In this letter, a compact high-gain full-space beam scanning leaky-wave antenna based on dual-layer spoof surface plasmon polaritons (SSPPs) is proposed. By exploiting the asymmetric characteristics of the dual-layer SSPPs unit, the proposed antenna enables independent eigenmode dispersion tuning and effective open-stopband (OSB) suppression, while maintaining mode-1-dominant propagation. Moreover, lateral slots are introduced to redistribute the surface current and stabilize the leakage characteristics near broadside. Compared with equal-length-stub SSPP units, the proposed unit provides a higher phase constant, leading to an enhanced scanning rate. The antenna is fabricated and measured, and the measured results agree well with the simulations. With an overall length of 5.23λ0, the antenna achieves continuous full-space beam scanning over 7.53 − 11.31 GHz with a scanning rate of 4.67°/%. The realized gain ranges from 8.28 to 11.32 dBi, corresponding to a realized gain per unit length of 2.01 dBi/λ0. Owing to effective OSB suppression, the gain fluctuation around broadside is less than 0.5 dB, while the radiation-efficiency variation is below 5%. In addition, the radiation efficiency exceeds 70% across the entire operating band, and the sidelobe level remains below −8 dB.

Multimodal Emotion Recognition in Conversations (MERC) represents a pivotal research avenue within the realms of human-computer interaction and affective computing. The fundamental challenge is to develop effective models for multimodal contextual information and to integrate complementary multimodal data. Given the superior performance of Graph Neural Networks (GNNs) in relation modelling, this paper puts forward a Graph Based Cross-modal Information Fusion (GCIF) for conversational emotion recognition to tackle the problems of redundant information generation, information loss and over-smoothing that have been identified in the existing GNN methods for multimodal information fusion. The GCIF builds a graph network with a configurable fixed context window and integrates speaker information for multimodal data. The generality space and individuality space are created in order to maintain the consistency and specificity of multimodal features. GCIF employs a module based on the improved graph attention networks to achieve pairwise fusion of multimodal information, thereby reducing the difficulty of multimodal fusion and effectively alleviating the problems of heterogeneity and over-smoothing. The experimental results on the two public benchmark datasets demonstrate that our GCIF can effectively promote performance for MERC. Furthermore, this paper conducts extensive experiments to discuss and analyze the impact of different settings on the performance of GCIF, thereby validating the effectiveness of the model.

This study addresses the critical challenge of incomplete downhole drilling tool attitude data in rotary steerable drilling systems, a significant issue impacting drilling trajectory control, efficiency, and safety. Attitude data loss often occurs due to sensor malfunctions, communication errors, and harsh drilling environments, posing risks to accurate geological modeling and decision-making. Traditional interpolation methods struggle with the nonlinear, time-dependent nature of drilling data, resulting in poor accuracy and reliability. To overcome these limitations, a Long Short-Term Memory (LSTM) model, a type of deep learning algorithm known for capturing complex time-series dependencies, is proposed for data completion. This model was trained and validated using a dataset collected from a rotary steerable system laboratory test platform, including acceleration, magnetic flux, and Hall sensor pulse signals under varying inclination and rotational speeds. Experimental results show that the LSTM model achieves superior performance, with a coefficient of determination (R2) exceeding 0.95 and significantly lower mean squared errors compared to Fully Connected Neural Networks (R2 = 0.88) and other regression-based methods. The completed attitude data closely replicates original signal trends, accurately reconstructing gravitational acceleration, magnetic flux, and TICK signals despite noise and data loss. This work goes beyond previous studies by applying deep learning specifically to downhole drilling tool attitude data completion, a novel approach not previously explored in petroleum exploration literature. The demonstrated accuracy and robustness of the LSTM model provide a reliable, data-driven solution for enhancing drilling path optimization, reducing non-productive time, and minimizing operational risks. This research contributes a scalable method to improve drilling efficiency and resource recovery rates, with potential applicability to broader data-driven drilling automation systems.

Existing multimodal sentiment analysis (MSA) methods usually adopt fixed convolution kernels or static windows to model features from limited or fixed scales, making it difficult to dynamically model emotional features under different scale combinations. Furthermore, the absence of mechanisms to suppress redundant information in non-linguistic (video and audio) modalities hinders further performance improvements. To address these limitations, we propose a text guided multimodal scale path fusion network (TMSPF-Net). TMSPF-Net contains three main modules: Multi-scale Adaptive Transformer (MAT), Text-guided Conflict Elimination Module (TGCEM), and Channel Fusion Module. MAT captures the interaction of intra-modal and inter-modal through the combination of patches of different sizes and the dual attention mechanism, fully extracting multi-level global and local emotional information. Meanwhile, the adaptive routing module in MAT dynamically optimizes the feature paths through a learnable mechanism, enabling MAT to adaptively select the optimal path and increasing the flexibility of the model when dealing with heterogeneous data. TGCEM leverages multi-scale text-guided dynamic memory in MAT to filter conflicting signals and selectively preserve emotionally salient patterns in non-linguistic modalities, thereby improving the consistency and semantic richness of multimodal representations. Channel Fusion Module fuses the output results of these two modules and inputs them into the pre-trained language model to complete the MSA task. Extensive experiments on the MOSI and MOSEI datasets demonstrate that TMSPF-Net outperforms in most metrics than state-of-the-art methods. The results show that TMSPF-Net effectively guides the learning of non-linguistic modalities, integrates multi-level sentiment features, showing great potential in sentiment analysis.

This paper proposes a machine learning technique based on “training while drilling” to improve the integrity and accuracy of resistivity logging data in the oil and gas industry. Resistivity logging is a critical tool for assessing hydrocarbon content in geological formations; however, it often suffers from data loss due to hardware failures caused by the harsh high-temperature conditions encountered downhole. To address these challenges, this study introduces Long Short-Term Memory (LSTM) neural networks, a deep learning technique well-suited for handling time-series data and predicting missing values. Unlike traditional surface-based data processing, this method integrates the trained LSTM model into embedded devices, enabling real-time downhole data completion and significantly enhancing the efficiency and immediacy of data processing. “training while drilling” refers to the continuous learning and model updating based on newly acquired data during drilling operations. This process utilizes downtime, such as during drill pipe or bit changes, to input fresh data as a training set, allowing for real-time model optimization. Experimental results demonstrate that this method notably improves the predictive accuracy of the LSTM model, with enhanced capability to capture detailed information and adapt to changes in formation characteristics when compared to the initially trained model. Moreover, the LSTM model outperforms other deep learning algorithms, including Fully Connected Neural Networks (FCNN), in terms of prediction accuracy and stability. The dataset used in this study is based on field data from the Shengli Oilfield in Dongying, with measurements spanning depths of 1140 to 1690 m. Through comparative experiments, the LSTM model achieved a mean squared error (MSE) of 0.0610, significantly lower than the MSE of 0.0961 observed for the FCNN and other traditional methods. These findings highlight the superior performance of the “training while drilling” approach in improving prediction accuracy of electromagnetic wave logging curves, providing a novel solution with considerable practical value and broad potential for future development in the oil and gas exploration sector.

A novel grating-free tunable external cavity diode laser has been designed and optimized, featuring an impressive phase-continuous tuning range exceeding 6 THz. The laser cavity is constructed from two sets of right-angled mirrors arranged in a rectangular configuration. These mirrors are symmetrically positioned relative to the central axis and are parallel in pairs, effectively forming two periscopic assemblies. Wavelength adjustment is achieved by rotating a double-sided mirror positioned centrally within the mirror sets, and an etalon is utilized as the wavelength selector. This pivot-point-free rotation simultaneously tunes both the cavity length and the etalon mode wavelength, ensuring high synchronization between the two. With a low refractive index material used for the etalon, the tuning range can theoretically extend up to 6.022 THz.

Emotion Recognition in Conversations (ERC), a pivotal technology for augmenting the empathy capacity of human-computer interaction systems, continues to encounter substantial challenges in its multimodal fusion mechanism. Owing to their advanced relational modelling capabilities, graph neural networks (GNNs) have proven to be advantageous in the domain of multimodal learning. GNNs utilise graphs constructed with multimodal data to facilitate information interaction within and between modalities, thereby achieving comple mentarity and fusion of multimodal data. The majority of prevailing MERC methodologies founded upon GNNs are inherently incapable of leveraging the full potential of direct inter-modal cues. The simultaneous extraction of information from both within and between modalities invariably gives rise to conflicts in the context of multi modal data integration. In this work, we propose a dynamic alternating graphs based cross-modal complementary fusion for conversational emotion recognition (DAGCCF). The DAGCCF comprises two fundamental modules: the DAGP and RMGAT modules. The DAGP module models intra-modal and inter-modal relationships in an alternat ing, layer-wise manner, thereby reducing conflicts and redundancies during multimodal data fusion. Meanwhile, the RMGAT module alleviates the over-smoothing issue inherent in GNNs, while simultaneously enhancing the fusion of multimodal contextual information and inter-modal relational features. The application of DAGCCF can be performed in a concurrent manner with regard to multimodal emotion recognition in conversation (MERC) and multimodal sentiment analysis in conversation (MSAC). The new framework provides a unified network for multimodal learning. The experimental findings, derived from the analysis of numerous datasets, demonstrate the preeminence of our DAGCCF over contemporary state-of-the-art methodologies. This paper further explores and examines the impact of various settings on the efficacy of DAGCCF through a substantial number of experiments, thereby providing additional validation for the model.

Electromagnetic wave logging while drilling (LWD) technology is an important tool for evaluation of formation oil and gas content. It generally adopts multi-transmitter–receiver coil system structure and the symmetrical coil system arrangement with equal transmitter–receiver spacing can obtain the measurement results with borehole compensation. Here, we develop a method to realize wellbore compensation by deep learning for logging data inversion. This paper focuses on reasonable inversion of logging data through deep learning technology, which is combined with Levenberg–Marquardt (LM) algorithm, modular and fast construction of deep neural network (DNN) model. Under the condition of reducing the outermost transmitting coil, the logging data are inversed, and the inversion effect is evaluated. Our research shows that the combination of neural network and logging data can realize the measurement results with borehole compensation under the condition of reducing one transmitting coil, thereby shortening the instrument length to reduce drilling tool sticking risk and effectively reducing the LWD instrument structure complexity, high power and other problems. At the same time, the accuracy of logging data inversion is tested. The test results show that the DNN method can achieve high-precision inversion, and the average error is reduced by about 50

Emotion recognition in conversation (ERC) represents a pivotal research domain within affective computing, concentrating on discerning the emotional nuances embedded within individual utterances during conversational exchanges. The majority of current research focuses on modeling situational cues, with relatively little attention paid on affective tendencies inherent in emotional expression. Moreover, ensuring the fair representation of diverse modalities in emotional expression presents a significant challenge in effectively extracting synergies and insights from multi-modal data sources. To tackle these challenges, this study proposes a novel approach termed the Uni-Modal and Cross-Modal Encoding Network based on Multi-Head Attention (UCEMA) for ERC. The framework leverages two distinct encoding techniques, namely Uni-Modal Encoding based on Multi-Head Attention (UEMA) and Cross-Modal Encoding based on Multi-Head Attention (CEMA), to extract distinct emotional features from individual modes and facilitate the fusion of emotional attributes within the multi-modal context. Particular emphasis is placed on textual input as the primary mode of interaction. Additionally, this study employs Context Modeling (CM) to analyze the outcomes of emotion recognition in conversational contexts. A comprehensive comparative analysis of the UCEMA was conducted on two publicly available datasets, IEMOCAP and MELD. The results demonstrated that the UCEMA exhibited superior efficacy. It is noteworthy that the proposed framework effectively balances intra-modal emotional orientation information, inter-modal emotional association information, and context-related cues, thereby demonstrating superior performance in recognition accuracy compared to current state-of-the-art (SOTA) models.

In the paper, a circularly polarized (CP) frequency beam-scanning antenna based on photonic crystal (PC) is proposed, which possesses the characteristics of eliminating open-stopband (OSB). In the microwave field, the PC is typically constructed by embedding the air or metal column into the material. However, it is difficult for the PC with perfect magnetic conductor walls to accomplish the degeneracy of eigenmodes at the Gamma point. Additionally, antennas based on the PC with perfect electric conductor walls exhibit a narrow scanning angle. The proposed unit based on 1D PC possesses perfect electric and magnetic conductor walls. The unit can not only eliminate OSB but also has a wide scanning range. In addition, the unit breaks into the original symmetrical field distributions of 1D PC. Based on the perturbation theory, leaky-wave slots can be designed to generate electromagnetic waves with the CP characteristics. And a CP antenna based on the unit is designed. The antenna can achieve a 70 degrees beam-scanning within the entire operating band (7.6-9.2 GHz), and the variation of the realized gains is below 1.0 dBi around the broadside frequency. Simultaneously, the antenna is fabricated and measured. The measured results are in accordance with the simulation, which validates the success of the design.

With the diminishing availability of oil and gas resources, the Rotary Steerable System has become increasingly important. However, the vibrations and shocks during the drilling process pose challenges to the Measurement While Drilling. In recent years, the application of machine learning in the field of petroleum exploration has gradually expanded, especially in the estimation of geological parameters and lithology discrimination. However, there is still limited research on drilling tool attitude measurement. To address the above-mentioned issues, this study proposes a method that combines drilling tool attitude sensor data with artificial neural networks to improve the accuracy of dynamic inclination measurement. This method utilizes machine learning techniques, combining real-time z-axis acceleration signals and magnetic induction signals, and employs a deep learning model to invert the x and y-axis acceleration signals, thereby achieving high-precision measurement of dynamic inclination angles. Experimental results show that Long Short-Term Memory model, under simulated measurement conditions with different rotational speeds, yields dynamic inclination curve errors ranging from 0.4° to 0.7°, significantly reducing the errors compared to the original measurements. This method not only improves the accuracy of inclination angle measurement but also demonstrates strong adaptability to different rotational speeds, providing more accurate data support for drilling operations.

We experimentally demonstrate a silicon chiral mode switching device by dynamically encircling exceptional point at 2 μm waveband, with high purities (> 95%) for both TE0 and TE1 modes in a broad bandwidth (85 nm).

A new type of low terahertz (THz) leaky-wave antennas (LWAs) with circular polarization (CP) characteristic is proposed in this paper. The CP LWA is based on the asymmetric structure and the accidental degeneracy of modes at the Γ-point in 1D photonic crystals. Because the composite right/left handed and CP properties can be achieved by unit cells of the proposed antenna, as well as the refractive index is zero at the broadside frequency, the beam scanning of the proposed CP LWA is from backward direction to forward direction through the broadside. There won’t be any deterioration of broadside performances, especially the gain whose fluctuation is only 0.25 dBi at and around broadside frequency. The gain varies from 14.85 to 16.75 dBi and the scan range is from − 23.5° to 40.4° over the entire operating frequency band. By optimizing the “mouth” slots, the axial ratio (AR) is less than 3 dB from 1.387 to 1.639 THz.

Hydrogels are effective adsorbents for removal of heavy metal ions from wastewater due to their three-dimensional network and functional groups. In this paper, an eco-friendly hydrogel based on the polysaccharide extracted from the over-breeding macroalgae Enteromorpha prolifera was constructed and evaluated as an adsorbent for heavy metal ions. DFT calculation results confirmed that the carboxyl and sulfate groups of the Enteromorpha prolifera polysaccharides had certain chemical interactions with Pb2+, Cu2+ and Cd2+, which guarantee the high adsorption capacity for them. The interpenetration strategy used to construct the polysaccharide-based double network hydrogel, through which the mechanical properties, adsorption performance and recyclability of the hydrogel get strengthened synchronously. Overall, the constructed hydrogel is a competitive candidate material for heavy metal ion adsorption because of its low cost and biocompatibility, through which an innovative application of Enteromorpha prolifera from waste to treasure has been realized.

A dual-band dual-polarized frequency beam-scanning (FBS) antenna based on half-mode substrate integrated waveguide (HMSIW) and photonic crystal (PC) is proposed. HMSIW can decrease the width of the substrate, and each eigenmode of PC has own unique electric-field distribution, which can be good for generating different polarization types. Therefore, the proposed antenna not only has a narrow substrate, but also has the charac-teristic of circularly polarized (CP) and linearly polarized (LP) radiation, which can be used for overcoming effectively the multipath fading and enhancing the channel capacity. For increasing the bandwidth of 3 dB axial ratio (AR), the slots on top surface are not perfectly symmetrical in the improved unit cell. The asymmetrical antenna based on the improved unit is manufactured and measured. The antenna can accomplish 33 degrees scanning angle with the realized gains of 7.9-10.6 dBi in the CP working band (7.2 to 8.2 GHz), and can also achieve 24 degrees beam-scanning for the LP radiation (8.8 to 9.6 GHz), whose realized gains are from 9.5 to 11.5 dBi. In addition, the experimental results are highly consistent with the simulated results, verifying that the design is feasible.

This letter proposes a novel simple planar circularly polarized (CP) frequency beam-scanning (FBS) antenna using slotline (SL) and spoof surface plasmon polariton (SSPP) structure. A new SL-SSPP slow-wave transmission line (TL) is initiated by periodically introducing orthogonal tilted corrugated grooves on both sides of the SL, which can generate orthogonal electric-field components in the same cross-section along the TL, and its dispersion characteristic can be flexibly adjusted. In order to excite an effective radiating mode from the propagating slow-wave mode, a simple double-sided tilted modulation method is devised to periodically modulate the SL-SSPP TL. Furthermore, to obtain the CP characteristic, a π/2 phase displacement between the sinusoidal modulation structures on both sides of the TL is designed. The configuration of presented CP beam-scanning antenna owns the merits of simpler structure, low profile, and single dielectric layer. A prototype is manufactured and then tested to verify the proposed design. The experimental results confirm that, within the 3 dB axial ratio bandwidth (from 7.2 to 8.35 GHz), the designed antenna achieves a 35° beam-steering range with CP characteristic. This antenna also realizes appropriate gain and radiation efficiency in the working band.

Facing the increasingly serious water pollution problem, photocatalytic method has been recognized as one of the best solutions due to its convenience, efficiency and low consumption. Graphitic carbon nitride (g-C3N4), a metal-free and visible light-responsive photocatalyst, has shown great potential in the photocatalytic treatment of water pollution. This study describes a novel bottom-up strategy for constructing visible light-responsive photoactive nanocomposite g-C3N4/CNC/PAM aerogel with the g-C3N4 uniformly loaded throughout a polymeric network, perfect mechanical property and high performance in treating wastewater containing organic/heavy metal compound contaminants. Aldehyde-modified negatively charged natural nanoparticle cellulose nanocrystal (CNC-CHO) was used to achieve uniform dispersion of g-C3N4, and also participate in the subsequent formation of aerogel through chemical crosslinking with photoinitiated polyacrylamide (PAM) catalyzed by g-C3N4 without additional crosslinkers. The optimized porous structure of the prepared aerogel ensured high swelling rate of up to 2500%, excellent pollutant adsorption performance and high light transmittance, thus achieved high removal rate and removal efficiency for different types of pollutants including organic dyes (RhB, MB and CR) and heavy metal Cr(VI). The removal efficiency could get 99% for RhB (5 mg/L) in 6 h, and 95% for Cr (VI) (10 mg/L) in 100 min under visible light. Further more, the aerogel showed excellent mechanical stability and reusability, with no loss in activity observed after over 5 cycles of dye degradation.

Heavy metal ions produced by industrial activity have been a serious environmental problem, and their detection is critical for treatment of the heavy metal pollution. Among the variable heavy metal detection methods, fluorescent indicator methods have attracted wide attention due to the advantage of the convenience and nondestructive detection process. Carbon quantum dots have great application in this respect. In this study, nitrogen and sulfur doped fluorescent carbon quantum dots (N,S-CDs) were prepared based on wool keratin using a hydrothermal method, of which the morphology, chemical composition and optical properties were characterized. The prepared N,S-CDs were spherical nanoparticles with a diameter of 2-6 nm, showing wide fluorescence excitation and emission wavelength range and considerable quantum yield, and have sensitive response to Cr6+ and Fe3+. This study reveals a novel cost-effective and convenient synthetic route of green carbon quantum dots through naturally sourced materials, demonstrates the application potential of the wool keratin-based N,S-CDs in rapid detection of heavy metal, and opens up a new path for functional utilization of waste wool keratin.

Prussian blue (PB) is an anodic coloring candidate in the wide area of electrochromic (EC) applications. However, the co-influence of weak adhesion and low electrical conductivity leads to the poor stability and slow switching speed. To tackle this bottleneck, a novel TiO2/Au/PB nanorod array is designed through hydrothermal and electrodeposition approaches on fluorine-doped tin oxide (FTO) glass. Such a designed ternary array structure could not only increase reactive site and conductivity, but also improve ion storage capacity and promote charge transfer, attributed to the synergistic effect of TiO2/Au/PB core—shell heterostructure and the localized surface plasmon resonance (LSPR) effect of Au nanoparticles. Besides, density functional theory (DFT) calculation confirms the strong interaction between rutile TiO2 and FTO substrate, which contributes to the improvement of EC cycle stability. Benefiting from these effects, the TiO2/Au/PB film shows a fast coloration/bleaching response of 1.08/2.01 s (2.17/4.48 s, PB film) and ultra-stable EC performance of 86.8

In this letter, an explicit and absolutely stable finite-difference time-domain (FDTD) algorithm is designed for electromagnetic analysis. The algorithm works through a structure composed of several vectors and matrices acting on these vectors. Excitation source is linearly approximated in a time interval, fields in the computation domain are expressed by a vector, and these matrices are derived based on the FDTD method. The proposed algorithm solves electromagnetic problems in an explicit way, its time step size is beyond the Courant-Friedrich-Levy (CFL) stability condition and the computation efficiency of the proposed method is also higher than the conventional FDTD method. Two numerical examples are tested and validate that the proposed algorithm can solve electromagnetic problems correctly and also improves calculation efficiency.