This study establishes a three-dimensional stress-induced hydrogen diffusion model by using finite element analysis and simulates the plastic strain at the defects and its influence on the diffusion and aggregation behaviour of hydrogen atoms under large deformation conditions by using the nonlinear hardening function. The results show that the plastic strain significantly increases the local stress, the equivalent plastic strain, and the hydrogen concentration in the corrosion defect region. Plastic strain increases the overall von Mises stress in the specimen, while the significant increase in hydrostatic stress and equivalent plastic strain is mainly concentrated in the central region of the corrosion defect. In the absence of plastic strain, the diffusion of hydrogen atoms is mainly driven by the concentration gradient, and the flow direction is from the region of high concentration to the region of low concentration, which ultimately leads to a steady state of hydrogen atom concentration inside the specimen. However, when plastic strain exists, the diffusion behaviour of hydrogen atoms is driven by the hydrostatic stress gradient, which is manifested by the aggregation of hydrogen atoms towards the stress concentration region （i.e., at the corrosion defect）. In addition, narrow and deep corrosion defects are more likely to lead to local accumulation of hydrogen atoms, while the shape of the defect has relatively little effect on the distribution of hydrogen concentration.

In order to achieve the recycling of nutrients in photo-fermentation hydrogen production tailing liquid, the adsorption capacity of zeolite on NH₄⁺ and K⁺ in photo-fermentation hydrogen production tailing liquid was investigated using natural zeolite as adsorbent and kinetic and thermodynamic analyses were carried out. It is found that the adsorption of zeolite on NH₄⁺ and K⁺ in photo-fermentation hydrogen production tailing liquid exhibites a tendency of accelerated adsorption, then gradually reaches balance, which is attained at 2 h and 3 h of adsorption, respectively. When adsorption time is 2h and adsorption temperature is 35 ℃, zeolite demonstrates stronger adsorption capacity for NH₄⁺ with an adsorption of 1.82 mg/g. When adsorption time is 3 h and adsorption temperature is 45 ℃, zeolite exhibites its maximum adsorption capacity for K⁺, reaching 7.11 mg/g. The quasi-secondary kinetic model provides a superior description for the adsorption process. The adsorption process of zeolite on NH₄⁺ and K⁺ in the tailing liquid of photo-fermentation hydrogen production is non-spontaneous. The adsorption of NH₄⁺ by zeolite is characterised as an exothermic entropy reduction reaction, while the adsorption of K⁺ is identified as an adsorptive entropy increase reaction.

This study employed FLACS software to simulate hydrogen leakage scenarios involving a Toyota Mirai Gen II fuel cell vehicle in underground parking garages, to systematically investigate the impacts of ventilation scheme, vent orientation, and ventilation rate. Results indicate that the combined action of the suspended ceiling and ventilation system completely removes the flammable gas clouds, achieving the highest hydrogen removal efficiency. Upward-exhaust ventilation is more effective than other orientations. At a ventilation rate of 24 air changes per hour （ACH）, the peak equivalent stoichiometric cloud volume is 22.5% lower than that for no ventilation, and the cloud removal time is 43.4% shorter than that at 12 ACH. Further increases in ventilation rate provide limited enhancement in flammable cloud evacuation efficiency.

This study proposes an operational optimization framework for power-to-hydrogen systems that explicitly incorporates the influence of hydrogen price dynamics. Firstly, a fuzzy set is constructed using historical price data to simulate hydrogen price volatility. Subsequently, a seasonal production intensity forecasting mechanism is developed to coordinate hydrogen price volatility with renewable energy variability. Building on this seasonal forecasting, a distributionaly robust optimization model is formulated for the operational scheduling of power-to-hydrogen systems. Case study results demonstrate that the proposed forecasting mechanism enables rational quarterly adjustments of production intensity, and its integration with the distributionaly robust optimization model effectively enhances renewable energy consumption while reducing daily operational costs.

To improve the accuracy of power generation prediction in high penetration photovoltaic microgrids, this paper proposes an improved Markov chain displacement time series prediction method （MFM） to optimize the problems of insufficient feature extraction and inaccurate prediction in the conjugate gradient method （CG） - deep belief network （DBN） combined prediction model （CG-DBN）. Firstly, using Pearson correlation coefficient to analyze the influencing factors of high penetration photovoltaic microgrid power generation; Secondly, taking advantage of the inefficiency of Markov chain displacement time series prediction method, the residual correction process is applied to the CG-DBN prediction model to construct a short-term prediction model for the power generation of MFM-CG-DBN high penetration photovoltaic microgrids; Finally, the MFM-CG-DBN short-term prediction model is used to simulate the power generation data of high penetration photovoltaic microgrids under three types of weather conditions： sunny, cloudy, and rainy. The simulation results show that the proposed short-term prediction model has higher prediction accuracy than the traditional CG-DBN short-term prediction model, and can meet the demand for high penetration photovoltaic microgrid power generation prediction.

A panel heating terminal with photovoltaic direct drive electric heating phase change heat storage wall panel heating terminal is proposed, which uses photovoltaic DC to drive the electric heating film to produce heat and store the heat in the phase change material at the same time. The heat transfer model is established by using the multi physical field coupling software COMSOL, and the effect of adding flat fins （fin spacing, length and thickness） on the heat transfer process of phase change heat storage wall panel under natural convection conditions is discussed. The results show that properly reducing the fin spacing can effectively shorten the complete melting time of PCM, improve the average heat storage rate, and extend the heat release time, but the effect is not obvious. When the fin spacing is less than 25 mm, the total heat storage decreases significantly; Under the same fin spacing, the greater the proportion of fin length to the radial length of PCM, the higher the heat storage rate; When the thickness is 1 mm, the average heat storage rate reaches the maximum of 103.3 kJ/h, which increases by 27.2% compared with the structure without fins.

For the purpose of achieving a scientifically rigorous and methodologically sound clustering partition of photovoltaic clusters, this study incorporates the synergistic effects of meteorological factors and geographical positioning on the clustering of photovoltaic power stations. Employing output characteristics as the basis for subgroup delineation, the paper introduces a K-medoids clustering algorithm based upon an enhanced Manhattan distance metric. The two pivotal dimensions in photovoltaic cluster analysis encompass the robust identification of meteorological features and the precise quantification of inter-station similarity. Accordingly, an integrated approach combining the coefficient of variation method with the rank-sum ratio method is utilized to derive meteorological input features, thereby augmenting the efficacy of feature discrimination. Moreover, the refined Manhattan distance is embedded within the K-medoids framework to comprehensively elucidate the dynamic attributes inherent in the dataset. Three internally validated clustering metrics are leveraged to optimize cluster cardinality, effectively mitigating the dispersion and systematic bias endemic to conventional partitioning approaches while enhancing overall clustering robustness. Subsequent predictive validation is performed based on the established clusters. To resolve the challenge of parameter calibration within the forecasting model, metaheuristic optimization techniques are integrated. Specifically, an improved grey wolf optimization （IGWO） algorithm is deployed to fine-tune the hyperparameters of both the iTransformer and extreme gradient boosting （XGBoost） architectures. The power forecasting results verify the operational efficacy and superior predictive accuracy of the proposed clustering methodology.

This study aims to elucidate the influence of terrain slope on wind pressure distribution and wind load shape coefficients of PV modules through computational fluid dynamics （CFD） simulations using FLUENT software. The Reynolds-Averaged Navier-Stokes （RANS） equations coupled with the shear stress transport （SST） turbulence model were employed to investigate the wind pressure distribution patterns and wind load shape coefficients under various configurations, including different PV module inclination angles, wind direction angles, and terrain slopes. The findings reveal that both wind direction angle and mounting inclination angle significantly influence the wind load magnitude on PV modules across different terrain slopes. While terrain features induce wind bypass effects, these effects gradually diminish with increasing column height. Furthermore, under various slope conditions and mounting inclinations, single-string PV modules consistently exhibit higher shape coefficients compared to PV array configurations. The investigation also demonstrates substantial shading effects within PV arrays, resulting in distinctive wind load characteristics for subsequent rows. Specifically, the shape coefficients experience maximum attenuations of 70% and 66% for the second and third rows of PV modules, respectively. However, the presence of terrain slope significantly mitigates these shading effects, reducing the maximum attenuation by up to 45%.

Under conditions of partial shading caused by cloud cover, dust accumulation, building obstructions, and vessel movement during navigation, photovoltaic arrays deployed on ship decks exhibit multi-peak characteristics in their power-voltage output curves. This renders traditional maximum power point tracking （MPPT） algorithms inadequate for optimal power extraction. To address this issue, this study proposes a hybrid MPPT control method combining improved frilled lizard optimization algorithm with variable step perturbation and observation method （IFLO-IP&O） for shipboard photovoltaic systems. The methodology involves three key innovations： First, the integration of non-standard Circle chaotic mapping, Lévy flight strategy, and triangular random walk strategy into the Frilled Lizard Optimization algorithm creates an enhanced global search capability to locate the neighborhood of the global maximum power point （GMPP）. Second, a variable-step perturbation mechanism enables precise tracking of GMPP. Third, an adaptive switching logic seamlessly integrates these two algorithms into a unified framework, which is then applied to shipboard photovoltaic systems. Simulation results demonstrate that the proposed IFLO-IP&O hybrid method achieves superior performance in convergence speed, tracking accuracy, and oscillation suppression across various illumination scenarios.

To address the challenges of existing deep learning prediction models, such as long training times and a tendency to fall into local optimum, this paper proposes an innovative KOA-driven VMD-CNN-BiGRU-Attention method for short-term photovoltaic （PV） output power prediction. Firstly, the Pearson correlation coefficient is employed to identify key meteorological factors. Then, the grey relational analysis （GRA） method is used to determine the historical similarity days for the predicted days. The Keplerian Optimization Algorithm （KOA） is then used to optimize the parameters of variational mode decomposition （VMD）, which decomposes the output sequences of historical similarity days to generate a high-quality training sample set. Finally, the VMD-CNN-BiGRU-Attention model, driven by KOA, is constructed to achieve accurate PV output power prediction. Practical applications at PV power stations in Yunnan and Gansu show that the model achieves RMSE values of 0.2540 MW and 2.7981 MW, and MAPE values of 0.0234 and 1.1699, respectively. Compared with other combined prediction models, the proposed KOA-driven VMD-CNN-BiGRU-Attention model demonstrates superior ability to capture spatiotemporal features, offering significant improvements in prediction accuracy and stability. These results highlight the broad application potential of the model in PV power generation prediction.