Soft sensor technology has been widely applied in key areas of industrial process monitoring. To address challenges such as strong nonlinearity, complex temporal dependencies, and dynamic system behavior commonly encountered in industrial soft sensor data modeling, we propose a hybrid dynamic modeling method that integrates gated recurrent unit (GRU) with temporal convolutional network-Transformer (TCN-Transformer) architecture. TCN-Transformer module is employed to extract multi-scale temporal patterns and capture long-range dependencies among auxiliary variables, while GRU network processes the historical information of target variables through its gated memory mechanism. The complementary feature representations from both components are summed before being passed into a fully connected layer for prediction. To validate the effectiveness of GRU-TCN-Transformer framework, comprehensive case studies were conducted on two typical industrial processes: the prediction of butane (C4) concentration in a debutanizer column and the estimation of hydrogen sulfide (H2S) and sulfur dioxide (SO2) concentrations in a sulfur recovery unit (SRU). Experimental results demonstrate that the proposed hybrid dynamic modeling method significantly outperforms traditional dynamic modeling methods—convolutional neural network (CNN), long short-term memory (LSTM), and TCN—across multiple evaluation metrics. Specifically, for C4 concentration estimation, the proposed method reduced root mean squared error (RMSE), mean absolute error (MAE), and mean absolute percentage error (MAPE) by 55.0%, 51.0% and 50.1%, respectively, and improved R² by 2.3% compared to the best-performing TCN-Transformer model. For H2S estimation, it achieved reductions of 30%, 30.61% and 29.23% in RMSE, MAE, and MAPE, respectively, while increasing R² by 11.09% over the best LSTM-TCN-Transformer model. For SO2 estimation, the proposed model reduced RMSE, MAE, and MAPE by 7.91%, 9.09% and 9.64%, respectively, with a 0.87% increase in R². These comparative results further confirm the improvements in prediction accuracy, indicating that the proposed model is capable of meeting the stringent requirements of industrial applications.

Catalyst wettability plays a critical role in heterogeneous catalysis by governing mass transfer at active sites, thereby affecting reactivity and selectivity. In this study, we propose a strategy to modulate catalyst wettability by doping Cr into CeO2, applied to lignocellulose hydrodeoxygenation. Cr incorporation tailor oxygen vacancy concentration, hydroxyl density, surface roughness, and charge distribution, while also acting as an electronic promoter to enhance metal activity and the degree of hydrodeoxygenation. Consequently, the Cr-modified Pt/CeO2 catalyst achieves nearly complete conversion of lignocellulose into bio-oil and fuel gas with no biomass solid residues, significantly outperforming pristine Pt/CeO2. Density functional theory (DFT) simulations further reveal the formation of a Ce-Cr solid solution, which generates additional defect sites and lowers the energy barrier for ethanol adsorption. This study highlights a new role of Cr as a promoter for designing wettability-matched catalysts in heterogeneous reactions.

While confinement-induced structural disjoining pressure has been extensively studied, the role of surface charge and ionic strength in regulating nanoparticle self-organization and fluid flow in wedged confinements remains incomplete understood. Here, we employ dissipative particle dynamics to investigate the mesoscale behavior of charged nanoparticles suspended in an aqueous phase confined between wedge-shaped walls. By varying wall and particle surface charges and salt concentration, we examine how electrostatic interactions influence nanoparticle layering and fluid flow behavior. Our results demonstrate that electrostatic repulsion between like-charged walls and nanoparticles governs lateral particle migration and ordering, with stronger repulsion driving nanoparticles toward the wedge tip while suppressing their accumulation near charged boundaries. Increasing salt concentration or decreasing surface charge screens electrostatic interactions and hinders step-like nanoparticle layering. In addition to disjoining pressure gradients from spatially varying particle structure, pressure gradients also arise in particle-free regions, contributing to observed circulatory flow within the wedged confinement. SYNOPSIS Electrostatic interactions and ionic screening govern nanoparticle layering in wedge-shaped confinements, thereby modulating structural-pressure-driven water spreading

Direct one-pot upgrading of real lignocellulosic biomass into fuels or value-added chemicals remains a formidable challenge in biorefinery due to the intrinsic recalcitrance of biomass and the strong dependence of conventional hydrodeoxygenation processes on external hydrogen. Herein, we propose an external hydrogen-free dual-valorization catalytic strategy that simultaneously upgrades lignocellulosic biomass and solvent through in situ hydrogen transfer. The hydrothermally synthesized NiFe/TiO2-H catalyst exhibited superior catalytic performance, achieving near-complete biomass conversion with a liquid yield of 56.8 wt%, producing polyols and alkylphenols as the dominant liquid products, while ethanol simultaneously underwent self-condensation to generate higher alcohols. Structural characterization showed that the hydrothermal synthesis route markedly improved highly dispersed Ni-Fe alloy like species and oxygen-deficiency-related oxide signals, while subsequent Fe incorporation further modified the structural and electronic environment of Ni-containing species. Correlation analysis between catalyst structure and catalytic performance suggests that the synergistic interaction between Ni-Fe interfaces and oxygen-vacancy-rich TiO2 facilitates ethanol activation and selective hydrodeoxygenation of oxygenated intermediates. In addition, carbon-balance analysis and fuel-property evaluation indicate that the obtained liquid products possess significantly improved energy density compared with raw lignocellulosic feedstock. This work provides a sustainable and economically attractive strategy for the transfer-hydrogenative upgrading of real lignocellulosic biomass using non-noble metal catalysts.

Pitch-derived hard carbons (HC) are promising anodes for sodium-ion batteries (SIBs) due to their high carbonization yield and low cost. However, the inherent compositional heterogeneity of pitch induces non-uniform oxidative cross-linking during conventional pre-oxidation, which not only renders the microstructure of HC difficult to regulate but also significantly degrades its sodium storage performance. Here, we identify the "shielding effect" of oxidation-inert components in pitch as the root cause of this structural inhomogeneity. To overcome this limitation, we propose a novel "sieving-and-reinforcement strategy". This involves liquid-phase crosslinking to construct a polar three-dimensional (3D) carbon skeleton, followed by stepwise extraction as a molecular sieving process to remove inert components and expose the reactive skeleton, and finally, oxygen etching as a reinforcement step to drastically enhance the crosslinking density and defect population. This controllably engineered carbon skeleton in-situ evolves into an HC with a uniform hierarchical porous structure, featuring abundant ultramicropores, optimally sized closed pores (similar to 2.15 nm), and ultrathin pore walls during carbonization. The resulting HC anode delivers a high reversible capacity of 363.3 mAh g-1 at 50 mA g-1, with an impressive plateau capacity contribution of 71.5%. It also demonstrates exceptional cycling stability, retaining 203.1 mAh g-1 after 500 cycles at a high current density of 1000 mA g-1. This work provides a fundamental understanding of precursor engineering, paving the way for the rational design of advanced carbon materials for next-generation energy storage. (c) 2026 Science Press and Dalian Institute of Chemical Physics, Chinese Academy of Sciences. Published by Elsevier B.V. and Science Press. All rights are reserved, including those for text and data mining, AI training, and similar technologies.

In pharmaceutical manufacturing, there is a widespread presence of industrial waste liquids containing methyl ethyl ketone, isopropanol, and n-heptane. This work employs extractive distillation to separate this multiazeotropic system. Firstly, the conventional extractive distillation process (CED) is designed, and the optimal parameters of the three-objective trade-off are obtained by multi-objective optimization. Then, the side-stream distillation, intermediate reboiler and heat integration enhancement methods are introduced to reduce energy consumption and costs. The results demonstrate that the double side-stream extractive distillation with intermediate reboiler and heat integration process (DSSEDIR-HI) shows the best economic and environmental performance. Compared to the CED process, the total annual cost is reduced by 23.9 %, CO2 emissions are decreased by 42.5 %, and thermodynamic efficiency is improved by 68.4 %. This study provides a sustainable pathway for the separation of methyl ethyl ketone/isopropanol/n-heptane mixture.

Low-quality images, due to degradation issues such as noise and compression artifacts, often lead to feature extraction distortion in classification models, thereby reducing classification accuracy. This issue is particularly prominent in practical computer vision applications. To address this, this paper proposes the SR-PAN-EffNet model, which integrates a semantic-guided SRGAN restoration module with a noise-aware PANet attention module, and employs end-to-end joint training to achieve the collaborative optimization of “restoration serving classification”. Experimental results show that the model achieves Top-1 accuracy of 68.5% and 57.6% on the ImageNet-1K and CIFAR-100 low-quality datasets, respectively, improving by 4.7–6.4 percentage points over NFNet-F4, with PSNR and SSIM also leading. Ablation experiments reveal that removing core modules results in a 2.9–6.8 percentage point decrease in accuracy. Future work will focus on improving the model’s real-time performance and robustness through lightweight design, extreme degradation adaptive optimization, and category-adaptive guidance, aiming to promote its application in scenarios such as surveillance and medical imaging.

In the synthesis of liquid crystal monomers and the industrial production of norgestrel in the pharmaceutical industry, wastewater containing tetrahydrofuran and ethanol is generated, necessitating the recovery and reuse of these high-value organic compounds. This study identified dimethyl sulfoxide (DMSO) as the entrainer with optimal separation performance and sustainability through phase diagram analysis and ecotoxicity assessment. The mechanism by which DMSO enhances the separation of the ternary azeotrope was investigated via quantum chemical calculations. A three-column extractive distillation (TCED) process and an extractive dividing-wall column (EDWC) process were designed. Using the total annual cost, carbon dioxide emissions, and process route index as objective functions, both processes were optimized separately using the second-generation nondominated sorting genetic algorithm. The results indicate that the EDWC process exhibits a superior performance. The sustainability of the EDWC process was further demonstrated using the Eco-indicator 99 method based on a life cycle assessment. Exergy analysis results show that the exergy loss of the EDWC process is reduced by 25.77% compared with the TCED process. Finally, three control structures were designed for the EDWC process, and subsequent dynamic control performance tests revealed that the model predictive control scheme delivered the best control performance. The main contribution of this work lies in providing a multiscale analysis ranging from intermolecular interactions to process optimization and control, offering valuable insights for the practical industrial application of the EDWC process.

A novel bifunctional zeolite-encaged Ni–Pt catalyst enables efficient low-temperature methane activation and aromatization via Pt-induced electronic modification of Ni, achieving high methane conversion and aromatics selectivity.

To study the influence of mass and heat transfer on the microcrystalline structure and properties of mesophase pitch and resulting carbon fiber properties, mesophase pitches were synthesized via pressurized/N₂-blowing thermal condensation with different stirring rates, with experimental conditions optimized using response surface methodology (RSM). RSM analysis confirmed that mesophase content was highly dependent on stirring rate (p < 0.05), and the influencing factors on the formation of mesophase pitch is ranked as reaction temperature > duration time > stirring rate > reaction pressure. The results demonstrated that a moderate increase in stirring rate enhanced molecular diffusion and heat transfer, improving reaction kinetics and aromatic molecule interactions. This accelerated mesophase sphere growth and coalescence while inducing molecular orientation via shear, ultimately yielding a wide-domain optical texture with 100 vol

Acetonitrile (ACN) is an important organic compound and versatile chemical intermediate. Recovering ACN from pharmaceutical waste liquids can reduce production costs and mitigate environmental hazards. At atmospheric pressure, ACN forms an azeotrope with water, and extractive distillation is one of the most effective separation methods. In this study, the conductor-like screening model-segment activity coefficient (COSMO-SAC) was employed to identify deep eutectic solvents (DESs) prepared using choline chloride (ChCl) as the hydrogen bond acceptor and oxalic acid (OA) and glycolic acid (GA) as hydrogen bond donors for separating the ACN-water azeotrope. Vapor-liquid equilibrium (VLE) experiments demonstrate that adding 25 mol % DESs significantly enhances the relative volatility of ACN to water. The nonrandom two-liquid (NRTL) model was used to correlate the experimental data with satisfactory accuracy. Quantum chemical calculations were performed to analyze the types, sites, and strengths of weak interactions between DESs and azeotropic components. Results confirm that interactions between DESs and water are stronger than those with ACN, thereby reducing the activity coefficient of water in the liquid phase and enabling azeotrope separation.

Polymeric hydrogels, known as blocking gel or disproportionate-permeability-reducer, have been highly successful in improving sweep efficiency and reducing excessive water cut by adjusting reservoir heterogeneity (Al-Muntasheri and Zitha. Gel under dynamic stress in porous media: new insights using computed tomography. SPE Saudi Arabia Section Technical Symposium; 2009. Al-Sharji, Grattoni, Dawe, et al. Pore-scale study of the flow of oil and water through polymer gels. SPE Annual Technical Conference and Exhibition; 1999. Bai. Preformed particle gel for conformance control: factors affecting its properties and applications. SPE Reservoir Eval Eng. 2007;10:415-422). However, it remains an extremely challenging task to develop polymeric weak gel for in-depth conformance control, simply because most in situ synthetic weak hydrogels suffer from loosely structured network and lack of efficient energy dissipation mechanism in harsh temperature environment (Bai, Leng, and Wei. A comprehensive review of in-situ polymer gel simulation for conformance control. Pet Sci. 2022;19(1):189-202. Bhattacharya and Samanta. Soft-nanocomposites of nanoparticles and nanocarbons with supramolecular and polymer gels and their applications. ACS Appl Mater Interfaces. 2016;8(19):21512. Bai, Zhou, and Yin. Comprehensive review of polyacrylamide polymer gels for conformance control. Pet Explor Dev. 2015;42(4):525-532). This paper introduces a low-cost, high-temperature resistant, polymer covalent weak gel system. Polyethylenimine (PEI) and polyethylenimine-modified nano-SiO2 nano-crosslinker were used as crosslinkers, and the performance of the gel system under high temperatures was systematically evaluated. The studied gel properties include gel formation time, gel strength, thermal stability, infrared spectroscopy, rheological properties, long-term stability, and microstructure. Acrylamide/2-acrylamido-2-methyl propane sulfonate (named QC-9 in this paper) is used in the gel system at concentrations as low as 2250 mg/L, and the gel time can be controlled within 3-12 h by adjusting the ratio of the polymers and crosslinkers. The gel exhibits a viscosity of 100-200 mPas, withstands temperatures up to 130 degrees C, and has a salt tolerance of 214818.52 mg/L, with a viscosity retention rate of >= 80% after 70 days. At a shear rate of 1 Hz, the elastic modulus (G ') after aging for 70 days is 1.07 Pa, and the viscous modulus (G '') is 0.25 Pa. The test results indicate that the gel system remains in the high-viscosity range. Thermogravimetric Analysis tests indicate that the structural destruction temperature of the composite gel is 140 degrees C, the activation energy of the nanocomposite crosslinking polymer weak gel is 5.2 and 16.5 kJ/mol higher for the evaporation of free water and the escape of bound water, respectively, compared to that of the PEI-crosslinking microgel. Finally, scanning electron microscopy and atomic force microscopy were used to observe the gel's microstructure. Compared to traditional PEI crosslinked gel systems, the organic/inorganic nanocomposite crosslinked gel system has a dense, thickened network structure.

n-Hexane is widely used in industry, and methylcyclopentane is often generated as a byproduct during its production. The small boiling point difference (3.08 K) between these two compounds makes their separation challenging using traditional distillation methods. Extractive distillation, known for its high efficiency and energy savings, is effective in separating azeotropic and near-azeotropic mixtures. Deep Eutectic Solvents (DESs), a new class of green solvents, have shown promise in chemical separations. This study investigates the use of DESs as potential entrainers for the separation of n-hexane and methylcyclopentane in extractive distillation. The COSMO-SAC model was used to select two DESs: DES1 (tetrabutylammonium bromide (TBAB):decanoic acid = 1:2) and DES2 (TBAB:oleic acid = 1:3). Vapor-liquid equilibrium (VLE) data for the ternary system (n-hexane-methylcyclopentane-DES) were measured at 101.3 kPa. The results showed that both DESs effectively separated n-hexane and methylcyclopentane at a 25 mol % concentration, with DES1 demonstrating superior performance. The VLE data were fitted using the Non-Random Two-Liquid (NRTL) model, yielding satisfactory results. Quantum chemistry calculations further elucidated the molecular mechanisms behind the superior separation performance of DES1.

Data-driven soft sensor modeling has gained significant traction in modern chemical process monitoring and quality prediction. However, persistent challenges remain in accurately characterizing the complex dynamics inherent in chemical production systems, which typically exhibit significant time delays, strong nonlinearity, and time-varying characteristics. To address these critical challenges, a dynamic soft sensor modeling method based on temporal convolutional network (TCN) combined with channel spatiotemporal attention module and long short-term memory network (TCN-CBAM-LSTM) is proposed. Firstly, TCN is employed to extract deep nonlinear dynamic dependencies from process variables through its dilated causal convolution architecture, secondly, a convolutional block attention module (CBAM) is incorporated to enhance feature representation by adaptively focusing on critical spatiotemporal information across different sensor channels, finally, a long shortterm memory network (LSTM) is integrated to model intricate temporal patterns and long-range dependencies between process variables and quality indicators. This multi-stage architecture enables comprehensive learning of both local temporal features and global dynamic relationships within complex chemical processes. To verify the effectiveness of the proposed method, TCN-CBAM-LSTM was applied to a soft sensor modeling example for calculating the exhaust gas composition in a sulfur recovery unit (SRU). Under the same experimental conditions, it was also compared with convolutional neural network (CNN), variable weighted stacked autoencoder (VWSAE), spatiotemporal attention LSTM (STA-LSTM), CNN-LSTM, TCN, and TCN-LSTM. The results show that the TCN-CBAM-LSTM method has better performance and modeling accuracy, and its performance meets the needs of practical engineering applications.

This study established a rate-based model for the ternary amine (MEA/MDEA/AMP) CO2 capture process and employed a multiobjective genetic algorithm to achieve coordinated optimization of solvent formulation and process parameters. The basic process for CO2 capture using a ternary amine blend is constructed, and the absorber intercooler (AIC) configuration and split-flow (SF) configuration are designed. The results indicate that the AIC process is suitable for higher-concentration absorbents, achieving a 12.25% reduction in liquid-to-gas ratio compared to the basic process and 5.53% reduction in total annual cost (TAC). The SF process significantly decreases the water content in the gas phase of the stripper column, leading to a 22.75% reduction in TAC. When combined, the effects demonstrate cumulative superposition, resulting in a 27.11% reduction in TAC and a 37.39% reduction in carbon emissions. The integration of a multiobjective genetic algorithm for parameter optimization provides an efficient solution for improving composite amine-based carbon capture technology.

This study investigates a novel weak gel system prepared from hyperbranched nanowires for enhanced oil recovery (EOR) applications. The gel was synthesized using an environmentally friendly slow-release crosslinker and evaluated for its gelation behavior, viscosity, and structural stability under simulated reservoir conditions. Characterization techniques, including Fourier Transform Infrared Spectroscopy (FT-IR), Differential Scanning Calorimetry (DSC), and Scanning Electron Microscopy (SEM), were employed to analyze the chemical and physical properties of the gels. The results demonstrated that O-20@PAM@CNTs gels exhibited superior gelation rates, viscosity, and structural integrity compared to O-20@PAM gels, particularly under elevated temperatures. Core flooding experiments indicated that both gel systems effectively reduced permeability and improved oil recovery, with O-20@PAM@CNTs achieving a higher cumulative recovery rate of 89.9 % compared to 79.2 % for O-20@PAM. The findings highlight the potential of CNTs-enhanced gels in optimizing oil recovery and addressing reservoir heterogeneity. The findings of this research indicate the potential of O-20@PAM@CNTs to serve as promising candidates for profile control. The introduction of carbon nanotubes (CNTs) into the polymer matrix not only accelerated the gelation process but also significantly improved the mechanical properties and thermal stability of the gels. Core plugging and flooding tests revealed that O-20@PAM@CNTs gels provided superior performance in reducing permeability and improving the cumulative recovery rate, demonstrating their capability to effectively block high-permeability channels and redirect flow towards oil-rich areas. The degradation behavior of the gels also indicated their suitability for reservoir cleaning post-operation. Overall, this study underscores the potential of utilizing advanced polymer-based gels for efficient EOR, offering valuable insights for future applications in the field of petroleum engineering.

For the soft sensor modeling of chemical processes exhibiting pronounced nonlinearity and intricacy, this study proposes a soft sensor model – TCN-LSTM, which combines Temporal Convolutional Networks (TCN) and Long Short-term Memory Networks (LSTM). TCN-LSTM can extract the spatio-temporal features and dynamic response relationships of the input samples, addressing their time-varying delay issue by discerning dynamic response relationships. To confirm TCN-LSTM's efficacy, we apply it to modeling a soft sensor instance of a debutanizer column. The experiment outcomes demonstrate that TCN-LSTM exhibits superior measurement accuracy over Backpropagation Neural Network (BP), Radial Basis Function Neural Network (RBF), Convolutional Neural Networks (CNN), LSTM, and TCN approaches.

The process of separating ethylene glycol and 1,2-butanediol with reactive distillation is intricate, with many variables involved, leading to a time-consuming and inefficient optimization process. To expedite the optimization process, this study proposes a new parallel stochastic algorithm framework. This research shows that the framework is 36.8 times faster than the serial framework. The reactive distillation-assisted separation process of ethylene glycol and 1,2-butanediol is optimized using the parallel stochastic algorithm framework. The optimization process duration is 810 min. Based on the optimal design, the process is further intensified by incorporating the dividing wall column and intermediate reboiler. Compared to the base process, the intensified process exhibited reductions of 8.89 %, 10.18 %, and 9.83 % in total capital cost, total operating cost, and total annual cost, respectively.