Applied Sciences

17 pages, 589 KB

Open AccessReview

Biomarkers of Motor Recovery After Corticospinal Tract Damage in Stroke: A Scoping Review

by Błażej Cieślik, Pierre Bouquillon, Roberto De Almeida Lomba, Thibault Chatton, Davide Grillo, Mirko Zitti, Silvia Zangarini, Tomasz Rutkowski, Roberto Meroni and Pawel Kiper

Appl. Sci. 2026, 16(1), 317; https://doi.org/10.3390/app16010317 (registering DOI) - 28 Dec 2025

Abstract

Motor recovery after stroke is highly variable and closely linked to the extent of corticospinal tract (CST) damage. Neurophysiological biomarkers, such as motor evoked potentials (MEPs) and structural imaging markers, including CST lesion load and fractional anisotropy (FA), show promise for predicting motor [...] Read more.

Motor recovery after stroke is highly variable and closely linked to the extent of corticospinal tract (CST) damage. Neurophysiological biomarkers, such as motor evoked potentials (MEPs) and structural imaging markers, including CST lesion load and fractional anisotropy (FA), show promise for predicting motor outcomes. This scoping review evaluated the prognostic value of these biomarkers and the utility of multimodal models for individualized rehabilitation. A systematic search (April 2024) in PubMed, the Cochrane Library, and Scopus identified empirical studies examining biomarkers predictive of post-stroke motor recovery. Biomarkers were primarily derived from magnetic resonance imaging (resting-state functional connectivity and diffusion-weighted imaging) and transcranial magnetic stimulation. Nineteen studies (1219 patients) met the inclusion criteria. Structural biomarkers, particularly lower CST FA and higher weighted lesion load, were generally associated with poorer motor recovery. Combining neurophysiological measures, such as MEP status, with functional imaging and artificial intelligence-based analyses may improve prognostic precision. Multimodal approaches appeared promising in some studies, but evidence remains limited and heterogeneous. Integrating diverse biomarkers into multimodal prognostic models may enhance the prediction of motor recovery and support personalized rehabilitation after stroke, although heterogeneity in study design and outcome assessment highlights the need for standardized, large-scale longitudinal studies to enable clinical implementation. Full article

(This article belongs to the Section Applied Neuroscience and Neural Engineering)

► Show Figures

Figure 1

26 pages, 1203 KB

Open AccessSystematic Review

Radiation Dose Reduction in CT Exams with Iterative and Deep Learning Reconstruction: A Systematic Review

by Sandra Coelho, Maria de Lurdes Dinis, Marco Freitas and João Santos Baptista

Appl. Sci. 2026, 16(1), 316; https://doi.org/10.3390/app16010316 (registering DOI) - 28 Dec 2025

Abstract

This systematic review evaluated the effectiveness of iterative reconstruction (IR) and deep learning reconstruction (DLR) in reducing radiation dose in computed tomography (CT) while preserving diagnostic image quality. We systematically searched PubMed, Scopus, and Web of Science (last search 22 March 2025); the [...] Read more.

This systematic review evaluated the effectiveness of iterative reconstruction (IR) and deep learning reconstruction (DLR) in reducing radiation dose in computed tomography (CT) while preserving diagnostic image quality. We systematically searched PubMed, Scopus, and Web of Science (last search 22 March 2025); the protocol was registered in the OSF (DOI: 10.17605/OSF.IO/TUQDS). Eligible studies were English-language adult (≥18 years) investigations published between 2020 and 2025 that used IR or DLR and reported radiation-dose outcomes; studies on paediatric, phantom, cadaver, cone-beam, and spectral CT were excluded. In accordance with PRISMA 2020 guidelines, 4371 records were identified, and 30 met the inclusion criteria. Risk of bias was assessed using the NIH Quality Assessment Tool; most studies were deemed to be at low risk. Data were narratively synthesised and structured by a reconstruction approach and anatomical region. Across the 30 studies, IR achieved a dose reduction of 24–50% (mean ≈ 45%) and a DLR reduction of 34–89% (mean ≈ 58%); several DLR protocols enabled reductions of ≥75% without impairing diagnostic quality. Thirty studies in total were included (total N = 2581; range 24–289). It was determined that both approaches substantially reduce radiation exposure while maintaining diagnostic image quality; DLR generally demonstrates greater noise suppression and dose efficiency, especially in ultra-low-dose applications. However, heterogeneity in methods, designs, and scanner technologies limits the ability to draw uniform conclusions. Standardised protocols, multi-vendor prospective studies, and long-term evaluations are needed. Full article

► Show Figures

Figure 1

23 pages, 95864 KB

Open AccessArticle

ALUDARM: A Lightweight Universal Database-Assisted Registration Method for On-Board Remote Sensing Imagery

by Linhui Wang, Rui Liu, Guangyao Zhou, Hongjian You and Niangang Jiao

Appl. Sci. 2026, 16(1), 315; https://doi.org/10.3390/app16010315 (registering DOI) - 28 Dec 2025

Abstract

Satellite on-board registration is becoming increasingly prevalent since it shortens the data processing chain, enabling users to acquire actionable information more efficiently. However, current on-board processing hardware exhibits severely constrained storage and computational resources, making traditional ground-based methods infeasible in terms of storage [...] Read more.

Satellite on-board registration is becoming increasingly prevalent since it shortens the data processing chain, enabling users to acquire actionable information more efficiently. However, current on-board processing hardware exhibits severely constrained storage and computational resources, making traditional ground-based methods infeasible in terms of storage and time efficiency. Meanwhile, real-time orbit parameters are normally less accurate, causing a large initial geolocation offset. In this paper, we propose a novel registration framework based on a well-designed lightweight universal database to address the challenges of limited storage as well as poor initial accuracy. Firstly, for the global matching step, a lightweight universal database is designed by storing a feature vector of control points instead of a traditional basemap (such as Digital Orthophoto Map and Digital Surface Model) for on-board processing. We replace the keypoint detection stage with a sparse sampling strategy, which significantly improves time efficiency. In addition, the sparsely sampled control points avoid the problem of keypoint repeatability, allowing the proposed method to perform robust global matching with few control points and little storage usage. Then, for the local matching step, we introduce relative total variation to extract the most obvious and significant structures from the basemap, so that unimportant feature or noise can be omitted from the database. Combined with Run-Length Encoding, the masked binary edge feature yields high precision with considerably reduced storage. Quantitative experiments demonstrate that the proposed reference database occupies less than

5 %

of raw image storage, while maintaining efficiency and accuracy comparable to SOTA methods. Full article

(This article belongs to the Collection Space Applications)

► Show Figures

Figure 1

19 pages, 9564 KB

Open AccessArticle

High-Fidelity Colorimetry Using Cross-Polarized Hyperspectral Imaging and Machine Learning Calibration

by Zhihao He, Li Luo, Xiangyang Yu, Yuchen Guo and Weibin Hong

Appl. Sci. 2026, 16(1), 314; https://doi.org/10.3390/app16010314 (registering DOI) - 28 Dec 2025

Abstract

Accurate colorimetric quantification presents a significant challenge, as traditional imaging technologies fail to resolve metamerism and even hyperspectral imaging (HSI) is compromised by nonlinearities and specular reflections. This study introduces a high-fidelity colorimetric system using cross-polarized HSI to suppress specular reflections, integrated with [...] Read more.

Accurate colorimetric quantification presents a significant challenge, as traditional imaging technologies fail to resolve metamerism and even hyperspectral imaging (HSI) is compromised by nonlinearities and specular reflections. This study introduces a high-fidelity colorimetric system using cross-polarized HSI to suppress specular reflections, integrated with a Support Vector Regression (SVR) model to correct the system’s nonlinear response. The system’s performance was rigorously validated, demonstrating exceptional stability and repeatability (average

Δ E_{00} < 0.1

). The SVR calibration significantly enhanced accuracy, reducing the mean color error from

Δ E_{00} = 4.36

to

0.43

. Furthermore, when coupled with a Random Forest classifier, the system achieved 99.0% accuracy in discriminating visually indistinguishable (metameric) samples. In application-specific validation, it successfully quantified cosmetic color shifts and achieved high-precision skin-tone matching with a fidelity as low as

Δ E_{00} = 0.82

. This study demonstrates that the proposed system, by synergistically combining cross-polarization and machine learning, constitutes a robust tool for high-precision colorimetry, addressing long-standing challenges and showing significant potential in fields like cosmetic science. Full article

► Show Figures

Figure 1

19 pages, 9838 KB

Open AccessArticle

Processing of Large Underground Excavation System—Skeleton Based Section Segmentation for Point Cloud Regularization

by Przemysław Dąbek, Jacek Wodecki, Adam Wróblewski and Sebastian Gola

Appl. Sci. 2026, 16(1), 313; https://doi.org/10.3390/app16010313 (registering DOI) - 28 Dec 2025

Abstract

Numerical modelling of airflow in underground mines is gaining importance in modern ventilation system design and safety assessment. Computational Fluid Dynamics (CFD) simulations enable detailed analyses of air movement, contaminant dispersion, and heat transfer, yet their reliability depends strongly on the accuracy of [...] Read more.

Numerical modelling of airflow in underground mines is gaining importance in modern ventilation system design and safety assessment. Computational Fluid Dynamics (CFD) simulations enable detailed analyses of air movement, contaminant dispersion, and heat transfer, yet their reliability depends strongly on the accuracy of the geometric representation of excavations. Raw point cloud data obtained from laser scanning of underground workings are typically irregular, noisy, and contain discontinuities that must be processed before being used for CFD meshing. This study presents a methodology for automatic segmentation and regularization of large-scale point cloud data of underground excavation systems. The proposed approach is based on skeleton extraction and trajectory analysis, which enable the separation of excavation networks into individual tunnel segments and crossings. The workflow includes outlier removal, alpha-shape generation, voxelization, medial-axis skeletonization, and topology-based segmentation using neighbor relationships within the voxel grid. A proximity-based correction step is introduced to handle doubled crossings produced by the skeletonization process. The segmented sections are subsequently regularized through radial analysis and surface reconstruction to produce uniform and watertight models suitable for mesh generation in CFD software (Ansys 2024 R1). The methodology was tested on both synthetic datasets and real-world laser scans acquired in underground mine conditions. The results demonstrate that the proposed segmentation approach effectively isolates single-line drifts and crossings, ensuring continuous and smooth geometry while preserving the overall excavation topology. The developed method provides a robust preprocessing framework that bridges the gap between point cloud acquisition and numerical modelling, enabling automated transformation of raw data into CFD-ready geometric models for ventilation and safety analysis of complex underground excavation systems. Full article

(This article belongs to the Special Issue Mining Engineering: Present and Future Prospectives)

► Show Figures

Figure 1

22 pages, 4845 KB

Open AccessArticle

Predicting Nuclear Level Density Using a Physics-Informed Neural Network with Multi-Task Learning

by Bora Canbula

Appl. Sci. 2026, 16(1), 312; https://doi.org/10.3390/app16010312 (registering DOI) - 28 Dec 2025

Abstract

The accurate determination of nuclear level density (NLD) is essential for a wide range of applications in nuclear science, including reactor design, nuclear astrophysics, and nuclear data evaluation. Traditional phenomenological models often face challenges in capturing key physical effects, such as collective excitations [...] Read more.

The accurate determination of nuclear level density (NLD) is essential for a wide range of applications in nuclear science, including reactor design, nuclear astrophysics, and nuclear data evaluation. Traditional phenomenological models often face challenges in capturing key physical effects, such as collective excitations and shell structure, particularly in heavy and transitional nuclei, where the level density grows exponentially. Machine learning (ML) approaches have shown promise in improving predictive accuracy but are often limited by their purely data-driven nature, leading to challenges in interpretability and performance in regions with sparse experimental data. In this study, we propose a Physics-Informed Neural Network (PINN) framework, enhanced through multi-task learning (MTL), to address these limitations. The proposed model simultaneously predicts cumulative levels and mean resonance spacings by integrating experimental data with theoretical constraints, ensuring consistency with nuclear structure theory and robustness in extrapolating beyond the training data. Validation against both cumulative and yrast levels highlights the model’s ability to accurately capture rotational and vibrational excitations across a wide range of isotopes. Comparative evaluations demonstrate that the PINN model significantly outperforms traditional phenomenological models and purely data-driven approaches, offering a comprehensive and interpretable framework for advancing nuclear level density predictions and supporting practical applications in nuclear energy and astrophysics. Full article

(This article belongs to the Section Computing and Artificial Intelligence)

► Show Figures

Figure 1

29 pages, 9773 KB

Open AccessArticle

Prediction of Mean Fragmentation Size in Open-Pit Mine Blasting Operations Using Histogram-Based Gradient Boosting and Grey Wolf Optimization Approach

by Madalitso Mame, Shuai Huang, Chuanqi Li, Xiaoguang Zhou and Jian Zhou

Appl. Sci. 2026, 16(1), 311; https://doi.org/10.3390/app16010311 (registering DOI) - 28 Dec 2025

Abstract

Blast-induced rock fragmentation plays a critical role in mining and civil engineering. One of the primary objectives of blasting operations is to achieve the desired rock fragmentation size, which is a key indicator of the quality of the blasting process. Predicting the mean [...] Read more.

Blast-induced rock fragmentation plays a critical role in mining and civil engineering. One of the primary objectives of blasting operations is to achieve the desired rock fragmentation size, which is a key indicator of the quality of the blasting process. Predicting the mean fragmentation size (MFS) is crucial to avoid increased production costs, material loss, and ore dilution. This study integrates three tree-based regression techniques—gradient boosting regression (GBR), histogram-based gradient boosting machine (HGB), and extra trees (ET)—with two optimization algorithms, namely, grey wolf optimization (GWO) and particle swarm optimization (PSO), to predict the MFS. The performance of the resulting models was evaluated using four statistical measures: coefficient of determination (R²), root mean squared error (RMSE), mean absolute error (MAE), and mean absolute percentage error (MAPE). The results indicate that the GWO-HGB model outperformed all other models, achieving R², RMSE, MAE, and MAPE values of 0.9402, 0.0251, 0.0185, and 0.0560, respectively, in the testing phase. Additionally, the Shapley additive explanations (SHAP), local interpretable model-agnostic explanations (LIME), and neural network-based sensitivity analyses were applied to examine how input parameters influence model predictions. The analysis revealed that unconfined compressive strength (UCS) emerged as the most influential parameter affecting MFS prediction in the developed model. This study provides a novel hybrid intelligent model to predict MFS for optimized blasting operations in open-pit mines. Full article

(This article belongs to the Special Issue Advances and Technologies in Rock Mechanics and Rock Engineering)

► Show Figures

Figure 1

16 pages, 6233 KB

Open AccessArticle

A Tire Temperature Adaptive Extended Kalman Filter for Sideslip Angle Estimation: Experimental Validation on a Race Track

by Andrea Masoero, Raffaele Manca, Luis M. Castellanos Molina and Andrea Tonoli

Appl. Sci. 2026, 16(1), 310; https://doi.org/10.3390/app16010310 (registering DOI) - 28 Dec 2025

Abstract

Real-time estimation of vehicle sideslip angle is essential for both safety and performance applications. This study presents a temperature-adaptive Extended Kalman Filter (EKF) that estimates the sideslip angle of a racing vehicle by integrating dynamic and kinematic information. A temperature-dependent Pacejka tire model, [...] Read more.

Real-time estimation of vehicle sideslip angle is essential for both safety and performance applications. This study presents a temperature-adaptive Extended Kalman Filter (EKF) that estimates the sideslip angle of a racing vehicle by integrating dynamic and kinematic information. A temperature-dependent Pacejka tire model, derived directly from track tests, is embedded in a 3-degree-of-freedom dual-track vehicle model and used within the EKF to compensate for temperature-induced variations in tire behavior. The adaptive model parameters are identified from standard on-track maneuvers conducted at different tire temperatures, without the need for additional indoor rig testing. Experimental validation on a race track demonstrates that incorporating tire temperature adaptation and combining dynamic and kinematic estimation significantly enhance estimation accuracy, particularly underow-grip and high-performance driving conditions attested by a reduction of 40–50% in RMS error and a further reduction in maximum absolute error. Full article

(This article belongs to the Special Issue Advancements and Future Trends in Vehicle Electrification and Intelligent Control Systems)

► Show Figures

Figure 1

18 pages, 4575 KB

Open AccessArticle

Analysis of Echo Characteristics of Pulsed Laser Short-Range Detection Based on Light Cone Beam Expansion Mechanism

by Changkun Ke, Lin Gan, He Zhang and Miaomiao Chen

Appl. Sci. 2026, 16(1), 309; https://doi.org/10.3390/app16010309 (registering DOI) - 28 Dec 2025

Abstract

This study aims to fill the existing gap in laser detection research, particularly regarding how the waveform of outgoing laser pulses affects detection performance. Based on the mechanism of light cone beam expansion, this study emits three different laser pulse signals to detect [...] Read more.

This study aims to fill the existing gap in laser detection research, particularly regarding how the waveform of outgoing laser pulses affects detection performance. Based on the mechanism of light cone beam expansion, this study emits three different laser pulse signals to detect short-range targets. A theoretical model for short-range ranging of these lasers is established, and the effects of emission power, divergence angle, and equivalent root mean square noise voltage on circumferential detection accuracy are simulated and experimentally measured. As emission power decreases, both echo amplitude and detection accuracy decline for all three pulsed lasers. Additionally, except for the inverted parabolic function, both echo amplitude and detection accuracy decrease with reduced divergence angle. An increase in equivalent root mean square noise voltage broadens the half-width of the probability density distribution for pulsed laser detection. The mean central position deviation between the ideal and measured detection probability density distributions of the heavy-tailed function laser pulses shows the best performance and the highest fidelity, which are +0.01 m, +0.05 m, and +0.02 m, respectively, which is of great significance for the development of laser detection technology. Full article

(This article belongs to the Section Optics and Lasers)

► Show Figures

Figure 1

21 pages, 5675 KB

Open AccessArticle

Numerical Simulation of the Formation of Frozen Walls in Subway Cross Passages Under Seepage Conditions

by Xin Liu, Huijie Cheng, Juan Deng, Xuefu Zhang, Zhaohui Sun, Linfeng Wang, Fuping Zheng and Yuchao Xia

Appl. Sci. 2026, 16(1), 308; https://doi.org/10.3390/app16010308 (registering DOI) - 28 Dec 2025

Abstract

The artificial ground freezing (AGF) technique is widely used in the construction of subway cross passages due to its advantages of good water sealing, strong adaptability, and minimal environmental impact. However, groundwater seepage adversely affects the formation of the frozen wall. The functional [...] Read more.

The artificial ground freezing (AGF) technique is widely used in the construction of subway cross passages due to its advantages of good water sealing, strong adaptability, and minimal environmental impact. However, groundwater seepage adversely affects the formation of the frozen wall. The functional relationship between the content of unfrozen water and the temperature in saturated sandy gravel was obtained using frequency domain reflectometry (FDR). Based on the theories of heat transfer and seepage in porous media, a coupled hydrothermal mathematical model of saturated ground considering phase change was established. This model was verified using results from a model test and a freezing project for a subway cross passage. Building on this, the influence of seepage velocity on the formation and closure time of the frozen wall was studied, and prediction formulas for closure times under different seepage velocities were proposed. The results demonstrate the effectiveness of the VG–Clapeyron model in predicting the unfrozen water content in saturated sandy gravel. Groundwater seepage is the core factor affecting the formation of the frozen wall. As seepage velocity increases, closure times for both the cross passage and the pump room are significantly delayed, and the difference between their respective closure times increases. The upstream sidewall is the weak link in frozen wall expansion under seepage conditions. Monitoring of the temperature field in this area should be strengthened to track the formation of the frozen wall. Full article

► Show Figures

Figure 1

18 pages, 3043 KB

Open AccessArticle

Antibacterial and Antioxidant Performance of Natural Textile Dyes for Children’s Wear

by Diana Santiago, Behnaz Mehravani, Cátia Alves, Isabel Cabral, Joana Cunha, Andrea Zille and Jorge Padrão

Appl. Sci. 2026, 16(1), 307; https://doi.org/10.3390/app16010307 (registering DOI) - 28 Dec 2025

Abstract

Children’s skin is highly sensitive and prone to irritation, allergies, and infections, requiring special consideration in textile selection. Although clothing serves as a protective barrier, it can also pose a risk when dyed with toxic chemical colourants. This study explores the potential of [...] Read more.

Children’s skin is highly sensitive and prone to irritation, allergies, and infections, requiring special consideration in textile selection. Although clothing serves as a protective barrier, it can also pose a risk when dyed with toxic chemical colourants. This study explores the potential of multifunctional natural dyes as safer alternatives for children’s clothing, particularly for those with dermatological conditions. Cotton knitted fabrics were dyed through exhaustion with extracts of madder root (Rubia tinctorum L.), pomegranate peel (Ppe, Punica granatum L.), oxidised logwood (Logox, Haematoxylum campechianum L.), and tannin from quebracho (Schinopsis lorentzii Griseb.), both individually and in various combinations with or without potassium aluminium sulphate dodecahydrate (alum). The combination of madder and Ppe demonstrated the most promising multifunctional performance, being classified as a weak disinfectant against S. aureus (3.7 log reduction) and showing the highest antioxidant activity (92.6 ± 2.56% 2,2′-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid (ABTS) radical reduction), while maintaining excellent results after washing. Moreover, these natural formulations expanded the achievable colour palette from each dye while maintaining moderate wash fastness. The results highlight the relevance of these findings to textile and fashion designers, offering sustainable tools for creating health-conscious, visually appealing garments. This research reinforces the potential of natural dyes and biomordants in developing functional textiles that support children’s wellbeing and environmental responsibility. Full article

(This article belongs to the Special Issue Sustainability in Functional Textiles: Materials, Technology and Design)

► Show Figures

Figure 1

21 pages, 13601 KB

Open AccessArticle

Determination of Compressive Strength in Hemp–Lime Composites: Comparative Study of Testing Methodologies and Proposal of Improved Approach

by Wojciech Piątkiewicz, Andrzej Piotrowski and Piotr Narloch

Appl. Sci. 2026, 16(1), 306; https://doi.org/10.3390/app16010306 (registering DOI) - 28 Dec 2025

Abstract

The determination of compressive strength of hemp–lime composite (HLC) is currently based on diverse and non-unified approaches, which complicates the direct comparison of results obtained in different studies. This study compares commonly used strength-determination methods using a dedicated experimental dataset and proposes a [...] Read more.

The determination of compressive strength of hemp–lime composite (HLC) is currently based on diverse and non-unified approaches, which complicates the direct comparison of results obtained in different studies. This study compares commonly used strength-determination methods using a dedicated experimental dataset and proposes a new, more universal approach, defining compressive strength as the stress corresponding to 1% permanent strain. Cubic specimens (150 × 150 × 150 mm) were produced at three compaction levels (150%, 170%, 190%) and tested under uniaxial compression. Increasing compaction increased density and resulted in compressive strengths of 219, 316, and 349 kPa, respectively. Methods based on fixed strain levels (5% and 10%) showed good repeatability but primarily reflected serviceability limits, whereas those based on loss of linearity or stiffness reduction were less reliable and highly sensitive to curve interpretation. The proposed 1% permanent-strain method accurately captured the onset of irreversible deformation and aligned with the real behavior of hemp–lime infill in frame structures, supporting its use for standardizing compressive-strength testing of hemp–lime composites. Full article

(This article belongs to the Special Issue High-Performance Concrete: Preparation, Properties, Evaluation and Applications)

► Show Figures

Figure 1

22 pages, 4301 KB

Open AccessArticle

Intelligent Wind Power Forecasting for Sustainable Smart Cities

by Zhihao Xu, Youyong Kong and Aodong Shen

Appl. Sci. 2026, 16(1), 305; https://doi.org/10.3390/app16010305 (registering DOI) - 28 Dec 2025

Abstract

Wind power forecasting is critical to renewable energy generation, as accurate predictions are essential for the efficient and reliable operation of power systems. However, wind power output is inherently unstable and is strongly affected by meteorological factors such as wind speed, wind direction, [...] Read more.

Wind power forecasting is critical to renewable energy generation, as accurate predictions are essential for the efficient and reliable operation of power systems. However, wind power output is inherently unstable and is strongly affected by meteorological factors such as wind speed, wind direction, and atmospheric pressure. Weather conditions and wind power data are recorded by sensors installed in wind turbines, which may be damaged or malfunction during extreme or sudden weather events. Such failures can lead to inaccurate, incomplete, or missing data, thereby degrading data quality and, consequently, forecasting performance. To address these challenges, we propose a method that integrates a pre-trained large-scale language model (LLM) with the spatiotemporal characteristics of wind power networks, aiming to capture both meteorological variability and the complexity of wind farm terrain. Specifically, we design a spatiotemporal graph neural network based on multi-view maps as an encoder. The resulting embedded spatiotemporal map sequences are aligned with textual representations, concatenated with prompt embeddings, and then fed into a frozen LLM to predict future wind turbine power generation sequences. In addition, to mitigate anomalies and missing values caused by sensor malfunctions, we introduce a novel frequency-domain learning-based interpolation method that enhances data correlations and effectively reconstructs missing observations. Experiments conducted on real-world wind power datasets demonstrate that the proposed approach outperforms state-of-the-art methods, achieving root mean square errors of 17.776 kW and 50.029 kW for 24-h and 48-h forecasts, respectively. These results indicate substantial improvements in both accuracy and robustness, highlighting the strong practical potential of the proposed method for wind power forecasting in the renewable energy industry. Full article

► Show Figures

Figure 1

25 pages, 3113 KB

Open AccessArticle

Data-Driven Modeling for a Liquid Desiccant Dehumidification Air Conditioning System Based on BKA-BiTCN-BiLSTM-SA

by Xianhua Ou, Xinkai Wang, Zheyu Wang and Xiongxiong He

Appl. Sci. 2026, 16(1), 304; https://doi.org/10.3390/app16010304 (registering DOI) - 28 Dec 2025

Abstract

The model of a liquid desiccant dehumidification air conditioning (LDAC) system is one of the key foundations for achieving efficient cooling, dehumidification and regeneration, and saving energy consumption. The data-driven modeling method does not need to understand the complex heat and mass transfer [...] Read more.

The model of a liquid desiccant dehumidification air conditioning (LDAC) system is one of the key foundations for achieving efficient cooling, dehumidification and regeneration, and saving energy consumption. The data-driven modeling method does not need to understand the complex heat and mass transfer mechanism and equipment physical information, thus the modeling complexity is greatly reduced. This paper proposes a temperature and humidity prediction model integrating the Black Kite Algorithm (BKA), Bidirectional Temporal Convolutional Network (BiTCN), Bidirectional Long Short-Term Memory (BiLSTM), and Self-Attention mechanism (SA). The model extracts local spatiotemporal features from sequence data through BiTCN, enhances the understanding of contextual dependencies in temporal data using BiLSTM, and employs the SA to assign dynamic weights to different time steps. Furthermore, BKA is adopted to optimize the hyperparameter combinations of the neural network, thereby improving prediction accuracy. To validate the model performance, an experimental platform for an LDAC system was established to collect operational data under multiple working conditions, constructing a comprehensive dataset for simulation analysis. Experimental results demonstrate that compared to conventional time-series prediction models, the proposed model achieves higher accuracy in predicting outlet temperature and humidity across various operating conditions, providing reliable technical support for system real-time control and performance optimization. Full article

► Show Figures

Figure 1

22 pages, 7441 KB

Open AccessArticle

OpenBeePose: A Remote Sensing Framework for Bee Pose Estimation Using Deep Learning

by Sajad Sabzi, Ali Najar, Raziyeh Pourdarbani, Ginés García-Mateos, Ruben Fernandez-Beltran and Mohammad H. Rohban

Appl. Sci. 2026, 16(1), 303; https://doi.org/10.3390/app16010303 (registering DOI) - 28 Dec 2025

Abstract

The application of remote sensing technology for pose estimation in beekeeping has the potential to transform colony management, improve bee health and mitigate the decline in bee populations. This paper presents a novel bee pose estimation method that integrates the accuracy and efficiency [...] Read more.

The application of remote sensing technology for pose estimation in beekeeping has the potential to transform colony management, improve bee health and mitigate the decline in bee populations. This paper presents a novel bee pose estimation method that integrates the accuracy and efficiency of two existing deep learning models: a variant of the classic VGG-19 network architecture for feature extraction and an adaptation of OpenPose for part detection and assembly. The proposed approach, OpenBeePose, is compared with state-of-the-art methods, including YOLO11 and the original OpenPose. The dataset used consists of 400 high-resolution images of the hive ramp (1080 × 1920 pixels) taken during daylight hours from eight different hives, totaling more than 3600 bee samples. Each bee is annotated in the YOLO format with two key points labeled: the stinger and the head. The obtained results show that OpenBeePose achieves a high level of accuracy, similar to those of other methods, with a success rate exceeding 99%. However, the most substantial advantage is its computational efficiency, which makes it the fastest method among those compared for 540 × 960 images, and it is almost twice as fast as OpenPose. Full article

(This article belongs to the Special Issue Remote Sensing Applications in Agricultural, Earth and Environmental Science, 2nd Edition)

► Show Figures

Figure 1

26 pages, 944 KB

Open AccessArticle

Decentralized BIM Workflows with Smart Contract Execution

by Sara Antinozzi, Liliana Cecere, Francesco Colace, Angelo Lorusso, Domenico Santaniello and Carmine Valentino

Appl. Sci. 2026, 16(1), 302; https://doi.org/10.3390/app16010302 (registering DOI) - 28 Dec 2025

Abstract

The integration of Building Information Modelling (BIM), blockchain technology, and smart contracts presents a significant opportunity to fundamentally reevaluate the administration of information and contracts in construction projects. This article introduces a distributed system that amalgamates BIM models, decentralized storage via IPFS, semantic [...] Read more.

The integration of Building Information Modelling (BIM), blockchain technology, and smart contracts presents a significant opportunity to fundamentally reevaluate the administration of information and contracts in construction projects. This article introduces a distributed system that amalgamates BIM models, decentralized storage via IPFS, semantic oracles, and smart contracts to automate essential procedures such as versioning, design verification, and payment issuance contingent upon execution milestones. This proof of concept, built on a Proof-of-Authority blockchain using actual IFC models, demonstrates the technical viability of the method and evaluates its performance, constraints, and operational implications. The applications of SAL automation and design review demonstrate that integrating off-chain verification with on-chain documentation can reduce uncertainty, enhance accountability, and enable hitherto unattainable forms of contract automation. The suggested framework acknowledges the need for improved information standards and Oracle governance, demonstrating that integrating BIM and distributed technologies can significantly transform the digitalisation of the construction sector. Full article

(This article belongs to the Special Issue Smart Buildings: Integrating BIM and Artificial Intelligence for New Challenges of Design)

► Show Figures

Figure 1

30 pages, 23715 KB

Open AccessArticle

Intelligent Landslide Susceptibility Assessment Framework Using the Swin Transformer Technique: A Case Study of Changbai County, Jilin Province, China

by Jiachen Liu, Xiangjin Ran and Xi Wang

Appl. Sci. 2026, 16(1), 301; https://doi.org/10.3390/app16010301 (registering DOI) - 27 Dec 2025

Abstract

Frequent geological hazards such as landslides and rockfalls, intensified by human activities and extreme rainfall, highlight the urgent need for rapid, accurate, and interpretable susceptibility assessment. However, existing methods often struggle with insufficient characterization of spatial heterogeneity, fragmented spatial structures, and limited mechanistic [...] Read more.

Frequent geological hazards such as landslides and rockfalls, intensified by human activities and extreme rainfall, highlight the urgent need for rapid, accurate, and interpretable susceptibility assessment. However, existing methods often struggle with insufficient characterization of spatial heterogeneity, fragmented spatial structures, and limited mechanistic interpretability. To overcome these challenges, this study proposes an intelligent landslide susceptibility assessment framework based on the Swin-UNet architecture, which combines the window-based self-attention mechanism of the Swin Transformer with the encoder–decoder structure of U-Net. Eleven conditioning factors derived from remote sensing data were used to characterize the influencing conditions. Comprehensive experiments conducted in Changbai County, Jilin Province, China, demonstrate that the proposed Swin-UNet framework outperforms traditional models, including the information value method and the standard U-Net. It achieves a maximum overall accuracy of 99.87% and consistently yields higher AUROC, AUPRC, F1-score, and IoU metrics. The generated susceptibility maps exhibit enhanced spatial continuity, improved geomorphological coherence, and greater interpretability of contributing factors. These results confirm the robustness and generalizability of the proposed framework and highlight its potential as a powerful and interpretable tool for large-scale geological hazard assessment, providing a solid technical foundation for refined disaster prevention and mitigation strategies. Full article

(This article belongs to the Section Earth Sciences)

23 pages, 4848 KB

Open AccessArticle

Development Virtual Sensors for Vehicle In-Cabin Temperature Prediction Using Deep Learning

by Hanyong Lee, Woonki Na and Seongkeun Park

Appl. Sci. 2026, 16(1), 300; https://doi.org/10.3390/app16010300 (registering DOI) - 27 Dec 2025

Abstract

The internal temperature of a vehicle is influenced by various factors such as the external environment (temperature, solar radiation, and humidity) and the air conditioning habits of the driver. Even when the air conditioning system is set to a specific temperature, the internal [...] Read more.

The internal temperature of a vehicle is influenced by various factors such as the external environment (temperature, solar radiation, and humidity) and the air conditioning habits of the driver. Even when the air conditioning system is set to a specific temperature, the internal temperature can vary depending on the time, weather, and driver’s manipulation of the system. In this study, we developed and evaluated a deep learning-based vehicle cabin temperature prediction system using CAN (Controller Area Network) data collected from the vehicle and temperature data from thermometers installed on the roof and seats of an electric vehicle (EV). The models used in the temperature prediction system were evaluated by applying various deep learning architectures that consider the characteristics of time series data, and their accuracy was measured using the mean absolute percentage error (MAPE) metric. Additionally, a low-pass filter was applied to the prediction results, which reduced the MAPE from 4.2798% to 4.1433%, indicating an improvement in prediction accuracy. Among the deep learning models, the model with the highest performance achieved an MAPE of 3.5287%, corresponding to an approximate error of 0.88 °C at an actual temperature of 25 °C. The results of this study contribute significantly to enhancing the accuracy and reliability of EV interior temperature predictions, enabling more precise simulations, and improving the thermal comfort and energy efficiency of EVs. The proposed temperature-prediction system is expected to contribute to the comfort of EV users and overall performance of vehicles, thereby strengthening the role of EVs as a sustainable means of transportation. Full article

(This article belongs to the Section Transportation and Future Mobility)

► Show Figures

Figure 1

21 pages, 6350 KB

Open AccessArticle

An Experimental Investigation on the Barrier Performance of Complex-Modified Bentonite

by Jiangdong Xu, Hai Lin, Youshan Su and Shanke Tang

Appl. Sci. 2026, 16(1), 299; https://doi.org/10.3390/app16010299 (registering DOI) - 27 Dec 2025

Abstract

The barrier performance of containment liners against heavy metals and other contaminants is a critical element in ensuring environmental safety. However, the high concentration of multivalent cations in landfill leachate raises concerns about the effectiveness of conventional barriers (e.g., sodium bentonite). To address [...] Read more.

The barrier performance of containment liners against heavy metals and other contaminants is a critical element in ensuring environmental safety. However, the high concentration of multivalent cations in landfill leachate raises concerns about the effectiveness of conventional barriers (e.g., sodium bentonite). To address concerns regarding the high permeability and elevated heavy metal concentrations in effluents from sodium bentonite (Na-B) barriers, this study proposes the use of new complex-modified sorbent bentonite—specifically treated with disodium ethylenediaminetetraacetate (EDTA-2Na) and sodium tripolyphosphate (STPP). Batch adsorption and flexible-wall permeability tests in extreme synthetic leachate demonstrate that the complex-modified sodium bentonite not only maintains low permeability but also enhances contaminant adsorption capacity of barriers. When modified with 2% EDTA-2Na and 4% STPP (by mass), the maximum Zn(II) adsorption capacity of bentonite was measured at 43.22 and 48.22 μg/g, respectively. These values correspond to enhancements by a factor of 1.99 and 2.32 compared to the unmodified Na-B. Simultaneously, the hydraulic conductivity met the permeability requirements for engineering barrier systems (k < 1 × 10⁻⁷ cm/s) throughout the tested range of confining pressures. Microscopic analyses confirmed the successful incorporation of functional groups into bentonite by both EDTA-2Na and STPP. STPP-induced electrostatic repulsion, promoting ordered particle stacking and dense structure formation. EDTA-2Na physically filled pores to block ion migration pathways while electrochemically counteracting double-layer compression under high ionic strength. This effective strategy resolves the long-standing trade-off between permeability and adsorption capacity in conventional bentonite, providing a theoretical basis for designing barrier materials in complex contaminated sites. Full article

(This article belongs to the Special Issue Advanced Geotechnical Engineering Technology and Application in Environmental Detection)

► Show Figures

Figure 1

Journal Description

Applied Sciences

Latest Articles

Journal Menu

Journal Browser

Highly Accessed Articles

Latest Books

E-Mail Alert

News

Topics

Conferences

Special Issues

Topical Collections

Further Information

Guidelines

MDPI Initiatives

Follow MDPI