Search

With the continuous development of the smart energy systems for carbon reduction and energy saving, the problem of short-circuit current exceeding the standard has become increasingly prominent. While adjusting the system operating mode and installing current-limiting equipment can suppress short-circuit current effectively, these measures may influence the system stability and increase costs. Therefore, a method for short-circuit current suppression based on instant system reconstruction is proposed in this paper. The breaking current of circuit breaker for typical substations is firstly analyzed. Then three control strategies for transient topology reconstruction are proposed, and their applicability is analyzed to provide diverse options for practical engineering applications. Finally, considering the requirement for the system topology reconstruction, the hardware and software architecture for high-speed protection control is constructed. The fast estimation of current state is realized using the Kalman estimator to meet the requirements for short-circuit current suppression. The simulation result shows the effectiveness of the proposed method.

The goal of event detection is to identify instances of various event types within text. In real-world scenarios, multiple events often coexist within the same sentence, making the extraction of these events more challenging than extracting a single event. While graph neural networks operating over dependency parsing trees have shown some capability in handling multi-event scenarios and improving event detection effectiveness, their improvement is limited. This limitation arises because dependency trees cannot automatically establish connections between trigger words and other key words, which are crucial for recognizing and classifying trigger words. Additionally, syntactic-based methods typically focus on the closest neighbors in the dependency graphs to aggregate information for the trigger candidate word, even though relevant words are often multi-hop away. In this paper, we combine the word dependency graphs with our automatically induced latent graph structure for event detection and multiple events detection. Furthermore, we propose two regularizers to enhance the representation of the dependency graphs and the induced latent graph structure. Experimental results demonstrate the effectiveness of our model for events detection.

The presence of trace metal in transformer oil significantly affects its insulating properties, posing potential risks to the safe operation of transformers. Therefore, understanding the intrinsic mechanisms behind the formation of the trace copper within transformers is of critical importance. In this study, we investigated the two primary forms of trace copper in transformer oil—suspended particulate and soluble ionic states—and conducted detailed experimental analyses to elucidate their formation mechanisms. Through electrical arcing experiments, we examined the behavior of metal contacts in insulating oil, identifying the transfer of surface material during arc interruption as a crucial pathway for the generation of suspended metallic particulates. Additionally, we explored the impact of electrical arcing on the damage mechanisms of the contacts and the resulting microstructural characteristics of the ablated surfaces. Corrosion tests were employed to study the formation mechanisms of metal ions, with both macroscopic and microscopic analyses performed on different corrosion morphologies. Our findings suggest the presence of copper corrosion products and their soluble forms, specifically identifying copper-oleate as a major component, which demonstrates high solubility and diffusion capacity in the oil.

Tractors are among the most widely used machinery in agriculture. However, low-frequency vibrations during tractor operations pose significant health risks to drivers, such as musculoskeletal disorders, and negatively impact ride comfort. Current approaches rely on offline comfort prediction models, which lack real-time feedback, making them unsuitable for practical field applications. To address this gap, we propose a real-time recommendation system based on Internet of Things (IoT) and Machine Learning (ML) to enhance the driving comfort of agricultural tractors. Our low-cost IoT-enabled solution is compatible with existing tractors, requiring no expensive intelligent upgrades. Using the XGBoost model for ride comfort prediction, we achieved superior performance (R $^{2} =0.96$ , RMSE =0.015) compared to other ML models. Additionally, the Particle Swarm Optimization (PSO) algorithm is employed to recommend optimal operational parameters, reducing the ride comfort value (OVV) by 6.67% in real-time experiments. This study highlights a scalable, data-driven approach for improving tractor comfort and offers a reference for intelligent control strategies in Agriculture 4.0.

This paper introduces the POMA-C (Partial Observable Model for Anesthesia Control) framework, developed to address the challenge of anesthesia management in environments with incomplete physiological monitoring, such as low-resource settings where critical indicators like the Bispectral Index (BIS) are often unavailable. Unlike traditional methods that rely on fully observable data, POMA-C frames the problem of anesthesia control under incomplete observability within a Partially Observable Markov Decision Process (POMDP), enabling the precise control of anesthesia despite missing data. By establishing a formal correspondence between the anesthesia control process and POMDP, this framework provides a theoretical foundation for modeling anesthesia control in uncertain environments. The framework employs the POMCPOW (Partially Observable Monte Carlo Planning with Observation Weighting) algorithm, which integrates Monte Carlo Tree Search (MCTS) and particle filtering to estimate the patient’s true physiological state and guide optimal anesthetic decisions. Through comprehensive ablation experiments—where key observation dimensions are systematically reduced to simulate missing data—POMA-C demonstrates significantly higher decision accuracy and cumulative reward optimization compared to methods like Q-learning and human expertise, even in data-constrained environments. This work not only provides a robust solution for anesthesia control under incomplete observability but also bridges the gap between MDP and POMDP models, offering a foundation for future research in automated anesthesia management.

Automatic traffic accident detection has attracted the attention of the machine vision community for the rapid development of autonomous intelligent transportation systems (ITS). However, previous studies in this domain have been constrained by small-scale datasets with limited scope, impeding their effectiveness and applicability. Specifically, highway traffic accidents, often resulting in severe consequences due to higher speeds, require a more comprehensive approach to detection. The use of video surveillance provides a unique perspective, capturing the entire accident sequence. Unfortunately, existing traffic accident datasets are either not sourced from surveillance cameras, not publicly available, or not tailored for highway scenarios. An open-sourced traffic accident dataset with various scenes from surveillance cameras is in great need and of practical importance. To fulfill the above urgent need, we endeavor to collect abundant video data of real traffic accidents and propose a large-scale traffic accidents dataset, named TAD. Various experiments on image classification, video classification, and object detection tasks, using public mainstream vision algorithms or frameworks are conducted in this work to demonstrate the performance of different methods. The proposed dataset together with the experimental results are presented as a new benchmark to improve computer vision research, especially in ITS. The dataset is publicly available at https://github.com/UnicomAI/UnicomBenchmark/tree/main/TADBench.

The rapid advancement of brain-computer interface (BCI) technology has created interest in steady-state visual evoked potential (SSVEP)-based BCIs, which are valued for their high information transfer rates and ability to manage multiple targets. Nonetheless, the efficacy of SSVEP decoding is often constrained by the volume and duration of user calibration data, limiting its practical application. Generative adversarial networks (GANs) have shown promise in synthesizing SSVEP electroencephalogram (EEG) data. However, they face challenges, such as low signal-to-noise ratio and capturing both temporal and spatial features. To address the need for high-quality data generation, the Attention-Aided Classifier Wasserstein GAN (AAC-WGAN) is proposed, which is a novel GAN model that combines an Attention Mechanism and an Auxiliary Classifier to improve data quality and classification performance. Our experiments on the Direction and Dial datasets reveal that our model performs obviously better, particularly with a training set ratio of 25% synthetic data and 75% real data. It achieves a classification accuracy of 91.65% on the Direction dataset, which is a significant improvement over the baseline accuracy of 84.32% (p =0.008) and outperforms other comparable models (p

Air traffic controllers play a crucial role in overseeing and managing dynamic flight operations. The issue of their fatigue is intrinsically linked to aviation safety, thus detecting fatigue states in controllers is vital for operational control and training. This paper provides a comprehensive review of the methods and research on detecting fatigue in air traffic controllers (ATCOs). It initially reviews the historical advancements and pertinent literature concerning the management and prevention of controller fatigue. Moreover, it examines the behavioral traits, physiological signals, and current single-source data methods utilized for detecting fatigue in ATCOs. The paper also underscores the fusion detection methods, models, and multi-source data-based research, serving as references for exhaustive and real-time analysis of controller fatigue states. Finally, it suggests future research directions for fatigue risk prevention in ATCOs, based on the existing studies.

As university information management grows increasingly complex, the demand for seamless data sharing and interaction across various departments and faculties has surged. However, many universities lack a unified platform for data traceability and sensitive information protection, heightening security risks during frequent data exchanges. Furthermore, the uneven allocation of network resources and the absence of effective traffic analysis tools hinder universities from efficiently managing these resources, exacerbating the challenges of network management. Tight budgets further complicate cybersecurity infrastructure development, particularly as universities expand and diversify their disciplines, leading to even more imbalanced funding and constrained investment in network security.To address these challenges, this paper presents a data traceability system integrating Deep Packet Inspection (DPI) and Deep Flow Inspection (DFI) technologies. While DPI and DFI have been extensively studied in the literature, the innovation of this system lies in its hybrid approach, which combines the strengths of both technologies to form a complementary mechanism. This approach is particularly suited to the financial and technical limitations faced by universities. The system is designed for ease of implementation, management, and use, and has been optimized for the unique operational needs of higher education institutions.This research adopts a combination of action research and case study methodologies, starting with small-scale trials to assess system performance before expanding the scope to gather data from broader applications. In addition, this paper introduces the concept of tiered data management for the education sector, offering valuable insights into data classification and traceability within university information systems. The proposed system not only enhances the efficiency of data security management but also provides a practical solution for building future cybersecurity frameworks in higher education.

The importance of efficient path planning (PP) cannot be overstated in the domain of robots, as it involves the utilization of intelligent algorithms to determine the optimal trajectory for robot to navigate between two given points. The main target of PP is to determine potential trajectories for robot operating in a complex environment containing various obstacles. The implementation of these movements should facilitate robot in traversing a path without encountering any collisions, starting from its initial location and reaching the intended destination. In order to address the challenges associated with robot PP, this study applies the chimp optimization algorithm (CHOA) as a local searching (LS) technique and the evolutionary programming algorithm (EPA) to enhance the potential route discovered via a collection of LSs. In order to address CHOA’s tendency to converge to local minima, a new updating technique called twin-reinforced (TR) is developed. In order to assess the effectiveness of the TRCHOA, we conducted a comparative analysis with other widely used meta-heuristic algorithms that are typically employed for solving robot PP problems. Additionally, we included the conventional probabilistic roadmap method (PRM) in our evaluation. We evaluated the planning performances of these algorithms on a standardized set of benchmark problems. Our findings indicate that the TRCHOA outperforms the other algorithms in terms of its planning performance. The evaluation of planning effectiveness encompasses several key criteria, namely path length, consistency of scheduled paths, time complexity, and rate of success. The experiments conducted in this study provide evidence of the statistically significant value of the enhancements obtained through the implementation of the proposed method. The findings derived from the TRCHOA provide compelling evidence of its capacity to accurately determine the most optimal route within the specified test map.

The similarity between arc fault current waveforms and nonlinear load currents can lead to misjudgments in arc fault identification methods that rely on arc current. In response to this issue, this paper establishes a unified fault criterion based on the conclusion that different types of loads that exhibit similar magnitudes of load side voltage sag during arc faults. Combined the voltage sag criterion with waveform criteria, this paper proposes a novel arc fault detection method. The method takes the load side voltage sag value exceeding the threshold as the detection start condition. After detection starts, the method identifies faults by calculating the degree of load side voltage waveform distortion using Hausdorff distance algorithm. Simulations are conducted on five different loads to determine the startup threshold and detection threshold. Test results show that the fault identification accuracy of the proposed method is 100% under the above thresholds.

In this paper, we consider a simultaneous wireless information and power transfer (SWIPT) system adopting hybrid analog-digital beamforming where a base station (BS) with massive antennas transmits different data streams to multiple single-antenna devices. Our primary objective is to minimize the transmit-power of the BS while ensuring energy harvesting and signal-to-interference-plus-noise ratio (SINR) requirements of each device. We specifically highlight that our approach can accommodate the scenarios where users have different requirements on the harvested energy and the received SINR. We propose two suboptimal solutions in which digital and analog precoders are designed jointly. In the first solution (OptDig-ProRF), the objective function is approximated to be independent of the digital precoder and power splitting ratios, and the local minimum of the objective function is found to obtain an analog precoder. Subsequently, the optimal digital precoder and power splitting ratios are found for the obtained analog precoder using semidefinite relaxation of the original problem. In the second solution (ZfDig-SohRF), to reduce computational complexity, we newly design a zero-forcing (ZF) digital precoder which meets users’ requirements while applying the corresponding analog precoder proposed in the literature. The two precoders are iteratively found until the transmit power difference between two consecutive updates falls below a threshold. Illustrative results show that the OptDig-ProRF scheme yields a very close performance to the fully digital precoder in terms of transmit power, while the ZfDig-SohRF scheme outperforms the optimal digital precoder plus a conventional analog precoder scheme at high SINR constraints.

In short-term load forecasting, deep learning models with recurrent units are widely used. For Long Short-term Memory (LSTM) models, in terms of the input features at one particular time-step, there is a weight series for each input feature. Hence, an input feature with a larger value of weight can be regarded as more important to the final forecasting result. Since the weights of the features are time-invariant, the model considers the input features as equally important at each time step, thus lack of the capability of reflecting the importance fluctuation through the time span. To tackle this issue, this paper proposes a forecasting scheme, which integrates mutual information (MI) into bi-directional long short-term memory (BILSTM) network. The MI method is used to extract the importance value of input features at different time-steps, and constitute the fluctuation matrix of the importance values for input features, which is used as coefficients to correct the original input feature. The corrected features are substituted into the BILSTM network, which further improves the forecasting accuracy. Numerical experiments have been carried out based on real-world load data. The results demonstrate the superior performance of the proposed scheme compared with a series of state-of-the-art methods and thus indicate the importance and effectiveness of capturing the fluctuation of the importance value.

There are subtle differences between single regions of the same subcategory in fine-grained images. At present, many fine-grained image classification networks often focus on a single region to determine the target category. However, in many cases, most discriminative features in fine-grained images are distributed in multiple local regions of the image, and it is not often enough for fine-grained image to rely solely on one region.To solve these problems, a new method is proposed. This method generates discriminative features through reinforcement learning and obtains complementary regions through complementary network. The reinforcement network and the complementary network learn through adversarial learning and improve the accuracy of fine-grained images classification.The method is tested on CUB200-2011,fine-grained Visual Classification of Aircraft, and Stanford dogs datasets and the results show adequate performance.

In this work, a tapered fed circular-shaped monopole antenna for ultrawideband (UWB) applications integrated with four frequently-used band-notched properties is designed. By the design of notch structures inside the radiation patch, on the feedline, and on either side of the feedline, four notch bands covering from3.3–3.8 GHz, 5.25–5.75 GHz, 7.25–7.75 GHz, and 8.01–8.5 GHz are obtained possibly, which suppress the interference with WiMax, WLAN, satellite downlink and ITU applications, respectively. Based on these, the proposed antenna is operable from 1.4 GHz to 25 GHz with maximum coupling value of −20 dB covering almost operational band, and it has compact dimensions of $43\times 36\times 1.59$ mm3. Evolution and accomplishment of final results, involving the reduction of mutual coupling suppression, are illustrated gradually. The property of the monopole antenna considering reflection coefficient, isolation among different ports, realized gain, radiation pattern, efficiency, and envelope correlation coefficient (ECC) are studied. Moreover, experimental results are verified with the simulations. In addition, using the antenna inserted in chest badge, the chance of infection will greatly reduce. Ultimate, the proposed MIMO monopole antenna has a potential application in the medical, high precision machining, and biopharming relevant domains.

Truck-lifting accidents are common in container-lifting operations. The most common truck-lifting prevention system is based on lidar scanners which are very expensive, and it is more promising to use cameras to detect truck-lifting accidents than to use lidar scanners. However, little work has been conducted on visual detection of truck-lifting accidents, and the only previous visual approach fails in the common scenario where wheel hubs in images appear as non-standard or incomplete circles. To this end, this paper proposes a novel vision-based truck-lifting accident detection method for truck-lifting prevention system, which is free from the disturbance of distorted or incomplete wheel hubs in the image. The main idea of the proposed method is to utilize a deep learning-based object detection model to detect the truck body within which to extract many key-points whose vertical displacements are tracked to determine whether the truck is lifted. Based on this idea, the workflow for truck-lifting accident detection is delicately constructed. In addition, a YOLOv5-based modified detection model is proposed to reduce the computation cost of container and truck body detection, achieving 38.5% increase in inference speed on a single industrial personal computer without performance decrease. The experimental results demonstrate that the proposed truck-lifting accident detection method is capable of accurately recognizing the truck-lifting operations with the recall rate of 100% and the false alarm rate of 0.42%.

Rotated object detection in remote sensing images presents a highly challenging task due to the extensive fields of view and complex backgrounds. While Convolutional Neural Networks (CNNs) and Transformer networks have made progress in this area, there is still a lack of research on extracting and fusing features for small targets in complex backgrounds. To address this gap, we have extended the RTMDet framework by introducing three modules: the Focused Feature Context Aggregation Module, the Feature Context Information Enhancement Module, and the Multi-scale Feature Fusion Module. In the Focused Feature Context Aggregation Module, we replaced the Spatial Pyramid Pooling Bottleneck (SPPFBottleneck) to better extract small target features by focusing on contextual information. The Feature Context Information Enhancement Module enhances the model’s perception of multi-dimensional temporal and spatial information. Finally, we combined the original features with the fused ones to prevent the loss of specific features during the fusion process. Our proposed model, named the Multi-scale Feature Context Aggregation Network (MFCANet), was evaluated on four challenging remote sensing datasets (MAR20, SRSDD, HRSC, and DIOR-R). The experimental results demonstrate that our method outperforms baseline models, achieving improvements of 2.13%, 10.28%, 1.46%, and 1.13% in mAP for the MAR20, SRSDD, HRSC, and DIOR-R datasets, respectively.

Leakage current is one of the important parameters reflecting the operation status of distribution network-type surge arresters. At this stage, the polymagnetic current sensor has the advantages of miniaturization and high accuracy for leakage current measurement, but the complexity of electromagnetic interference in the field easily introduces more noise interference signals, which limits the performance of the polymagnetic current sensor for field application. To this end, this paper proposes a leakage current measurement method for distribution network-type surge arresters based on EEMD-SVD and low-rank RBF neural network methods. Firstly, the measured leakage current is decomposed by Ensemble Empirical Mode Decomposition (EEMD) to decompose the modal components containing eigenfrequencies, and then Singular Value Decomposition (SVD) is used to extract the non-eigenfrequency signals. Finally, a low-rank Radical Basis Function (RBF) neural network is used to approximate the leakage current signal after de-interference and combined with the Gaussian window function to remove the white noise interference. The experimental results show that the signal-to-noise ratio of the polymagnetic current sensor is improved by about 20dB and the maximum average absolute error is only 2.62%, which can truly reflect the leakage current of the network-type lightning arrester.

T-NNPC inverter is an attractive topology with fewer switches and good withstand voltage for medium voltage applications. In order to effectively achieve five-level output, the conventional method is an MPC strategy that can quickly track current, ensuring good output performance under fundamental frequency conditions. However, the flying capacitor voltage fluctuates, especially under low-frequency conditions. In addition, the CO-PWM is a potential method for this topology due to the broader voltage level distribution in the vertical direction. Nevertheless, the flying capacitor voltage balancing cannot be guaranteed under the high modulation indexes. To improve voltage balance under various modulation indexes and current frequencies, this paper proposes an improved CO-PWM (ICO-PWM) method whose voltage level distribution is modified to better apply in the T-NNPC topology. Compared to the conventional MPC method, the flying capacitor voltage is more stable, and the fluctuation is reduced by more than five times, especially under low-frequency conditions. The effectiveness of the proposed ICO-PWM method is verified by the simulations and experiments.

The change in laser output power caused by the temperature change of the pulse laser fuze drive circuit directly affects the detection distance and accuracy. In order to achieve accurate distance determination of the pulse laser fuze, it is necessary to study the influence of temperature changes on the output characteristics of the pulse laser fuze drive circuit. In this paper, we study the changes in peak power caused by changes in component parameters such as resistors, capacitors, and laser diodes in the pulse laser drive circuit, which is caused by typical temperature gradients in the transient state. Based on Simulink software, a simulation model of pulsed laser drive circuit is built, and based on the theory of RLC loop equation and laser rate equation, the temperature-dependent gain coefficients are introduced to obtain the laser output characteristics corresponding to typical temperature. As a main result, we found that the laser output power decreases by about 10% during the temperature change from -50°C to 50°C. A hardware test platform is built to verify that the error is no more than 1%. The above study provides a reliable method for the design of the temperature model of the pulsed laser fuze drive circuit.

Event detection is an important task for information extraction, which seeks to identify instances of specific event types from pieces of text. Recent studies have suggested that incorporating syntactic dependency graphs as feature representations for graph neural networks can significantly boost event detection performance. However, there are still challenges in leveraging multi-hop relationships within dependency parse trees to provide valuable additional information for keywords, as well as in effectively extracting relevant information from subordinate clauses, such as restrictive clauses. In this paper, we propose a novel Graph Convolutional Networks With Syntactic and Semantic (GCNWSS) structures for event detection task. Specifically, we construct a multi-hop matrix as the syntactic structure that calculates the hop distance between each word-pair. Besides, we propose a combination of biaffine attention and trigger-aware attention to generate semantic structures. In which, The biaffine attention mechanism is used to capture the global semantic information in a sentence. The trigger-aware attention mechanism enables the learning of trigger-related local semantics features of the text. Experimental results on benchmark dataset illustrate that our proposed model outperforms state-of-the-art methods.

In this paper, cascaded controllers, which combine adaptive sliding mode control and recursive techniques, are presented to control quadrotor unmanned aerial vehicles with time-varying load and unknown disturbances. In the outer-loop control subsystem, an adaptive sliding mode controller, which includes adaptive laws estimating the time-varying load and unknown disturbances, is presented to control the position of the quadrotor. In the inner-loop control subsystem, an adaptive recursive sliding mode controller, which includes a self-turning law estimating the unknown inner-loop disturbances, is designed to control the attitude of the quadrotor. Moreover, a nonsingular terminal sliding mode surface is used to converge the tracking error of attitude angles to zero in finite time; and an integral sliding mode surface is designed in the adaptive recursive sliding mode controller to improve the attitude control performance of the quadrotor. Simultaneously, adaptive gain adjustment laws were applied to estimate those unknown upper bounds of the disturbances in the system. In the end, the efficiency and feasibility of the proposed control method are demonstrated by some computer simulations.

With the continuous development of the automobile industry, fast and efficient body styling methods have become crucial. Although traditional hand-drawn sketches and 3D modeling are feasible, they often consume lots of time, manpower and financial resources when faced with complex design requirements. Therefore, a quick method for car body modeling has become a hot topic in automotive design. This paper proposes a three-dimensional automotive wireframe model generation and surface reconstruction method based on a single image in any view. Firstly, 3D wireframe model generation for a single car image in any view. By the MobileNet V2 network trained based on the established 2D car body library USMV3085, the visual angle of a specific single car image in any view can be identified, following the relevant wireframe model can be generated with the weight coefficients computed by constrained least squares method according to the key points and the corresponding rotated key points of the 3D wireframe library. Secondly, the corresponding 3D surface is reconstructed automatically. With the skeleton line firstly computed by the Skeleton Generation Algorithm for every 2D projected split part of the 3D wireframe model, corresponding 3D boundary relation pairs and control polygons are obtained, following the 3D surface is reconstructed integrating all split parts. Compared to traditional computer-aided shape design, this automatic method greatly shortens the research and development cycle and has certain reference significance in car body design.

With the increasing challenges posed by line-commutated converter-based high-voltage direct current (LCC-HVDC) in the era of widespread renewable energy utilization, AC/DC dynamic interaction problems under controller regulations appeared more often. However traditional approaches often fall short in identifying the features of signals involved in AC/DC conversion in grid dynamics, and the characteristics of the conversion process in the LCC main circuit are overlooked. Consequently, this paper aims to fill this gap to help reveal the mechanism of grid dynamic problems dominated by controller regulations. Based on the recognition that the amplitude/frequency of AC electrical variables has time-varying characteristics in practice during grid dynamics, this paper focuses on the conversion mechanism between AC and DC electrical variables and then the basic features of electrical variables in LCC main circuits during grid dynamics are expressed using time-varying amplitude/frequency rotating vectors. First, the six-pulse firing pattern is expressed using the rotating vector with time-varying frequency. Then the time-varying amplitude/frequency rotating vectors are utilized to discuss how AC/DC voltage and current converse, including the power produced from them during grid dynamics. Moreover, simulations are performed to validate the efficacy of the mechanism, alongside elucidating its potential applications to benefit grid dynamic analysis. Finally, the salient contributions of this proposed framework are clarified.

Pose change and environmental interaction have been the primary directions of recent research in the field of physics-based characters. To address the problems of a high production threshold, high real-time arithmetic consumption when combined with deep reinforcement learning, and a lack of realism in existing solutions, in this paper, a real-time automatic motion control model is designed for virtual characters that provides posture control through physical response linkages and that can generate smooth and natural interactions with the environment through small amounts of body adjustments. The proposed model consists of a mixture of kinematic controllers and lightweight physical interaction modules. It uses a designed posture correction module to correct the posture of a hybrid model, automatically adjusting the body posture, and it can respond reasonably to the environment when the action generation encounters environmental obstacles. The designed constraint scheme can reduce the tuning of joint parameters while allowing a character to have realistic human-like motion perception. The proposed model is verified by experiments, and the experimental results demonstrate that the proposed virtual human model can be adjusted to achieve the human-like effect using real action data and keyframe animation. Ball bouncing and walking on irregular terrain experiments verify that the proposed virtual character model can interact with the environment effectively in real-time. As the carrier of character animation, the proposed driving framework model can rapidly and easily generate interactive motions based on given input parameters.

The task of detecting surface defects on substation equipment faces several challenges, including a variety of target categories, the scarcity of original defect image data, complex environmental conditions, low accuracy in existing algorithms, as well as notable issues with false alarms and missed detections. Overcoming these obstacles is crucial for the successful implementation of intelligent inspection systems for substations. To address the problem of limited original data, we first employ the method of ADD-GAN to augment the image training set. Furthermore, this paper proposes a target detection model called YOLO-SD to detect various equipment defects in complex real-world scenarios. In order to enhance the network’s feature extraction capabilities in the presence of complex backgrounds and to improve detection accuracy, a novel deep perceptual feature extraction module named C3+ was introduced in this research. Furthermore, we incorporates SimAM into the neck network of YOLO-SD. This integration not only bolsters the network’s learning capacity but also equips it with the capability to autonomously learn and dynamically fine-tune attention weights to suit different input scenarios. To tackle the challenges posed by variations in size and shapes of different substation equipment defects in images, a novel fusion loss function NWD-CIoU is designed. The improvements enhance the accuracy and robustness of YOLO-SD in defect target detection across different scales. The experiment demonstrated that the YOLO-SD model achieved an mAP@0.5 of 90.3% and mAP@0.5:0.95 of 63.9% in detecting defects in substation equipment. The F1 score reached 81.1%, IoU value was 90.5%. This model realized accurate detection of multi-scale substation defect targets, reaching the state-of-the-art level in substation defects detection.

A strong and reliable Traffic Sign Detection and Recognition (TSDR) system is essential for the effective deployment of autonomous driving technology. In this field, numerous scholars have conducted extensive research, but current studies only consider TSDR under ideal conditions, neglecting scenarios such as rain, snow, fog, which can cause image blurring. This paper investigates the challenges of TSDR performance degradation caused by five adverse environmental conditions: rain, snow, fog, lens dirt, and lens blur. To overcome the adverse effects of these conditions on TSDR, this paper proposes a Convolutional Neural Network (CNN)-based TSDR method, consisting of three modules: adverse environment classification module, image enhancement module, and traffic sign detection module. The adverse environment classifier, based on the VGG19 architecture, identifies whether the image includes the aforementioned five adverse weather conditions. The image enhancement module named Enhance-Net enhances each of the five adverse environments separately and specifically enhances the traffic sign regions within the image, rather than the entire image area. To increase the speed of the proposed method, the traffic sign detection module utilizes the YOLOv4 framework. The proposed method’s effectiveness is assessed using the CURE-TSD dataset, which includes traffic videos recorded in various adverse environmental conditions. Experimental results demonstrate that under five different levels of adverse environments, the pro-posed method achieves 95.03% accuracy and runs at a rate of 12.79 fps (frames per second). In contrast to the current benchmark, although there is a 2.81% reduction in accuracy resulting from training the proposed method on a subset of the dataset, the speed has increased by 12.03 fps, demonstrating the efficacy of the proposed approach.

Road abandoned objects are potential safety hazards in modern traffic transport, especially in highway scenes. Promptly detecting such obstacles on the road is of great significance for driving safety and Intelligent Transportation Systems (ITS). Current research primarily focuses on developing diverse road anomaly detection approaches to discriminate the unknown objects regarded as abandoned ones. However, previous efforts have been largely inadequate due to the absence of abundant datasets. In addition, prevailing benchmarks mainly provide data pertinent to autonomous driving, which might not effectively generalize to highway scenarios owing to camera perspective and scope limitations. To address these challenges, we introduce a large-scale Road Abandoned Object Detection (RAOD) benchmark derived from video surveillance. First, we collect abundant real-world video clips containing various potential abandoned object categories on the road from our commercial ITS, then assemble a road abandoned object dataset comprising 557 video sequences and 18,953 images with pixel-level manual annotations. Second, we conduct exhaustive evaluation experiments employing a range of baseline models from mainstream algorithms on our dataset to illustrate the performance of different approaches. Third, we propose a novel image segmentation framework based on an area-aware attention mechanism. Experimental results reveal that our method outperforms the UNet-based model by nearly 9% in terms of dice score. Our dataset represents the most extensive open-source resource dedicated to road abandoned object detection, accessible publicly at https://github.com/yajunbaby/A-Benchmark-for-Road-Abandoned-Object-Detection-from-Video-Surveillance.

This paper presents an X-in-Loop test platform designed for the Proof-of-Concept testing of intelligent braking ECPBS in commercial vehicle. Firstly, the concept of intelligent braking is introduced, outlining its research scope and defining the necessary test requirements for the proof of concept. Based on these experimental criteria, the composition and operational principles of the Proof-of-Concept experimental platform are developed in accordance with X-in-Loop testing methodologies. The various subsystems and functionalities of the X-in-Loop Proof-of-Concept test platform are subsequently detailed. The working principle of the X-in-Loop Proof-of-Concept test platform is thoroughly explained using a loop example, and its workflow is presented. Finally, the platform’s fundamental functionalities are validated through a vehicle braking test conducted under specific working conditions, followed by an analysis of its potential applications.

The aging population in China is increasing the prevalence of degenerative diseases such as knee osteoarthritis (KOA), which significantly impacts the elderly’s quality of life. Traditional Chinese Medicine (TCM) massage has been effective in alleviating KOA symptoms; however, the physician-patient ratio and the physical demands on practitioners pose challenges. This study introduces the KOA massage robot, designed to replicate the seated knee adjustment manipulation, a specific TCM technique. The robot’s structural design and a Particle Swarm Optimization-Back Propagation (PSO-BP) algorithm are integrated to reduce the manipulation required by physicians and to assist in KOA massage treatment. The robot’s force prediction accuracy was determined to be 5.34N on average, and its therapeutic efficacy was supported by Surface Electromyography (sEMG) and Visual Analog Scale (VAS) assessments, demonstrating pain relief and improved quadriceps muscle activation in KOA patients. The experimental validation involved a comparison between traditional manual massage and the robotic intervention. The results showed that the robot could achieve an average increase in integrated EMG (iEMG) of 47%, closely mirroring the 45.37% increase observed in the manual treatment group. Similarly on VAS scores, the robotic intervention group obtained a decrease of 28.12%. This study’s findings not only highlight the potential of integrating TCM principles with modern robotics but also pave the way for a new paradigm in elderly care, where personalized and efficient KOA management is within reach.

The extraction of lake water bodies from remote sensing images is vital for water resource management, environmental protection, climate change research, disaster monitoring, and land use planning. This study presents a novel model, AMFF-LWBENet (A Lake Water Body Extraction Network based on ASPP and Multi-scale Feature Fusion), designed to tackle challenges such as insufficient spatial detail, poor edge recognition, and low anti-noise performance found in existing lake water body extraction models. The model utilizes an encoder-decoder architecture, leveraging ResNet50 for downsampling to extract deep features, which are then fused more effectively through the multi-scale dense fusion (MDF) module, incorporating depthwise separable convolution and ASPP to enhance the integration of multi-scale features. Additionally, position and channel correlations in the feature map are captured using DANet, minimizing noise interference at the lake edge and improving segmentation accuracy. Bilinear interpolation is employed to upsample the feature map, and feature fusion is achieved through the cross-layer feature fusion (CFF) module, enabling precise lake water body extraction. The datasets utilized include a self-built W-H dataset, derived from Landsat images collected between 2015 and 2023, as well as the publicly available TP dataset. The AMFF-LWBENet achieved MIoU values of 97.52% on the W-H dataset and 97.5% on the TP dataset, surpassing current state-of-the-art semantic segmentation networks. These results suggest that AMFF-LWBENet is even more effective than many other networks in extracting lake water bodies, offering significantly enhanced support for a wide range of environmental and urban planning applications.

Manual laborers are often at an increased risk for work-related musculoskeletal disorders (WMSDs) due to improper work postures or the lifting of excessively heavy objects. Therefore, conducting an effective ergonomic assessment of workers is crucial for enhancing productivity and minimizing the occurrence of WMSDs. The Ergonomic Posture Risk Assessment (EPRA) is a widely used method for this assessment. However, traditional EPRA methods rely on human or sensor input, leading to subjective bias, instability, and reduced accuracy. This study addresses these issues by proposing an objective machine-learning approach. It employs a non-intrusive computer vision technique for posture capture, enabling rapid analysis of the worker’s activities through Random Forest analysis. The dataset for risk assessment is generated from the worker’s skeletal joint posture data, collected using Movenet Thunder in conjunction with an inertial motion capture device and the Rapid Entire Body Assessment (REBA) standard. The Particle Swarm Optimization Random Forest (PSO-RF) model is then utilized to predict risk scores for various postures, incorporating limb length ratios to tackle challenges associated with observing torsional joints. The model’s effectiveness in detecting poor posture is subsequently evaluated. The findings indicate that the PSO-RF model successfully identifies poor postures and computes the corresponding REBA scores with 89% accuracy, 93% precision, 91% F1 score, 89% recall, and a kappa value of 82%. This research demonstrates that the machine learning approach, utilizing computer vision and Random Forest, can effectively conduct EPRA to prevent musculoskeletal injuries, providing a data-driven and accurate method for enhancing workplace safety and health.

Nowadays, the electromagnetic environment is becoming more and more complex, and it is increasingly difficult to accurately identify electromagnetic signals. At the same time, the quality of models and algorithms largely determines the final result of electromagnetic signal recognition, and a comprehensive and holistic evaluation system is needed to assess the quality of model or algorithm performance in the field of electromagnetic signal recognition. To address this phenomenon, this paper uses different machine learning models and AdaBoost algorithms to identify 100 classes of WiFi radiation source electromagnetic signals. In this paper, the classification performance, complexity and robustness of the models are evaluated comprehensively in three aspects respectively, and a multi-dimensional evaluation system for signal recognition is constructed, and the differences between different model algorithms in signal recognition effects are compared and analyzed, and the correlations between each evaluation index in different dimensions are explored. The experimental results show that different machine learning models and algorithms have different recognition effects on electromagnetic signals, among which the recognition accuracy of ResNet can reach more than 90%, but its computational complexity is high and easily affected by noise. DNN has poor recognition effect and the highest computational complexity, but it is not easily affected by noise. The AdaBoost algorithm does not necessarily improve the recognition classification accuracy of the underlying classifier. The evaluation system established in this paper is meaningful for assessing the performance of models and algorithms.

Search and rescue (SAR) is an important part of joint operations and a key support for combat effectiveness. Because of the complex composition of the SAR system of systems (SoS), sensitivity analysis method is usually used to carry out sensitivity analysis of SoS capability, so as to determine the main design indicators affecting SoS capability. Reliable sensitivity analysis results are often based on the analysis for sufficient data. However, the SAR SoS capability is affected by many factors and there are numerous design indicators. Even if a small number of design points are selected for each design indicator, tens of thousands of test schemes will be formed, and carrying out all simulation tests will bring huge workload and time cost. To solve this contradiction, this paper introduces a bidirectional deep neural network (BDNN), and takes advantage of its better self-learning and adaptive features and unique structure to train the existing test data, Through strong feature extraction ability of BDNN, the network model between the design indicator and capability indicator is formed, namely, the capability fitting and data reconstruction (CFDR) model, so that the implicit relationship between the two is fixed into the model. In the training process, the number of hidden layers and neurons in each hidden layer, and the amount of training data are explored according to the training effect, so as to obtain a better parameter combination. In order to avoid introducing large cumulative errors accumulated during BDNN pre-training into DNN, particle swarm optimization (PSO) was introduced to optimize weight parameters and avoid large training errors being transmitted to deep neural network (DNN). Meanwhile, three basic functions were used to verify the strong global optimization and convergence abilities of the BDNN after optimized by the PSO (PSO-BDNN). Finally, the new test scheme is applied to the CFDR model to obtain the SoS capability value. The reconstructed data obtained from the CFDR model based on BDNN and PSO-BDNN respectively were compared with the simulation test data.The results show that the reconstruction accuracy of the CFDR model based on the PSO-BDNN is greatly improved than that of the BDNN. And the feasibility of this model as a reconstruction data generation model and the effectiveness of this model as an analysis data extension method applied to the sensitivity analysis of insufficient data to obtain reliable analysis results are verified.

Noise, vibration and harshness are a big issue for hybrid vehicles especially in range extenders. Traditional vibration control and active damping control are generally used for suppressing vibration. Compared with traditional vibration control, active damping control is more flexible and effective in suppressing vibration. The basic idea of active damping control is controlling electric motor compensating for the torque of engine and reducing the vibration from the source. The reduction of vibration is highly depended on the compensation torque of the electric motor. A differently designed compensation torque has a significant impact on the active damping control effect. In this paper, an active damping control method for online application is proposed. And several kinds of compensation torque are designed for this method. Experiments on vibration performance are conducted to compare the effect of different compensation torque design in reducing vibration. And the results show that the compensation torque in the shape of square wave has potential for reducing vibration in all directions. Finally, a guiding tool for compensation design is proposed by correlation analysis and the relationship between designing factors and vibration is revealed.

The Brillouin optical time-domain reflectometer (BOTDR) system, which is based on the Brillouin frequency shift (BFS) of the scattering spectrum, has been widely applied to monitor the temperature and strain in many fields. The measurement accuracy of the BFS directly affects the monitoring and location of the temperature change or strain events. In this study, the characteristics of the least-squares fitting optimal solution of the Brillouin power spectrum are theoretically investigated, and a half-peak fitting algorithm is proposed to measure the BFS with high accuracy and stability. In particular, it can precisely determine the double-peak spectrum and detect the position in the transition Section of a temperature change or strain event. Furthermore, it can reduce calculation complexity and enhance measurement speed by dropping most of the data. For the fiber under test (FUT) with a length of 12 km, 200 groups of time-domain data were processed using a probe pulse of width 50 ns and sampling rate 1 GHz. The half-peak fitting algorithm increased the temperature measurement accuracy by $\sim $ 1 fold, with $\sim $ 20% calculation of full fitting. Meanwhile, it effectively eliminated the influence of minor peak in the double-peak spectrum, and optimized the spatial resolution of the temperature change position to 0.1 m, which is the maximum limitation by sampling rate.

With the rapid development of intelligent computing technologies, the smart agricultural systems has gradually received more research attention. Precisely controlling planting conditions via intelligent means, it is expected to technically improve reliability of smart agricultural systems. For this purpose, this paper introduces the idea of big data fusion, and proposes a reliable data-driven control method for planting temperature in smart agricultural systems. Firstly, a variety of sensors are used to collect the values of environmental factors such as air temperature and light intensity during the experiment, and the “Digi” wireless communication technology is used to transmit the data. Then, the collected data is sent to the server through the 5G module, and the important data of the growth process is stored in the MySQL database. Taking the cultivation of yams as the investigation object, experiments on the real-world data are utilized to evaluate the proposal. And it is shown that the propose framework can work well to control temperature for yam planting from the perspective of data fusion.

It is well known that significant computational power and a large amount of memory are required for deep neural networks, which makes them difficult to apply in resource-limited environments. So, many network compression and acceleration technologies have emerged, of which connection pruning is widely applied due to its effectiveness and convenience. A novel connection pruning method for full model capacity on multiple sparse structures is proposed in this paper. We design a simple and efficient function called Dynamic Processing Unit (DPU) for handling the evaluated weights. Our method has the following features: 1) Instead of being pruned directly or set to 0, the weights are controlled by the DPU to determine whether they will be used during subsequent forward passes of the network during the iteration of pruning training. 2) It supports the traditional multi-steps prune method as well as the end-to-end training mode that can get a compressed network in a single stage by fusing training and pruning. 3) It can learn multiple useful sparse structures, including, but not limited to, depth-wise, filter-wise, channel-wise, 2D-filter-wise, row-wise, column-wise, connection-wise and mixed sparse structures. Our method is tested on various widely-used datasets and models, such as the LeNet and the ResNet on MNIST and CIFAR-10. Importantly, it demonstrates good performance in all these cases. Some details about our method can be found at the following URL: https://github.com/hujie369/DPU

This work presents an in-situ readout of the transient photoresponse of a graphene-oxide-semiconductor heterostructure by utilizing graphene’s field-effect coupling with the silicon photogate. The reported device which acts as a graphene charge-coupled device (GCCD) pixel is set into pre-avalanche condition by dynamic sinusoidal biasing and then exposed with pulsed illumination. The initial photo-ionized charge packet experiences pre-integration carrier multiplication, boosting the signal-to-noise ratio (SNR) before interacting with the surface traps and defects. The maximum multiplication factor of $\mathrm {\sim }8.5$ and responsivity of $\mathrm {350 A/W}$ are achieved. Arrhenius plots show the viability of operating the device at room temperature as thermal charge contribution is negligible. Moreover, a detailed discussion on reduced power consumption for such a sinusoidal charge-coupled device (CCD) drive concept is also incorporated. The presented technique paves the way for futuristic sinusoidally-driven graphene-silicon-based electron-multiplying CCDs with low-power surveillance and adaptive optics applications.

Object detection for remote sensing images has problems such as complex backgrounds, multi-scale and difficulty in the detection of small targets. Because of the above problems, an improved object detection algorithm for remote-sensing images is proposed. Firstly, fusing the characteristics of shallow and deep networks, the original feature pyramid structure is reconstructed, the improved adaptive feature fusion structure is introduced before the network prediction, and the location and category information of different feature maps are fused. Secondly, the attention mechanism is introduced into the network to reduce the interference caused by complex backgrounds; Finally, the up-sampling module is improved to expand the receptive field and realize up-sampling based on semantic information. Through experiments, we show that our network achieves 93.9% and 98.5% detection accuracy on Levir and RSOD dataset, respectively, with the detection speed, reaching 98.26fps. Compared with the original network, the accuracy is increasing by 4% and 1.2%, respectively. Experimental results show that the detection accuracy of our algorithm is improved while maintaining the light weights and efficiencies of the original network.

Along with the development of underground space and the increasing maturity of automation technology, a new type of transport and supply system called “Underground Logistics System (ULS)” is gradually gaining attention in various countries as an effective means of solving the problem of urban traffic congestion. However, the high cost and risk of underground construction are the main factors that have led to ULS remaining in the theoretical stage for the time being. Therefore, the paper aims to design a solution for implementing a metro-based underground logistics system (M-ULS). Firstly, the paper uses the K-mean clustering algorithm and genetic algorithm for path planning of conventional ground transportation and M-ULS. Secondly, the article is a retrofitting design for a site that combines the metro with an underground logistics system and simulates the underground logistics nodes with the help of AnyLogic simulation software. Finally, the paper explores the economic and environmental value of M-ULS in terms of carbon emissions, time costs, fuel costs, and environmental pollution costs. The research provides a solution to alleviate urban traffic congestion, reduce environmental pollution and promote sustainable urban development.

The current stereo warehouse in the cold chain logistics industry faces problems such as high product loss rate, severe temperature deviations, and low level of informatization. To ensure the food quality and safety of cold chain logistics warehouses, and achieve cost reduction and efficiency improvement, this study focuses on the operational scenarios of stereo warehouses in cold chain logistics, and designs a five-dimensional digital twin model with full mapping and multidimensional simulation. By integrating heterogeneous data from multiple sources, it achieves real-time visualization and monitoring of all elements in warehouses, and proposes a model optimization strategy that supports system-linked decision-making. This strategy intelligently adjusts the load and thermal distribution in the temperature-controlled zones of warehouses to maximize operational efficiency and reduce operating costs. Using this approach, a digital twin system for cold chain logistics warehouses was constructed. By comparing the warehouse operation data before and after optimization, the loss rate of fresh vegetables under the same storage conditions was reduced by 25-30%. This system achieves precise mapping and coordinated operation of real-time data from cold chain logistics warehouses and dynamic virtual models. The practical application of this solution is demonstrated in a cold-chain warehouse of a supply chain company in Guangdong, forming an integrated control model for cold chain logistics warehouses. This model provides a reference for the application of digital twin technology in the operation, scheduling, decision-making, and collaborative operations of cold chain logistics warehouses and offers new possibilities for establishing improved packaging for perishable products and ensuring cold chain food quality and safety.

Micro-expressions are brief, involuntary facial movements that reveal genuine emotions. However, extracting and learning features from micro-expressions poses challenges due to their short duration and low intensity. To address this problem, we propose the ADMME (Action Decouple Multi-tasking for Micro-Expression Recognition) method. In our model, we adopt a pseudo-siamese network architecture and leverage contrastive learning to obtain a better representation of micro-expression motion features. During model training, we utilize focal loss to handle the class imbalance issue in micro-expression datasets. Additionally, we introduce an AU (Action Unit) detection task, which provides a new inductive bias for micro-expression detection, enhancing the model’s generalization and robustness. Through five-class classification experiments conducted on the CASMEII and SAMM datasets, we achieve accuracy rates of 86.34% and 81.28%, with F1 scores of 0.8635 and 0.8168, respectively. These results validate the effectiveness of our method in micro-expression recognition tasks. Furthermore, we validate the effectiveness of each component of our approach through a series of ablation experiments.

Chlorophyll-a (Chl-a) is an important parameter of water bodies, but due to the complexity of optics in water bodies, it is currently difficult to accurately predict Chl-a concentration in water bodies by traditional methods. In this paper, Sentinel-2 remote sensing images is used as the data source combined with measured data, and Ulansuhai Lake is taken as the study area. An adaptive ant colony exhaustive optimization (A-ACEO) algorithm is proposed for feature selection and combined with a novel intelligent algorithm of optimizing support vector regression (SVR) by genetic algorithm (GA) for prediction of Chl-a concentration. The ant colony optimization (ACO) algorithm is improved to select remote sensing feature bands for Chl-a concentration by introducing relevant optimization strategies. The GA-SVR model is built by optimizing SVR using GA with the selected feature bands as input, and comparing with the traditional SVR model. The simulation results show that under the same conditions, using A-ACEO algorithm to select feature bands as inputs can effectively reduce the model complexity, and improve the model prediction performance, which provides a valuable reference for monitoring Chl-a concentration in lakes.

This paper investigates the critical issue of maintaining system stability for multipacket reception (MPR) $p$ -persistent carrier sense multiple access (CSMA) systems. When multiple users with individual queues simultaneously transmit a packet to an access point (AP) via CSMA, the number of successful transmissions and their identities are determined probabilistically in the MPR channel. Stability signifies that none of the users’ queues grows unbounded. The stability region is a significant measure, addressing all potential permutations of mean packet arrival rates for the users to maintain bounded queue lengths. The work begins by considering a system with two users characterized by differing mean packet arrival rates and (re)transmission probabilities. Subsequently, it examines an $N$ -user system in which each user shares an identical packet arrival rate and retransmission probability. A backoff algorithm is then proposed for these $N$ users to utilize in order to stabilize their queues. The paper concludes with numerical studies illustrating the stability region as a function of user parameters, demonstrating how the proposed backoff algorithm can be used to maximize throughput.

High-resolution multi-spectral (HRMS) images have been widely used in various fields, however, they can not be directly obtained due to the physical hardware limits of existing remote sensing satellites. Therefore, the pansharpening technique has been widely explored as an effective tool to generate HRMS images by fusing the complementary information of low-resolution multi-spectral (LRMS) images and high-resolution panchromatic (PAN) images. Existing deep learning-based pansharpening methods mainly focus on enhancing the spatial representation ability of the network, while paying little attention to modeling spectral dependencies in spite of its significance for remote sensing data. In this paper, we propose a simple yet effective channel group shuffling (CGS) module to explore the implicit relationships with regard to the adjacent and cross-channels while considering spatial information. To be specific, the proposed CGS module consists of two components: the channel group module and the feature shuffle fusion module. The former enhances the diversity of spectral information and cross-channel information communications while ensuring the spectral order of the input feature. The latter integrates the cross-group feature maps with rich spatial-spectral information. Equipped with the proposed functional module, our image fusion network, dubbed CGSNet, produces favorable results against existing state-of-the-art counterparts over various satellite datasets. Ablation studies further verify the flexibility and effectiveness of our core design.

High-resolution networks have demonstrated significant advantages in multi-scale feature extraction for human pose estimation task. However, this often leads to the problems of large network parameter quantity and high computational complexity. On the other hand, some existing lightweight high-resolution networks have lower model parameter quantity and computational complexity, but they lack accuracy in handling keypoint position information. To address these issues, this study proposes a lightweight High-resolution network called DB-HRNet (Dual Branch High-Resolution Network) for human pose estimation. Specifically. based on HRNet, a more efficient network framework is designed as the backbone framework of DB-HRNet. Additionally, we propose a position-sensitive basic module called DBBlock (Dual Branch Block) to enhance the model’s ability to capture positional information accurately, as the basic block of DB-HRNet. Finally, we propose a Pose Refine Module to establish the connections between keypoints and make the final adjustments to the model’s output. To validate the effectiveness of DB-HRNet, we conduct experimental studies on the COCO2017 and MPII human pose estimation datasets, comparing DB-HRNet with several advanced existing lightweight human pose estimation models. The results demonstrate that the proposed network model achieves a 1.8 points improvement in detection accuracy on the COCO2017 dataset compared to the advanced lightweight high-resolution network Dite-HRNet, while also achieving a 38% increase in model inference speed.

Intelligent design is a trend in geotechnical engineering. However, a middle link between demand identification and intelligent design is missing in anchor engineering, resulting in lower efficiency in terms of the interaction between the anchor components, geological body and modeling information. To address this challenge, this paper proposes a method for modeling a three-dimensional anchor engineering information model based on sparse boreholes. This method includes three parts: modeling the geological body, integrated modeling of the anchor components and the geological body, and multiple information assignments based on the anchor model. The feasibility was validated in actual engineering scenarios by using neural networks and rendering engines. The results showed that the prediction accuracy of the strata improved by 17.69%, the modeling efficiency increased by double, and engineering information was fit to the model well. This paper has the application potential to achieve intelligent design in anchor engineering.

Faced with increasingly complex industrial data, standard machine learning algorithms struggle to effectively extract both linear and nonlinear features. In this study, an improved residual network (ResNet) called Residual network with Independent Multi-Channel Weighting (RMCW) to tackle the nonlinear, temporally uncertain, and unevenly distributed fault. Firstly, a strategy for constructing the multi-channel vibration intrinsic mode function (IMF) images is designed to obtain the primary features by combing the empirical mode decomposition (EMD) and the gramian angular field (GAF). Secondly, a dynamic receptive field (DRF) with independent channel weighting is utilized to adaptively fuse the multi-channel features. This renders both initialization parameters for each individual channel and DRF parameters mutually independently adaptive to the fault features in the different batch. Thirdly, the RMCW model is built by inputting the fused features to the network of 9 residual building blocks. Two experimental cases verify that the propose method is effective for the machinery fault diagnosis and is superior to the comparing methods.

The aerodynamic noise test platform for a certain type of unmanned aerial vehicle (UAV) imposes strict requirements on the drive motor, demanding high torque, small diameter, and low vibration. Compared to DC motors and induction motors, permanent magnet motors have advantages in terms of torque and volume, making them well-suited to meet the platform’s requirements. In response to the development needs of the test platform for the drive motor, this study systematically analyzes the key technologies that require special attention in the design process of the permanent magnet motor. Through proper electromagnetic design, structural design, cooling design, and vibration design, particularly by verifying the electromagnetic performance under rated load and weak magnetic conditions, a 180 kW permanent magnet motor was manufactured for experimental validation. The motor was successfully utilized for propeller aerodynamic noise testing. The experimental results demonstrate that the 180kW permanent magnet motor meets the technical requirements of the test platform, confirming the accuracy of the permanent magnet motor design methodology. It provides a valuable reference for the design of similar propeller drive motors and holds practical engineering significance.