMFC applications.

Proton exchange membrane fuel cell (PEMFC) degradation involves complex interactions among multiple aging mechanisms, presenting challenges for accurate voltage prediction due to nonlinear dynamics and stochastic features. This study proposes a framework combining optimized data decomposition and multimodal hybrid neural network Res-CGA for PEMFC voltage degradation prognostics. It develops an optimized decomposition technique integrating the slime mold algorithm with variational mode decomposition to iteratively decompose aging voltage data into time-varying intrinsic mode functions (IMFs). Subsequently, entropy-driven k-means clustering consolidates IMFs into three cooperative IMFs: high-frequency, low-frequency, and trend sequences. The dedicated hybrid neural network Res-CGA (incorporating residual connections, convolutional operations, gating mechanisms, and attention mechanisms) is developed to learn decomposed features. The proposed method demonstrates superior capability in resolving multi-factor superimposed aging features and strong nonstationary. It effectively extracts trend features, long-period patterns, and high time-varying characteristics from aging voltage data, thereby achieving high-precision predictions. Validated across four independent PEMFC aging datasets, the framework achieves satisfactory accuracy with normalized root mean square error (NRMSE) < 0.05 and normalized mean absolute percentage error (NMAPE) < 10 %, outperforming benchmark machine learning and deep learning models by >3.94 % NRMSE and >1.68 % NMAPE reductions. Ablation studies confirm the critical contributions of the optimized decomposition module, the Res-CGA model, and the highfrequency sequence re-decomposition in the proposed method. These modules improve precision of the proposed method, achieving >0.41 % NRMSE and >0.20 % NMAPE reductions. This work establishes a novel technical pathway for PEMFC performance degradation prediction. Future research will focus on integrating this framework into prognostic and health management systems to validate its engineering applicability.

Transformer and its variants show significant potential for predicting proton exchange membrane fuel cell performance degradation, enabling accurate capture of degradation patterns to inform control strategies and extend lifespan. However, despite advancements, their applicability to commercial high-power fuel cells remains unclear, as existing researches focus primarily on small-scale laboratory stacks. Addressing this gap, this study investigates a 60 kW commercial fuel cell system under two 1000-hour aging test modes with different hydrogen supply conditions (ambient vs. low temperature). A characteristic current-based data extraction method was employed for the raw data associated with each mode. Three representative characteristic currents were selected based on the current distribution of dynamic load cycles for data extraction. Following the preprocessing, aging datasets for three characteristic currents were obtained. Single cell voltage was selected as the aging feature parameter to construct Transformer and four variants (Informer, Half-Transformer, Half-Informer, and Auto-former) for degradation prediction. Comparative analysis revealed Autoformer's superior aging voltage prediction accuracy. Its robustness was further validated under multi-step prediction, training set missing, and multivariate input scenarios, maintaining high accuracy across diverse conditions. The deviation of absolute prediction errors at the 80 % and 90 % cumulative distribution levels remained below 10 mV. These results demonstrate Autoformer's strong potential for integration into fuel cell control systems, offering promising applications in health management to enhance practical value.

Significant performance enhancement in nanofiber-based PEMFCs featuring highly O 2 permeable ionomer films, highly proton-conductive ionomer fibers, and increased active sites.

This review paper aims to explore the degradation mechanism and control strategies employed for onboard fuel cells operating under changing load conditions. The performance of fuel cells is crucial for various applications, particularly in vehicles, where factors like road conditions and driving habits lead to fluctuating power demands. While current research examines the impact of operating conditions on fuel cell durability and discusses various mitigation approaches, there remains a critical gap in correlating degradation mechanisms with control system strategy optimization. Through systematic analysis, this review identifies three primary degradation mechanisms under variable loads: potential cycle, parameter alternation, and localized gas starvation-each influenced by changes in power demand. These mechanisms are found to directly correspond to the optimization requirements of three key systems: Power electrical system, Water and thermal management systems, and Reactant system, respectively. The complex interplay of parameters such as temperature, pressure, and humidity can fluctuate beyond the optimal range, resulting in irreversible damage to fuel cell components. Such parameter variations cause irreversible damage to fuel cell components. Traditional control methods have limitations in handling multiple parameter changes. The integration of artificial intelligence (AI) with conventional control methods shows better results. This combined approach offers improved adaptability and efficiency for dynamic operations. Understanding degradation mechanisms is essential for control system optimization. The implementation of integrated control strategies can extend fuel cell life and enhance their performance under variable load conditions.

To obtain the temperature sensitivity characteristics of the proton exchange membrane fuel cell under different degradation levels, accelerated durability tests and temperature sensitivity tests of the fuel cell are first performed, and the temperature characteristics under different degradation states are analyzed by electrochemical impedance spectroscopy and polarization curve. Besides, an adaptive state-of-health transient thermal model of fuel cells is developed to investigate the effect of operating temperature on the internal gas concentration and hydrothermal distribution characteristics of the fuel cell from a mechanistic perspective. The experimental results of temperature sensitivity validate that the adaptive state-of-health transient thermal model can effectively monitor the steady and transient performance of the fuel cell. Subsequently, a mathematical model is proposed to describe the relationship among the load current, state of health, and optimal temperature using the temperature sensitivity test data and the transient thermal model of the adaptive state of health, which provides important insights into the optimal temperature range tailored to the current state of health of fuel cells to ensure efficient and stable operation of fuel cells all the time and contributes to the fine design of an optimal degradation adaptive temperature control strategy.

Large-area ultra-thin patterned membrane with finely ordered micropillars is fabricated with structured poly (dimethyl siloxane) mold as template. The performance and stability are evaluated for fuel cell automotive application. It shows 38 mVs higher performance than its flat counterpart (0.66 V versus 0.622 V at 2000 mA cm(-2)) in a 25 cm(2) MEA with 0.15 mg(Pt) cm(-2) cathode loading, 100 % humidified air. Five pieces of 11-mu m-thickness patterned membrane are further applied in a 300 cm(2) stack. With 0.15 mg(Pt) cm(-2) cathode loading, the stack exhibits a high performance of 0.634 V at 2000 mA cm(-2) under real fuel cell vehicle operation conditions. To our best knowledge, it is not only the first time the patterned membrane is utilized in a full-size stack for fuel cell vehicles, but also the highest performance of patterned membrane ever reported at such low Pt loading. The reasons for the exceptional performance of the patterned membrane could be due to the direct proton transport path formed by the pillars, resulting in a more effectively utilized catalyst layer. After humidity and voltage cycling, the membrane's three-dimensional structure shows excellent stability, neither micropillars collapsing nor performance decaying observed, demonstrating a great potential for industrial applications.

Proton exchange membrane fuel cell (PEMFC) with its zero emission and high efficiency is gradually being applied for clean transportation in China. In the quest to achieve higher economic efficiency of PEMFC hybrid vehicles, data-driven modelling methods are being developed in response to the complicated physicochemical phenomena of PEMFC systems. However, there is little research detailing the importance of balance of plants (BOP) features of the hydrogen anode, air cathode and cooling subsystems regarding PEMFC system efficiency at different driving styles. Furthermore, most research applies neural networks based on simulation and bench data rather than dynamic vehicle operation data, which leads to low robustness and unreliable practical results. Accordingly, this paper provides a novel application of the combination of a power-related feature extraction method, an unsupervised dimension reduction method, an unsupervised cluster method and an ensemble learning method, named PCA-Kmeans-XGBoost, to explore the relationship among controllable BOP features, PEMFC system efficiency and driving styles using real-time vehicle datasets. A case study of a PEMFC logistics vehicle is conducted based on the data at the size of 312,641 running in Shanghai in November 2022. The economic analysis explores the clustered driving styles with high power ranges and frequent power requests take 40.9

As a high efficiency hydrogen-to-power device, proton exchange membrane fuel cell (PEMFC) attracts much attention, especially for the automotive applications. Real-time prediction of output voltage and area specific resistance (ASR) via the on-board model is critical to monitor the health state of the automotive PEMFC stack. In this study, we use a transient PEMFC system model for dynamic process simulation of PEMFC to generate the dataset, and a long short-term memory (LSTM) deep learning model is developed to predict the dynamic performance of PEMFC. The results show that the developed LSTM deep learning model has much better performance than other models. A sensitivity analysis on the input features is performed, and three insensitive features are removed, that could slightly improve the prediction accuracy and significantly reduce the data volume. The neural structure, sequence duration, and sampling frequency are optimized. We find that the optimal sequence data duration for predicting ASR is 5 s or 20 s, and that for predicting output voltage is 40 s. The sampling frequency can be reduced from 10 Hz to 0.5 Hz and 0.25 Hz, which slightly affects the prediction accuracy, but obviously reduces the data volume and computation amount.

To improve the utilization of electric energy and heat energy of the fuel cell stack, it is necessary to explore the composition and availability of stack energy under different operating conditions. In this paper, a dynamic stack model that can quantitatively calculate the energy flow and thermal voltage is established, and the actual operation data of a 90 kW stack is used as model input and parameter identifications. Besides, the energy flow and thermal voltage of the stack are simulated and compared under different operating conditions, including different cathode inlet relative humidity, stack temperature, pressure, and oxygen excess ratio. According to the analysis results, for the 90 kW fuel cell stack, the gas emissions can take away about 30 kW of waste heat under the maximum power condition, which accounts for as considerably large as 28% of the heat generated by the stack. Moreover, it is indicated that the self-humidifying stack has a relatively small thermal voltage of about 1.24 V, and the thermal voltage increases by about 0.02 V for every 10% increase in the cathode inlet humidity after the cathode humidity reaches saturation. These conclusions are of great economic significance for improving the electric energy and waste heat utilization of water-cooled PEMFC.

车载质子交换膜燃料电池催化层的孔结构识别效率低、精度差且实验要求严格,无法适应日趋规模化的行业发展体系,因此针对该问题,本文提出基于遗传粒子群的最大化类间方差(GA-PSO-Otsu)优化算法,实现对催化层扫描电镜图孔径分布和孔隙率高效、精确且自适应的识别和测算.首先,协同引入高斯卷积核与二值化阈值最大化类间方差,有效降低噪声和手动调参对精度和效率的影响,实现自动化去噪和孔结构识别;其次,进一步提出遗传粒子群算法,有效解决传统方法遍历参数耗时长和易陷入局部优化的问题,兼具高精度和高效率的优点;最后,通过对催化层结构和灰度分布差异明显的扫描电镜图的对比实验验证,表明该方法具备良好的鲁棒性、自适应性和实用性,与遍历所有参数的传统Otsu算法的孔隙率误差小于0.5%,测算耗时降低约26.2%.

Paper-type gas diffusion layer (GDL) plays the role of structural support, mass transport, electron and heat transfer in the fuel cell. The deformation and fracture behavior of papertype GDL directly determines its physical characteristics. In this paper, we obtained the distribution function of fiber length among intersections using statistical analysis. Based on the bending theory and fracture analysis, the deformation model and the maximum allowable pressure criterion of paper-type GDL are established. The deformation modulus is related to the modulus and the initial volume fraction of carbon fiber, with an ideal value of 1.4 MPa. The tensile fracture is the main form of fiber fracture. The maximum allowable pressure is related to the expected fracture probability, the strength and the initial volume fraction of carbon fiber. The residual strain increases and the deformation modulus decreases with load range, which means that the strength of the paper-type GDL is weakened when residual strain occurs. (c) 2023 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.

Data-driven modelling methods are being developed in the quest to achieve more accurate performance prediction of protons exchange membrane fuel cell (PEMFC) systems in response to their complicated physicochemical phenomena. However, there is little research in this field detailing the pre-processing and selection of balance of plants (BOP) features for the input layer of system performance prediction at different current densities. Furthermore, most of the previous research applies neural networks based on simulation data rather than real-time bench or vehicle operation datasets which leads to low robustness and unreliable practical results. This paper details the application of a novel algorithm denoted XGBoost-Boruta, which utilises the combination of an ensemble learning approach and a wrapping approach, to improve the robustness of feature selection and to increase the accuracy and robustness of PEMFC system performance prediction. By introduction of the Z score and shadow features to eliminate the randomness of conventional ensemble learning methods, seven key controllable BOP variables of the hydrogen anode, air cathode and cooling subsystems are selected as the original input variables to determine their dependency on the stack voltage. Two case studies are presented for verification and validation of the proposed algorithm based on the real-time dataset of bench experimental data and data obtained from heavy truck operation at current densities ranging from 100 to 1500 mA/cm2. The feature selection strategy, based on the proposed XGBoost-Boruta algorithm, largely decreases the RMSE by 23.8% and 14.1% and the R2 increases by 0.06 and 0.04 of both the bench experimental and the heavy truck validation datasets respectively.