Challenges such as adapting to renewable energy power fluctuations and coping with start-stop mechanisms exist during the operation of electrolyzers. Capacity configuration models based on long time scales struggle to accurately capture these dynamic characteristics, which reduces the accuracy of capacity planning for hydrogen production systems.

Taking a hybrid hydrogen production system composed of alkaline (ALK) electrolyzers and proton exchange membrane (PEM) electrolyzers as the research object, this paper proposes a minute-level time-scale capacity planning method for hybrid hydrogen production. First, a minute-level electrolyzer start-stop control model is designed to accurately describe the operating states of the two types of electrolyzers. Second, considering ALK electrolyzers’ poor adaptability to power fluctuations and long start-stop time, we develop a power allocation strategy that prioritizes the stable operation of ALK electrolyzers. Finally, we conduct multi-objective optimization for the capacity planning problem of the hybrid hydrogen production system, and the Pareto solution set of the model is obtained via the augmented ε-constraint method.

Simulation results show that under the condition of 20 MW installed wind power capacity and 20 MW installed photovoltaic capacity, with a total system investment cost of 25 million yuan, the proposed 1-minute time-scale model increases the average daily hydrogen production by 14.7%, reduces the unit hydrogen production cost by 7.5%, and decreases the renewable energy curtailment rate by 63.6% compared with the traditional 15-minute time-scale model. In addition, with 1-minute scheduling accuracy, the ALK/PEM electrolyzer capacity ratio is gradually optimized as investment increases: when the total investment is below 25 million yuan, the proportion of ALK electrolyzers exceeds 90%; when the total investment exceeds 30 million yuan, the investment proportion of PEM electrolyzers rises to 19.4%. In contrast, for the 15-minute time-scale model, the ALK/PEM capacity ratio reaches 4:1 even when the investment is only 20 million yuan. This prematurely increased proportion of PEM electrolyzers not only deviates from practical engineering conditions but also degrades overall system performance, indicating that coarse time-scale scheduling may lead to capacity mismatch.

The stability of the steam inlet system in solid oxide electrolysis cell (SOEC) systems is crucial for enhancing the electrolysis efficiency of the electrolytic stack and prolonging its service life. However, the nonlinear coupling between steam pressure and flow rate imposes high demands on the control strategy.

An experimental platform for the steam inlet system tailored for the 50 kW-class SOEC system was established to investigate the control of steam flow rate error and pressure fluctuation. Based on the collected operation data of the experimental platform, a double-delay deep deterministic policy gradient agent was trained. A multi-objective intelligent control (IC) strategy based on deep reinforcement learning was proposed, aiming to achieve the control goals of the system output flow error not exceeding 3% and pressure fluctuation not exceeding 1 kPa.

The experimental results show that the maximum error of the output steam flow rate under the IC method is 1.4%, and the maximum pressure fluctuation is ±0.67 kPa. While under the PID control method, the maximum error of the steady-state output flow rate of the system is 3.8%, and the maximum fluctuation of the pressure is ±1.25 kPa. Compared with PID control, the IC method reduces the maximum flow rate error by 63.2% and the pressure fluctuation by 46.4%.

The proposed IC method demonstrates significantly superior control performance compared to the PID method.

With the continuous development of the wind power industry toward high power and large capacity, the soaring unit capacity and expanding blade radius of large-scale wind turbines have resulted in increasingly complex spatial distributions of the inflow wind field in front of the turbine and significantly enhanced vertical wind shear effects. The conventional method of characterizing wind conditions using single-point wind speed at hub height can no longer fully reflect the wind speed distribution differences and dynamic patterns within the ultra-large rotor swept area, which is prone to causing issues such as wind power prediction deviations and inadequate adaptability of operation control strategies. To address these challenges, this study proposes a multi-point feature wind speed selection and estimation method for large-scale wind turbines, which can accurately capture key wind speed information in the rotor swept area, overcome the limitations of single-point feature wind speed, and provide data support for the optimal operation of wind turbines.

To achieve the aforementioned research objective, this study adopts a step-by-step technical approach for systematic investigation. Firstly, based on high-precision grid data of the inflow wind field in front of the turbine, a two-stage stepwise feature selection algorithm is proposed, which first carries out preliminary selection with random forest and then implements refined selection via Boruta (RF-Boruta). The Boruta algorithm is employed to conduct significance tests on the feature importance scores output by the random forest model, thereby eliminating redundant and irrelevant wind speed grid points and realizing stable and accurate selection of feature wind speed points within the ultra-large rotor swept area. Secondly, for the selected feature wind speed points, the extended long short-term memory neural network (xLSTM-Mixer) algorithm is introduced, combined with an embedded feature engineering strategy that accounts for input-output delay orders. This strategy fully exploits the temporal correlation and spatial correlation of wind speed sequences, and constructs a unit dynamics-driven ultra-short-term multi-step dynamic estimation model for multi-point feature wind speed points. Finally, to verify the effectiveness and superiority of the proposed method, large eddy simulation (LES) of a 10 MW wind turbine is performed on the SOWFA platform. Meanwhile, 7 typical wind conditions covering the full wind speed range specified in the IEC standards (including complex wind conditions such as shear wind and turbulent wind) are configured for numerical simulation and flow field data collection. The feature selection performance and speed estimation accuracy of the proposed method are comprehensively validated based on the collected high-fidelity data.

The numerical simulation and verification results demonstrate that four representative feature wind speed points, including the hub center, are identified via the RF-Boruta stepwise algorithm. These feature points are arranged at a radius of 50~60 m with an angular interval of 120°, which can effectively cover the key regions of the rotor swept area and fully characterize the spatial distribution features of the inflow wind field. The constructed xLSTM-Mixer model exhibits excellent performance in the multi-point feature wind speed estimation task: the relative error of 80-step-ahead (second-level) prediction for multi-point wind speeds is ≤2.8%, achieving second-level high-precision estimation. Statistical characteristic analysis shows that the Kolmogorov-Smirnov (KS) statistic between the model estimation results and the actual wind speed data is ≤0.2, and the structural similarity index (SSIM) is ≥0.96, indicating a high degree of consistency in both distribution characteristics and structural features between the two datasets. Comparative experiments with mainstream time-series prediction models such as the conventional LSTM and Transformer reveal that the estimation accuracy of the xLSTM-Mixer model is improved by approximately 10%, with distinct advantages in wind speed distribution matching and spatial structure capture capabilities.

The multi-point feature wind speed selection and estimation method proposed in this study effectively breaks through the limitations of conventional single-point feature wind speed, realizing accurate selection and efficient estimation of key wind speed information within the ultra-large rotor swept area. The high-precision multi-point wind speed data provided by this method can reliably support wind power prediction, operation control optimization, and power generation evaluation of wind turbines, helping to enhance the operational stability and energy utilization efficiency of wind turbines. It holds important theoretical significance and engineering application value for promoting the high-quality development of the wind power industry.

Full-spectrum solar energy utilization through spectral splitting offers an effective pathway to improve overall solar energy conversion efficiency by allocating different wavelength bands to suitable energy conversion devices. Linear Fresnel lens-based systems are particularly attractive due to their structural simplicity and scalability. However, the optical efficiency and optical distribution uniformity of such systems are highly sensitive to structural parameters and tracking deviations. The objective of this study is to enhance the optical performance of a linear Fresnel lens-based full-spectrum solar splitting system by optimizing the installation configuration of the photovoltaic (PV) module and by systematically evaluating the influence of incident angle deviations on system performance.

An optical ray-tracing model of the proposed system was established using TracePro software. The model incorporated the geometric configuration of the linear Fresnel lens, spectral splitting characteristics, PV module positioning, and reflective components. To ensure model reliability, a prototype system was constructed, and experimental measurements were conducted under controlled conditions. The simulation results were validated against experimental data by comparing optical efficiency values. Subsequently, a parametric study was performed to investigate the influence of PV module installation height and tilt angle on the optical distribution uniformity and total optical efficiency. In addition, the effects of lateral and longitudinal incident angle deviations, which represent practical solar tracking errors, were quantitatively analyzed. Key performance indicators included total optical efficiency and optical distribution uniformity on the PV surface.

The comparison between simulation and experimental results showed a relative error within 1%, confirming the accuracy and validity of the established optical model. Parametric optimization revealed that when the PV module installation height was set to 540 mm and the inclination angle was 135°, the system achieved optimal optical performance. Under these conditions, the PV surface attained a maximum optical distribution uniformity of 0.86, and the total optical efficiency reached 80.1%. The sensitivity analysis demonstrated that optical performance is significantly affected by incident angle deviations. When the lateral deviation angle increased from 0° to 3.0°, the total optical efficiency decreased from 80.1% to 67.1%, while the optical distribution uniformity declined from 0.86 to 0.79. The influence of longitudinal deviation was even more pronounced. As the longitudinal deviation angle increased from 0° to 30.0°, the total optical efficiency sharply decreased from 80.1% to 26.4%, and the optical distribution uniformity dropped from 0.86 to 0.74. These results indicate that longitudinal tracking errors have a more severe impact on optical performance than lateral deviations, highlighting the importance of precise solar tracking in practical operation.

This study establishes and experimentally validates an accurate optical model for a linear Fresnel lens-based full-spectrum solar splitting system. The results demonstrate that appropriate configuration of PV installation parameters can significantly enhance optical distribution uniformity and overall optical efficiency. Furthermore, the system exhibits strong sensitivity to incident angle deviations, particularly in the longitudinal direction, which must be carefully controlled in engineering applications. The findings provide theoretical support and quantitative guidance for the structural design, parameter optimization, and operational control of full-spectrum solar splitting systems, contributing to the advancement of high-efficiency solar energy utilization technologies.

Temperature fluctuations within the underground cavern have a significant effect on the efficiency of compressed air energy storage power stations and the structural safety of the cavern. Installing the heat exchanger inside the cavern is one of the effective methods to suppress air temperature fluctuations.

A compressed air thermodynamic model that takes into account the heat transfer of internal heat exchangers is established to investigate the effects of different cold and hot water configuration strategies on air temperature and pressure changes inside the cavern.

The results show that by using low-temperature water during the charging phase and high-temperature water during the discharging phase, the internal heat exchanger can effectively suppress the compression heat effect and expansion cooling effect of the air, thereby reducing the range of air temperature fluctuations. Specifically, when cold water (33 ℃) and hot water (90 ℃) are introduced into the heat exchanger during the charging and discharging, respectively, the temperature difference of air can be reduced from 43.9 ℃ without using heat exchangers to below 15.0 ℃. Further analysis indicates that adjusting the cold water utilization period to the latter half of the charging phase and concentrating the hot water utilization time towards the end of the discharging phase can effectively increase the heat transfer temperature difference between the heat exchanger and the air, further reducing the air temperature difference at the end of charging and discharging.

In summary, the reasonable configuration of the operating strategy of the internal heat exchanger, especially the optimization of cold and hot water utilization times, can effectively improve the energy storage capacity and power generation capacity of compressed air energy storage systems.

Under the rigorous guidance of China’s national strategic goals of “dual carbon” (carbon peaking and carbon neutrality), the transformation of energy structures in heavy industries has become a critical priority. In particular, establishing an innovative energy supply system for the alumina digestion process that is predominantly powered by green electricity is essential. This transition is pivotal for promoting a comprehensive zero-carbon transformation, significantly enhancing the utilization efficiency of renewable energy resources, and effectively reducing the operational costs of the system.

To address these challenges, this study constructs a novel green electricity-molten salt synergistic hybrid system, supported by advanced wind and solar forecasting techniques, to supply reliable power for the alumina digestion process. Regarding the methodological framework, a sophisticated hybrid prediction model combining the autoregressive integrated moving average model (ARIMA) and a long short-term memory (LSTM) network is developed to achieve precise meteorological data forecasting. Furthermore, the non-dominated sorting genetic algorithm III (NSGA-III), integrated with a fuzzy satisfaction function, is utilized to conduct a rigorous multi-objective optimization configuration study. This comprehensive simulation covers a continuous period of 1 week (168 h) and evaluates performance across 12 typical operating scenarios to ensure robustness.

The comprehensive empirical results indicate that the prediction accuracy of the proposed model is exceptionally high, with the average coefficient of determination (R²) value for meteorological data exceeding 97.5%, thereby providing reliable data input for system control. The optimized full-equipment configuration demonstrates significant advantages in maintaining a cross-seasonal stable energy supply, achieving an optimal balance among economic viability, environmental impact, and overall energy efficiency. Specifically, the minimum weekly operating cost is recorded at 6.322 9 million yuan, the lowest carbon emission is reduced to 44.97 t, and the maximum effective green electricity rate reaches an impressive 99.67%. Additionally, the configuration of molten salt thermal energy storage equipment effectively smooths the inherent fluctuations of green electricity and drastically reduces reliance on external grid power purchases, resulting in an average green electricity supply proportion of 98.23%. The strong synergistic effect between wind turbine equipment and photovoltaic equipment successfully compensates for the temporal and intensity limitations of single energy sources, significantly improving both the effective green electricity rate and the stability of the system’s energy supply.

The proposed green electricity-molten salt synergistic hybrid system successfully realizes a stable green power supply, providing robust theoretical support for the optimization of low-carbon, low-cost alumina digestion processes powered directly by green electricity.

Hybrid tower structures have emerged as crucial supporting structures for wind turbine generators in low-wind-speed regions. However, the hybrid towers have a complicated construction process, leading to frequent occurrences of defects such as tower step misalignment and structural adhesive deficiency. These defects pose a significant threat to the structural integrity of the tower, thereby impacting the overall reliability and economic efficiency of wind power projects. There is an urgent need for safe construction and efficient operation and maintenance guidance.

Taking the damage of a 4.2 MW concrete-steel hybrid tower as a case, ABAQUS was used to establish finite element models of the tower structure under various defect scenarios, including joint misalignment, insufficient adhesive at joint interfaces, and excessive pad height, to investigate the influence mechanism on structural performance and derive engineering recommendations.

Excessive pad height, insufficient adhesive application, inadequate concrete strength, joint misalignment, and insufficient pretensioning of steel strands collectively contribute to the degradation of the tower structure's safety. Based on the aforementioned research findings, this paper proposes targeted engineering optimization recommendations.

At present, the greenhouse effect is becoming increasingly severe, making it crucial to control CO₂ emissions from fossil fuel combustion. Carbon dioxide capture technology represents both the primary step and the critical pathway, serving as a vital means for reducing carbon emissions in the future.

Based on a novel double-contact carbon capture gas-liquid two-phase absorption bed, and to investigate its enhanced mass transfer performance for ammonia-based carbon capture, this study employs an Eulerian-Lagrangian CFD framework. By integrating dual-film theory into a secondary development of Fluent, a mass transfer model for ammonia-based carbon capture is constructed.

The droplet load exhibits a unimodal distribution along the tower height, being low near the walls and high towards the center. As the liquid-to-gas ratio decreases, the net CO₂ flux increases, leading to a higher droplet load. Significant flow deviation exists at the flue gas inlet, with its severity diminishing as the liquid-to-gas ratio decreases. The uneven droplet distribution causes non-uniformity in CO₂ absorption and concentration field distribution. When the liquid-to-gas ratio decreases from 0.40 m³/m³ to 0.17 m³/m³, the CO₂ capture efficiency drops from 89.19% to 77.29%, a reduction of 13.34%, while the outlet CO₂ molar fraction rises from 1.45% to 3.00%. The overall gas-phase mass transfer coefficient (K_G) remains relatively high below three-quarters of the bed height. A banded region of low K_G forms beneath the nozzle manifold. In the region above the nozzles, K_G gradually decreases with increasing bed height due to insufficient mass transfer driving force and the influence of mass transfer resistance from the gas film side. As the liquid-to-gas ratio decreases, the high K_G zone contracts, and the banded low K_G zone exhibits a parabolic upward trajectory. Regarding the flow field, two high-velocity zones and vortices form within the bed due to the inlet and Venturi effect. The average gas phase velocity in the Z-direction exhibits a bimodal distribution along the bed height, a symmetrical three-segment oscillation along the X-cross section, and a non-monotonic distribution along the Y-cross section. As the liquid-to-gas ratio decreases, the disturbance between the gas and liquid phases intensifies, causing the average gas phase velocity along the Y=0 cross section and the X-cross section to gradually increase.

The double-contact carbon capture gas-liquid two-phase absorption bed exhibits favourable mass transfer characteristics, with the liquid-to-gas ratio exerting a significant regulatory effect on both flow and mass transfer. These findings provide a theoretical basis for optimizing carbon capture equipment.

Against the backdrop of global efforts to address climate change and actively promote the strategic goals of “carbon peak and carbon neutrality”, the clean and low-carbon transformation of the energy system has become a core issue for national development. Accelerating the low-carbon transformation of the coal-fired power industry and precisely reducing carbon emission intensity are key challenges in achieving climate goals. However, there are significant differences in the carbon emission characteristics of different types of coal-fired units, and their carbon emission levels and the emission reduction effects of coupling carbon capture technology have not been clearly compared. To reveal and compare the carbon emission intensities of different types of coalfired units, a carbon emission intensity calculation model applicable to different types of coal-fired units coupled with carbon capture and storage (CCS) systems was constructed.

The carbon emission intensities of typical coal-fired units such as 300 MW, 600 MW, 1 000 MW, double-reheat and IGCC at different load rates, as well as the carbon emission intensities after coupling with CCS systems, were compared and analyzed.

The research results show that at higher load rates, IGCC units have a lower carbon emission intensity, reaching 703 g/(kW·h) at 100% load rate, while the 300 MW unit has the highest carbon emission intensity, reaching 812 g/(kW·h). When the load rate decreases, the carbon emission intensity of the IGCC unit increases rapidly, reaching 948 g/(kW·h) at 50% load rate. The double-reheat unit has the lowest carbon emission intensity at 50% load rate, which is 781 g/(kW·h). CCS technology has a strong carbon emission reduction capacity and is an important means for the low-carbon transformation of coal-fired power. At 100% load rate, a 50% carbon capture rate can reduce the carbon emission intensities of 1 000 MW units, double-reheat units and IGCC units by 334, 329 and 295 g/(kW·h) respectively. Similarly, at a 50% load rate, a 50% carbon capture rate can respectively reduce the carbon emission intensity of 1 000 MW units, double-reheat units and IGCC units by 352, 358 and 379 g/(kW·h).

This study, through the construction of analytical models and systematic comparisons, quantitatively reveals the compound influence mechanism of the technical route of coal-fired units, operating load rate, and CCS coupling strategy on carbon emission intensity. In future power systems with a high proportion of renewable energy, coal-fired units will undertake more peak shaving and frequency regulation tasks. Quantifying the carbon emission differences of coal-fired units not only helps optimize the development path of low-carbon transformation in coalfired power, but also provides solid theoretical support and a decision-making basis for achieving the “dual carbon” goals.

This study aims to satisfy the growing demand for peak-load regulation in power systems and low-carbon hydrogen production.

A detailed numerical model for the coupled ammonia-hydrogen combustion and decomposition process is established by employing ammonia as an energy storage and hydrogen carrier medium. The model systematically investigates the influences of different burner configurations, such as conventional burners, single-layer porous burners, double-layer porous burners, and staged burners, as well as the inlet ammonia velocity within the decomposition zone on NO_x emission characteristics and ammonia decomposition efficiency. By integrating heterogeneous catalytic kinetics of the Ni-Pt/Al₂O₃ catalyst with porous-medium resistance and heat-transfer models, the simulation framework captures the complex thermo-chemical interactions within the integrated reactor. The reliability of the numerical model is validated through comparison with experimental data reported in the literature, showing an average absolute error of less than 4.4%, which confirms its capability to accurately predict the coupled combustion-decomposition behavior.

The simulation results reveal that the endothermic ammonia decomposition process significantly alters the thermal field within the reactor. The strong heat absorption associated with catalytic decomposition reduces the peak temperature in the combustion zone, thereby effectively suppressing the formation of thermal NO. Although the concentration of N₂O exhibits a slight increase (approximately 7×10^–5%), the overall NO_x emissions are substantially reduced due to the dominant decrease in NO formation. All four burner configurations can achieve an ammonia decomposition rate up to 99.99%. However, notable differences exist in the spatial distribution of regions with high decomposition rates and in the associated emission characteristics. Specifically, the staged burner demonstrates strong capability in NO_x mitigation because of the distributed combustion strategy. Nevertheless, the secondary injection of relatively cold ammonia leads to a delayed initiation of the decomposition reaction, which may influence the system stability under certain operating conditions. The double-layer porous burner exhibits superior thermal storage capacity, enabling sustained catalytic activity. However, localized high-temperature zones within the porous matrix tend to promote the formation of NO. In contrast, the single-layer porous burner provides a more balanced thermal environment, achieving an optimal compromise between NO_x suppression and efficient heat supply for ammonia decomposition, thus demonstrating the most favorable integrated performance. Further parametric analysis indicates that increasing the inlet ammonia velocity in the decomposition zone enhances convective heat transfer and strengthens the heat-absorption effect of the decomposition reaction. As a result, the combustion temperature is further reduced, leading to a more pronounced decrease in NO formation compared with the slight increase in N₂O. Consequently, the overall NO_x emissions continue to decline with the increasing inlet ammonia velocity. Notably, even at a relatively high inlet ammonia velocity of 10 m/s, the ammonia decomposition rate remains above 90%, indicating robust catalytic performance under intensified flow conditions.

This work elucidates the thermal-chemical synergy mechanism underlying ammonia-hydrogen combustion-decomposition integration. It identifies the single-layer porous burner as the most suitable configuration for power-generation-side peak-load regulation scenarios. The findings provide a solid theoretical foundation and valuable engineering guidance for the integrated design of ammonia energy storage, hydrogen production, and ultra-low-NO_x combustion systems.

Laser-induced breakdown spectroscopy (LIBS) holds significant potential for application in the field of coal property analysis, due to its advantages of eliminating complex sample pretreatment, enabling multi-parameter synchronous detection, and offering rapid analysis. However, discrepancies in spectral responses exist among different instruments. These discrepancies cause severe accuracy degradation when a quantitative model trained on spectra acquired by a master instrument is applied to slave instruments. Therefore, this study constructed cross-instrument LIBS quantitative analysis models of coal property by integrating TrAdaBoost transfer learning with various machine-learning algorithms.

Two LIBS-based coal analyzers were designated as the master and slave instruments respectively, and LIBS spectra were collected from different numbers of coal samples on both devices. Random forest (RF), support-vector regression (SVR), and their TrAdaBoost-enhanced counterparts (TrA-RF and TrA-SVR) were employed to build quantitative analysis models. Model performance was evaluated by predicting the coal properties of unknown coal samples on the slave instrument.

The results indicated that both TrA-RF and TrA-SVR models significantly outperformed their non-transfer counterparts. The TrA-RF model achieved the highest accuracy for calorific value, ash content, and carbon content. Compared with RF model, the mean absolute errors decreased from 1.390 MJ/kg, 4.774 %, and 3.826 % to 0.654 MJ/kg, 2.338%, and 1.927%, respectively. TrA-SVR model yielded the highest accuracy for volatile matter prediction. Compared with the SVR model, the mean absolute error decreased from 2.722% (SVR) to 2.524%.

These findings demonstrate that coupling transfer learning with an appropriate base learner markedly enhances the adaptability of LIBS-based coal property models across different instruments.

Traditional time-series forecasting methods often struggle to simultaneously capture cross-scale nonlinear fluctuations and long-range temporal dependencies, which leads to limited accuracy in short-term electricity price prediction for spot markets, especially when prices exhibit spikes, volatility clustering, and pronounced non-stationarity.

To address these challenges, this study proposes a short-term electricity price forecasting framework based on a fused global-residual Mamba model that combines series decomposition, dual-branch selective state-space encoding, and global residual learning to strengthen representation power and improve training stability. First, a moving average filter is applied to the normalized electricity price sequence to decouple it into a trend component and a residual component, separating relatively stable low-frequency movements from high-frequency stochastic variations. The decomposed sequences are concatenated along the temporal dimension and mapped through a high-dimensional embedding layer to obtain a richer latent representation capable of characterizing complex market dynamics. To better reflect the multi-factor formation mechanism of spot prices, the model also incorporates exogenous variables, such as regional load and weather-related information (e.g., temperature and meteorological conditions). Building on these inputs, a parallel dual-branch Mamba encoder is designed to extract local-to-global dynamic features from complementary perspectives. The variable-correlation branch focuses on learning time-varying interdependencies between electricity prices and exogenous drivers, explicitly modeling cross-variable coupling and market co-movements. In parallel, the feature-interaction branch targets nonlinear transformations and interactions within the embedded feature space; by permuting tensor dimensions so that selective scanning operates along the embedding dimension rather than only along time, it uncovers abstract interaction patterns that conventional temporal scanning may overlook. To integrate heterogeneous information from both branches, their outputs are concatenated and passed to a global residual learning module, which performs additive fusion between the fused representations and the original embedded input. This global residual pathway provides a stable channel for information flow, alleviates gradient degradation in deeper state-space architectures, and enhances the model’s ability to capture multi-scale patterns by preserving original signals while enriching them with learned cross-variable and cross-feature dynamics. For robust performance and reduced manual effort, Bayesian optimization is applied under a time-series cross-validation (TS-CV) setting to tune key hyperparameters, while training further adopts learning-rate scheduling and early stopping to improve efficiency and stability.

Experiments on real operational data from the Australian Energy Market Operator (AEMO) for 24-hour-ahead forecasting demonstrate clear performance gains: the proposed Residual Mamba reduces RMSE by 29.22% relative to the baseline Mamba model and by 35.33% relative to LSTM, confirming superior accuracy and robustness.

Ablation results further highlight the essential role of the series decomposition module, the importance of the variable-correlation branch in the dual-branch design, and the effectiveness of global residual connections in stabilizing training and improving feature expression for highly volatile spot-market price series.

High-efficiency film cooling technology is an important means of increasing the turbine inlet temperature of gas turbines, and how to obtain the optimal film cooling hole and flow channel geometries has become a critical engineering issue to enhance the film cooling effectiveness of hot-section components in gas turbines.

In this study, the hole geometry of film cooling holes was parameterized, wherein relative coordinates and angles were adopted as the input parameters. Sampling was performed within the selected ranges of input parameters, and parameter optimization was conducted by coupling the BP neural network with the genetic algorithm, with the objective of maximizing the film cooling effectiveness of the film cooling holes. The influences of the shape and inclination angle of the film cooling hole channel on the film cooling effectiveness were investigated.

Compared with cylindrical holes, the film cooling hole with the optimal configuration has a bigger spanwise width, smaller edge expansion angles on both spanwise sides, the same streamwise length as the cylindrical hole, and small-angle protrusion at the trailing edge. Under the condition of the same spanwise width and smooth flow channel, the area-averaged film cooling effectiveness of the optimal configuration is 18.28% higher than that of the dustpan-shaped hole. For the optimal configuration, the area-averaged film cooling effectiveness of the smooth flow channel is 5.3% higher than that of the unsmooth case. The optimized film cooling hole suppresses the trend of the mainstream entraining the cooling flow from both sides below, and alters the rotation direction of the kidney vortex pairs. Specifically, under the condition of the blowing ratio of 1 and hole inclination angles of 30°, 45° and 60° respectively, the area-averaged film cooling effectiveness of this configuration is 814.6%, 1 002.4% and 772.7% higher than that of the cylindrical hole.

The novel film cooling holes finally optimized in this study can reduce the intensity of kidney vortex pairs, enhance the wall adherence of coolant, and simultaneously delay the damping of film cooling effectiveness on the flat plate downstream of the film cooling hole. The optimized configuration maintains high film cooling effectiveness on the flat plate wall even in case of an increasing hole inclination angle, thus exhibiting broader adaptability to hole inclination angles. The optimal configuration with a smooth flow channel obtained in this study has certain engineering reference value.