This paper proposes a long short-term memory (LSTM) network to predict the power degradation of proton exchange membrane fuel cells (PEMFCs), and in order to promote the performance of the LSTM network, the ant colony algorithm (ACO) is introduced to optimize the hyperparameters of the LSTM network. First, the degradation mechanism of PEMFCs is analyzed. Second, the ACO algorithm is used to set the learning rate and dropout probability of the LSTM network combined with partial aging data, which can show the characteristics of the dataset. After that, the aging prediction model is built by using the LSTM and ACO (ACO-LSTM) method. Moreover, the convergence of the method is verified with previous studies. Finally, the fuel cell aging data provided by the Xiangyang Da’an Automotive Testing Center are used for verification. The results show that, compared with the traditional LSTM network, ACO-LSTM can predict the aging process of PEMFCs more accurately, and its prediction accuracy is improved by about 35%, especially when the training data are less. At the same time, the performance of the model trained by ACO-LSTM is also excellent under other operating conditions of the same fuel cell, and it has strong versatility.

Durability is the key issue for the proton exchange membrane fuel cell application and its commercialization. Current research usually uses the accelerated stress test to decrease the experiment time, whereas the performance evolution—especially the internal state evolution—under real use may be different from that under the accelerated stress test. In addition, studies rarely report this kind of durability in real decay scenarios. This paper investigates the seldom-reported impact of dry–wet cycles on durability in terms of open circuit voltage (OCV), inner resistance, and hydrogen crossover current at the condition of 20,000 cycles or the equivalent 400 h, while simultaneously running the test for the same time interval in the control experiment. The mechanical and chemical test is independent. Frequent dry–wet cycles make the OCV decay over 14% compared to 6.9% under the normal decay. Meanwhile, the dry–wet cycle helps to alleviate deterioration in terms of the inner resistance decline (61% vs. 37%) and in terms of the hydrogen crossover current increase (−64% vs. 15%). The inner state evolution is irregular and against common sense. The relationship between the crack, platinum transfer, and the moisture which heals the crack is the potential reason for the above-mentioned phenomena. These findings are beneficial to navigating fuel cell storage.

In the development of actual vehicles, manufacturers usually adopt a simplified control strategy to ensure the reliability of the control strategy based on the application scenarios. There are two main working modes for the fuel cell system in fuel cell electric vehicles in China. One is the stepped power type, and the other is the following power type. Therefore, the analysis and comparison of these two typical working modes in the power test of fuel cell electric vehicles helps determine how the fuel cell system works in the actual vehicle and how to choose energy management strategies in different application scenarios. We do the actual tests to explore how the two typical control strategies perform in actual vehicles. These two typical control strategies show different characteristics in the same test. It shows that the energy management strategies should be adopted according to the application scenarios and optimization goals. In the stepped power control strategy, the fluctuation of the fuel cell system and the frequency of starting and stopping are significantly reduced, which is beneficial to the durability of the fuel cell system. Compared with the stepped control strategy, the fluctuation of the output power of fuel cell electric vehicles with the following power control strategy increased significantly. At the same time, a simplified state of charge (SOC) test method is proposed. Due to the particularity of the stepped power control strategy, the change of capacity can be used to replace the change of SOC. For the following power control strategy, the change of electric energy can be used instead of SOC changes.

Mathematical modeling of unitized regenerative fuel cells (URFCs) faces significant challenges in reconciling parameter conflicts between fuel cell (FC) and electrolysis cell (EC) modes. This study establishes a COMSOL-based multi-physics framework coupling water–gas–heat–electric transport for both operational states. The critical factors associated with the model were identified through a systematic sensitivity analysis of structural and operational parameters, including temperature, exchange current density, conductivity, porosity, and flow rates. FC modes exhibited strong sensitivity to exchange current density (27.8–40.5% performance variation) and conductivity of membrane (10.1–35.6%), while temperature degraded performance (−4.2% to −4.0%). Spatial analysis revealed temperature-induced membrane dehydration and accelerated gas depletion at electrodes, thus explaining the negative correlation. EC modes were dominantly governed by temperature (8.6–9.4%), exchange current density (13.0–16.4%), and conductivity (2.5–13.3%). Channel simulations revealed that elevated temperature contributed to enhanced liquid water fluidity, while high flow rates had a relatively limited effect on mitigating species concentration gradients. Parameter optimization guided by sensitivity thresholds (e.g., porosity > 0.4 in FC GDLs, conductivity > 222 S/m in EC modes) enabled dual-mode calibration. The model achieved <4% error in polarization curve validation under experimental conditions, demonstrating robust prediction of voltage–current dynamics. This work resolves key conflicts of URFC modeling through physics-informed parameterization to provide a foundation for efficient dual-mode system design.