Abstract: The study of deep neural networks has recently gained widespread attention in recent years, with many researchers proposing network structures that exhibit exceptional performance. A current trend in artificial intelligence (AI) technology involves using deep learning and its applications via large-scale pretrained deep neural network models. This approach aims to improve the generalization capability and task-specific performance of the model, particularly in areas such as computer vision and natural language processing. Despite their success, the deployment of high-performance deep neural network models on edge hardware platforms, such as household appliances and smartphones, remains challenging owing to the high complexity of the neural network architecture, substantial storage overhead, and computational costs. These factors hinder the availability of AI technologies to the public. Therefore, compressing and accelerating deep neural network models have become a critical issue in the promotion of their large-scale commercial applications. Owing to the growing support for low-precision computation technology provided by AI hardware manufacturers, model quantization has emerged as a promising approach for the compression and acceleration of machine learning models. By reducing the bit width of deep neural network model parameters and intermediate feature maps during the forward propagation of the model, memory usage, computation efficiency, and energy consumption can be substantially reduced, enabling the utilization of quantized deep neural network models in resource-limited edge devices. However, this approach involves a critical tradeoff between task performance and hardware deployment, which directly impacts its potential for practical application. Quantizing the model to a low-bit precision can lead to considerable information loss, often resulting in a catastrophic degradation of the task performance of the model. Thus, alleviating the challenges of model quantization while maintaining task performance has become a critical research topic in AI. Furthermore, because of the differences in hardware devices, constraints of application scenarios, and data accessibility, model quantization has become a multibranch problem, including data-dependent, data-free, mixed-precision, and extremely low-bit quantization, among others. By comprehensively investigating various quantization methods for deep neural networks proposed based on different perspectives, and summarizing their advantages and disadvantages thoroughly, the essential problems that are associated with the quantization of deep neural network quantization can be explored, which points out the directions for possible future developments.
Abstract: In nature, flying creatures flap their wings to generate lift, which is necessary for flight. Most birds change flight patterns by moving their wings using their wing muscles and adjusting their tail states. Insects, which are without tails, can achieve maneuverable flight using their chest and abdomen muscles and other structures such as hind wings. Owing to high mobility and high energy efficiency, researchers have developed various flapping-wing aerial vehicles according to the bionic principle to improve flight performance. However, a flapping-wing aerial vehicle is a nonlinear and time-variable system. The low Reynolds number and unsteady eddy are important characteristics of the flapping-wing aerial vehicle, and the values are different from those of traditional aircraft. The Reynolds number of the traditional aircraft is larger; thus, the air viscosity is small enough to be ignored. However, the air viscosity of the bionic flapping-wing aerial vehicle is high at low Reynolds number conditions. Adopting a conventional aerodynamic configuration will result in insufficient lift. In addition, the traditional aerodynamics theory cannot explain the high lift of the flapping-wing aerial vehicles, and the mature technologies in traditional aircraft design cannot be directly applied owing to the low Reynolds number. Owing to the periodic movement of the flapping wing, it is difficult for researchers to accurately analyze the aerodynamic model. The autonomous flight of a flapping-wing aerial vehicle is limited by several challenges. To solve this problem, researchers have studied the flight principle of birds and insects. Moreover, the attitude control, position control, and stability analysis of flapping-wing aerial vehicles have been studied. Several control strategies based on robust control, neural networks, and other methods have been proposed to realize the autonomous flight of flapping-wing aerial vehicles. Researchers have also adopted control methods such as adaptive controllers combined with linearization techniques to control attitude. Position control has been achieved using a hierarchical controller and other approaches. In addition, perturbation observation is used to deal with the uncertainty of the system to improve stability. In this paper, the flight control strategies of flapping-wing aerial vehicles of different scales are reviewed. The current research on the flight control of the flapping-wing aerial vehicle is mostly in the prototype phase. Most of these theories have not been verified in actual flight. Therefore, the flight control theory needs to be combined with actual missions to promote the application of the flapping-wing aerial vehicle. Finally, the future trend of the flight control of the flapping-wing aerial vehicle is highlighted.
Abstract: With the proliferation of the industrial internet, passive optical networks (PON) have attracted attention from industry and academia. Industrial PONs need to meet the requirements of ultra-high speed, ultra-large connections, and low cost to be able to cope with diverse industrial internet scenarios. Physical layer modulation formats, multiplexing modes, transmission schemes, and digital signal processing algorithms have always been the focus of research in this field. Direct detection and 10 G devices are extensively studied due to their low cost and ease of deployment. However, meeting the power budget targets of a high-speed PON can be challenging. Transmission of a high-order modulation signal can improve the system’s capacity; however, it may also decrease the receiver’s sensitivity. Polarization multiplexing technology, coherent detection, and advanced digital signal processing technology can be utilized to meet the demands of industrial PON, such as ultra-high speeds, ultra-large connections, ultra-high capacities, high receiver sensitivity, and high power budgets. The use of intensity modulation schemes at the transmitter would reduce the overall cost of the system. Herein, we propose a digital coherent detection system with polarization multiplexing intensity modulation suitable for industrial PON transmission. The 50 Gbit·s?1 signals modulated with four-level pulse amplitude modulation (PAM 4) and on–off keying (OOK) are transmitted on the simulation platform. Performance assessments of back to back (BTB) transmission and fiber transmission over 20 km are studied to prove the superiority of this proposed scheme. A power budget of more than 32 and 36 dB can be achieved with the polarization multiplexing coherent detection system-based transmission of 50 Gbit·s?1 PAM 4 and NRZ signals, respectively, on a 20 km C-band single-mode fiber at a threshold of 3.8×10?3, which is a considerable improvement over intensity modulation direct detection systems. The superiority of the proposed scheme is further verified by the fact that the 100 Gbit·s?1 PAM 4 signal has better transmission performance in the polarization multiplexing intensity modulation coherent detection system than the NRZ signal. Considering the system performance, the polarization multiplexing intensity modulation coherent detection scheme is expected to be one of the promising solutions for high-speed PON systems. Finally, the development trends of the physical layer of industrial PON are discussed.
Abstract: Decrease in the cost of acquiring 3D point cloud data coupled with the rapid advancements in GPU computing power have resulted in an increased demand for 3D point cloud semantic segmentation in numerous 3D visual applications, including but not limited to autonomous driving, industrial control, and MR/XR, which further advances the development of deep learning methods in 3D point cloud semantic segmentation. Recently, many novel deep learning network architectures, such as RandLA-Net and Point Transformer, have been proposed and have achieved notable improvements in semantic segmentation accuracy while decreasing the computational load. However, previous research on 3D point cloud semantic segmentation methods has focused primarily on relatively early works, whose approaches have been gradually abandoned over the years and cannot accurately reflect the current research status. Moreover, the existing methods have been categorized based on their input data types, making it difficult to compare the segmentation performance of different techniques and not providing a comprehensive view of the relationship between methods using different network architectures. Therefore, this paper reviews the mainstream 3D semantic segmentation methods developed in the last three years using different deep learning network architectures and is organized into three levels. First, the two principal 3D point cloud data acquisition methods, including their customary datasets and metrics to evaluate model performance, are introduced. Second, a systematic review of 3D semantic segmentation methods based on different network architectures is organized, followed by a statistical analysis of the evaluation of performance between different models on two 3D segmentation datasets—S3DIS and ScanNet. The analysis of model performance on these two commonly used datasets includes model structure relevance, strengths, and limitations. Finally, an insightful discussion of the remaining methodological and application challenges and potential research directions is provided. This paper offers an extensive overview of the recent three-year research progress in 3D point cloud semantic segmentation and summarizes various network architecture pipelines, elucidates their fundamental operations, compares the model performance across multiple architectures, discusses their notable strengths and limitations, most importantly, concludes the current challenges and promising research directions for future investigations. Furthermore, this paper enables researchers to effortlessly identify the relevant research and research hotspots among different 3D point cloud semantic segmentation methods based on the analyses presented and aims to update the reviews on 3D point cloud semantic segmentation methods with a better viewpoint and highlight key properties and contributions of proposed methods, providing promising research directions for the main challenges.
Abstract: Intellectualization and unmanned manufacturing have been an inevitable trend in industrial development. The landing of intelligent applications is one of the current challenges in the industry. Due to the hierarchical architecture of the industrial automation pyramid, traditional programmable logic controllers (PLCs) that are usually employed in the field cannot cooperate with artificial intelligence (AI) algorithms that require massive data and computing resources. Therefore, it is necessary to research the virtualization of traditional PLCs as dockers, which can be deployed in the cloud, edge, or field. Cloud PLCs can be easily integrated with AI, big data, and cloud computing to achieve intelligent decision-making and control and break down data islands. The visual sorting system has attracted increasing attention for its ability to accurately detect the position of objects. Many deep learning–based methods have achieved remarkable performance in computer vision. Additionally, the requirement of a network is fundamental for guaranteeing data transmission with low latency and high reliability. The combination of 5G and time-sensitive networking (TSN) can achieve the deterministic transmission of several industrial applications. According to the above challenges, joint control between cloud PLCs of low-level devices and visual sorting systems in a reliable network is critical and has industry potential. In this study, we propose a deep learning–based material recognition and location system with a cloud PLC, which is demonstrated in a 5G-TSN network. First, traditional PLC is virtualized to realize flexible PLC function deployment in the field and cloud. Second, we establish a cloud-based AI platform and design a You only look once v5 (YOLOv5)-based object detection algorithm to locate the position and recognize the types of materials to obtain pixel coordinates. Third, the camera calibration method is used to transform pixel and world coordinates, and the material information consists of the world coordinates, types, and timestamps that are sent to cloud PLC. Finally, the commands are transmitted by the 5G-TSN environment from cloud PLC to the low-level devices for real-time control of the multi-crane cooperative. We establish an experimental system to demonstrate the significance and effectiveness of the proposed scheme, which synergistically controls multi-crane operation. The mean average precision (mAP) of material location is up to 99.65%, sorting accuracy reaches 96.67%, and the average consuming time is 25.99 s, which meets the requirements of low latency and high precision in industrial applications.
Abstract: The vehicle-to-everything (V2X) network has the potential to revolutionize the way we interact with vehicles and the surrounding. By utilizing innovative information and communication technologies, V2X networks can connect human beings, vehicles, roadside units, and even the cloud. In the near future, beyond 5G (B5G) and 6G technologies will enable the next-generation V2X networks to achieve superior communication and sensing capabilities, which is expected to offer advanced technologies such as intelligent driving and transportation. However, the strong Doppler effects arising from the high mobility of vehicles may lead to significant inter-carrier interference and pilot overheads in the existing orthogonal frequency division multiplexing (OFDM) systems, particularly as the millimeter wave and terahertz technologies dominate the B5G/6G era. In recent years, orthogonal time frequency space (OTFS) techniques have attracted attention owing to their ability to resist doubly-selective fading. In addition, the integrated sensing and communication (ISAC) based on OTFS (OTFS-ISAC) has emerged as a promising approach for V2X networks. In this context, our objective is to investigate the system structure, application and challenges of OTFS-ISAC in V2X networks, along with the related key techniques such as frame structure, pilot design and signal processing. First, we will explore the structures and fundamental theories of OTFS-ISAC systems, followed by the evaluation of communication and sensing performance. In particular, we will investigate the system architecture of OTFS-ISAC in monostatic and bistatic radar modes, respectively. Secondly, we will provide an overview of the state-of-the-art of OTFS techniques and further discuss the challenges and key techniques of OTFS-ISAC concerning the frame structure in the physical layer, pilot mechanism design, communication and radar signal analyses, etc. Finally, we will examine the case studies of OTFS-ISAC utilization in V2X networks to address corresponding major issues such as the inadequacy of Doppler resolution, low overhead beam scanning and target detection, and cooperative resource management. The ISAC system is in developmental stages, and this is the first comprehensive review that investigates the OTFS-ISAC system in detail. Although OTFS-ISAC offers significant advantages over OFDM-enabled ISAC in V2X characterized by high mobility, it faces numerous challenges in practical applications, including the well-known fractional Doppler effect and high peak-to-average ratio. However, with continuous development and technological advancements, it is anticipated that the OTFS-ISAC system will gain wide acceptance.
Abstract: Polarimetric SAR ship detection is an important application of the polarimetric SAR system. Existing polarimetric SAR ship detection methods are plagued by erroneous detection of strong clutter and missed detection of small targets in multiscale situations. Particularly, the existing methods easily detect strong clutter as the target under strong background clutter, resulting in false alarms; in the case of multiscale ship detection, small ships are easily submerged in background clutter, resulting in missed detection of small targets. To solve these problems, this paper proposes a polarimetric SAR ship detection method based on superpixels and sparse reconstruction saliency. This method has two stages. In the first stage, the large polarimetric SAR ship detection scene image is segmented using the superpixel segmentation method to obtain a superpixel image. With the superpixel as the basic unit, a saliency detection method based on sparse reconstruction is used to obtain the saliency value of each superpixel in the image. Then, the superpixels that may contain ship targets are retained using the sea surface ship density defined in this paper. Accordingly, in the first stage, the superpixel regions that may contain ship targets are obtained through superpixel segmentation and sparse reconstruction saliency detection. Next, in the second stage, a saliency detection method based on sparse reconstruction is used to obtain the saliency value of each pixel in these reserved superpixel regions. Finally, the global threshold segmentation method is used for the pixels in these regions to obtain the final detection results of ship targets. In this paper, two polarimetric SAR images of the ALOS-2 satellite with different scenes were selected for an experiment. One image contains strong clutter on the sea surface; the other contains ships of different sizes and many small ships. The experimental results show that the proposed method can well determine the superpixel regions that may contain ship targets in the first stage and successfully obtain the ship detection results in the second stage. In addition, in both scenarios, the classic constant false alarm rate (CFAR) methods and a saliency detection method are used for comparison with the proposed method. The experimental results show that the proposed method produces almost no false alarms because it is insensitive to strong clutter; moreover, this method rarely misses small ship targets in the multiscale ship detection scene. The figure of merit of the proposed method reaches 94.87% in the strong clutter scene and 94.05% in the multiscale ship detection scene.
Abstract: Deep rock mass is in a complex mechanical environment characterized by high ground stress, high ground temperature, high karst water pressure, and strong mining disturbance, resulting in difficult support and high levels of failure in the pump chamber group. To solve the problem of the instability of the deep pump chamber group, this paper takes the ?890-level pump absorbing well chamber group of the Daqiang coal mine as the engineering background. Through theoretical analysis, numerical simulation, and field tests, the reasons for the failure of the chamber group are analyzed, and the effects of intensive design and traditional design on the stability control of the surrounding rock are compared. According to the characteristics of high constant resistance, high elongation, and energy absorption of the negative Poisson’s ratio (NPR) cable, the instability energy criterion of intersection under the NPR cable support is established, where the chamber is stable at KN ≤ 1. The intensive control strategy of the pump absorbing well with a high prestressed NPR cable and three-dimensional truss as the core is presented and applied in the field. The results show that high ground stress, low surrounding rock strength, dense chamber group distribution, unreasonable excavation sequence of the chamber group, and inappropriate support are the main reasons for the failure of the deep pump absorbing well chamber group. Compared with the traditional design, the intensive design simplifies the layout and construction procedure of the chamber by considering the absorbing well, improves the stress conditions of the chamber, reduces the displacement and stress of the roadway, makes the plastic zone range smaller and more uniform, and eliminates the spatial effect. The deformation energy of the surrounding rock is released through the high constant resistance and large deformation of the NPR cable and the reserved gap between the truss and the surrounding rock, and the deformation of the surrounding rock is limited through the application of high prestress to the NPR cable and the strength of the three-dimensional truss material, which allows for the full use of the self-bearing capacity of the surrounding rock and effectively ensures the roadway stability. The field application shows that this strategy can effectively ensure the stability of the chamber group; the deformation of the surrounding rock is controlled within 70 mm, and there is no shedding, cracking, or destruction of the sprayed layer after concrete sealing, which indicates that the technology plays an important role in controlling the stability of the deep roadway and can provide a reference for similar projects.
Abstract: Considering that fluctuations in temperature can cause variations in both the internal structure as well as the mineral composition of rocks, their fracture characteristics must be impacted accordingly. With the exponential development of geotechnical engineering in cold regions, it is urgent to study the influence of the sub-zero temperature environment on the mechanical properties and dynamic properties of rocks. In order to investigate the influence of sub-zero temperature gradient on the dynamic fracture characteristics of rocks, red sandstone was used for the preparation of notched semi-circular bend specimens. First, a water-saturated machine and a sub-zero temperature incubator were utilized to pretreat the rock for 48 h, conducive for both satiation and freezing processes. Subsequently, the dynamic tests were carried out utilizing an improved split Hopkinson bar experimental system with a cryogenic sub-system. Concurrently, the striker velocity was modulated by setting distinctive air pressures, following which the rock was loaded at various loading rates. The test results demonstrate that the fracture toughness of the rock has an evident loading rate effect, and the fracture toughness proliferates exponentially with the increase in the loading rate. In the event that the loading rate is certain, the fracture toughness of the rock primarily increases gradually and then expeditiously over the course of advancement from room temperature to ?20 ℃. Contradictorily, the rock fracture toughness diminishes abruptly with plummeting temperature. Analysis of the rock fracture process, accommodated by a high-speed camera, revealed that the fracture process of the rock at distinctive temperatures is fundamentally equivalent, and the crack propagation speed is negligibly influenced by the temperature. Furthermore, the rock fracture mode was analyzed by employing a scanning electron microscope (SEM) system. The SEM images of the rock depicted that the fracture of red sandstone at sub-zero temperature is predominantly intergranular fracture and cement tearing, accompanied by a trace of transgranular fracture. Meanwhile, the experimentation also revealed that the number of micro-cracks in the rock significantly multiplied when the temperature declined to ?25 ℃, illustrating that sub-zero temperature has a deteriorating effect on the rock. Conclusively, the influence mechanism of temperature on the internal structure of the rock is discussed, and it is assumed that the change in the internal structure of the rock is the collaborative effect of thermal expansion-cold contraction and ice-water phase transition. The interpretation of this study has substantial reference significance for the further consequential analysis of frigidity on the fracture properties of the rock.
Abstract: As a high-temperature, high-pressure, multi-phase reaction vessel, the converter is vulnerable to splashing or slag overflow. Good molten pool surge can expand the slag–gold reaction area and enhance steelmaking efficiency. Abnormal molten pool surge can cause metal loss, damage the furnace body and its auxiliary equipment, and even threaten the personal safety of workers working in front of the furnace. This paper summarizes the previous research findings on splashing mechanisms and influencing factors. According to the occurrence principle, converter splashes can be classified into explosive splashes, foam splashes, metallic splashes, and other splashes, among which explosive splashes are the most dangerous and foam splashes occur most frequently. The occurrence of splashing accidents can be generally attributed to the high-temperature melt splashing caused by bubbles produced during the vigorous chemical reaction in the furnace and the splashing produced by the flow energy generated during the top–bottom combined blowing of the molten pool. The influencing factors of splashing are discussed based on six aspects: loading system, slag making system, oxygen supply system, bottom blowing system, temperature system, and safety system, and the foam of slag, oxygen lance blowing parameters, and bottom blowing parameters are thoroughly examined. It is observed that the occurrence of a splashing accident is frequently caused by the coupling of multiple factors. It is mainly one-sided, and hence the cause of the splashing accident cannot be unilaterally analyzed. Currently, no methods are present that can effectively quantify the effect of each factor on the splashing. Thus, developing a set of safety evaluation models suitable for converter splashing is imperative. Furthermore, the author summarizes the existing splash prediction models, examines the benefits and drawbacks of some splash prediction models, and summarizes the prediction principles and some application outcomes of the furnace gas analysis, audio analysis, and image analysis methods. Although preliminary progress has been made in the study of prediction models, there are still challenges that need to be overcome. It is pointed out that the reason why the existing prediction models have not been widely used is due to the low prediction accuracy, short prediction time, high cost, and low practicability. Several researchers have used a combination of several models to predict splash in converter. The findings reveal that various models can learn from each other, and the prediction accuracy of the comprehensive model is higher than that of the single model. Furthermore, the splash prediction model will become more intelligent and refined in the future.
Abstract: Microalloying and heat treatment are the most important ways to improve the steel properties . In this study, the precipitation behavior of NbC precipitates with Nb content of 210 × 10–6, 430 × 10–6, and 690 × 10–6 and heat treatment temperature of 1000, 1100, 1200, and 1300 ℃ were investigated. DH980 slab was melted in a silicon–molybdenum heating furnace with different Nb additions. The water-quenched steel samples were heated at different temperature in a furnace. The morphology and chemical composition of inclusions in steel samples were determined using an inclusion analysis system. The main inclusions in the Nb micro-alloyed high-Al high-strength steel were Al2O3, MnS, and NbC. The measured diameter of NbC precipitates ranged from 0.7 to 6.0 μm, which mainly concentrated on 1.0–2.0 μm. The precipitation temperature and amount of NbC were calculated using the thermodynamic calculation software Factsage. The initial precipitation temperature of NbC precipitates gradually increased to 1125, 1200, and 1260 ℃ as the Nb content increased from 210 × 10–6 to 690 × 10–6, respectively, and the NbC precipitation rate (the ratio of NbC mass to the mass of all inclusions) increased to 0.023%, 0.047%, and 0.076%, respectively. The precipitation temperature of MnS was 1450 ℃, which changed little with the content of Nb. Al2O3 was present at the melting temperature of steel. The amount of the precipitated NbC in steel increased with an increase in Nb content and heat treatment temperature. The NbC was dissolved in steel when the heat treatment temperature was 1300 ℃, resulting in a decrease in the precipitation of NbC. The size of the NbC precipitates was mainly influenced by the Nb content, heat treatment temperature, and heating time. With the increase in the initial Nb content, the difference in Nb content between the steel matrix and reaction boundary became larger, the diffusion driving force increased, and thus the size of NbC precipitates increased. The diffusion coefficient of Nb varied with the heat treatment temperature, which was hardly influenced by the Nb content. The diffusion coefficient increased with the increase in temperature, which promoted the diffusion of Nb. Consequently, the size of NbC increased with the temperature increased. The diffusivity of Nb increased with an increase in heating time, which also increased the size of NbC. Therefore, the size of NbC precipitates increased as the Nb content, heat treatment temperature, and heating time increased. A kinetic model of NbC precipitation in high-Al high-strength steel was developed to predict the effects of Nb content, heat treatment temperature, and heating time on the size of NbC precipitates.
Abstract: NH3-selective catalytic reduction of NOx over a V2O5–WO3/TiO2 catalyst is the major control method of NOx and has been successfully promoted and applied in various large steel enterprises in China. The production of waste catalysts (hazardous waste) from flue gas denitrification in iron and steel enterprises increases annually. Harmless landfills and wet purification are widely-employed methods for the treatment of waste catalysts. However, these methods pose environmental problems such as resource wastefulness, excessive amounts of acid/alkali, and considerable secondary pollution. Optimizing the effective use and disposal of such waste catalysts has become a key common problem in the industry. In this work, a novel method for producing titanium-bearing pellets by adding waste catalysts to pellet material was introduced. The feasibility of using waste catalysts to prepare titanium-bearing pellets was comprehensively evaluated by comparing the preparation process and metallurgical properties of the resulting pellets with those of commercially available titanium-containing pellets. The findings of this study reveal that the addition of 5.0% waste catalyst to the raw material can substantially improve the overall comprehensive performance of green pellets. Moreover, the drop number (dropped from 0.5 m height), average compressive strength, and burst temperature of the green pellets increased from 3.8 times, 16.5 N, and 487 ℃ (without waste catalyst addition) to 7.7 times, 21.5 N, and 553 ℃, clearly outperforming the ordinary titanium-bearing pellets prepared using vanadium–titanium magnetite (1.4 times, 15.0 N, and 542 ℃). These results could be attributed to the physical properties of the waste catalyst, which is a porous material with abundant hydrophilic groups on the surface. These hydrophilic groups, comprising hydroxyl groups, lead to the presence of more capillary water on the catalyst particle surfaces. Furthermore, the capillary force played an important role in various interactions in the pelleting process, thus improving the performance of mixtures. After roasting, the average compressive strength of the pellets containing the waste catalyst was 3083 N, higher than the 2630 N for ordinary titanium-bearing pellets. However, the short preheating and roasting time resulted in partially unreacted TiO2 being present in the internal pores of the pellets as rutile-type particles. The consolidation mechanism of pellets containing waste catalysts demonstrated that TiO2 in the waste catalyst reacts with iron oxide to form a Fe2TiO5 bond, while unreacted TiO2 reduces the compressive strength of the pellets. The metallurgical properties of the two titanium-bearing pellets are virtually identical to those of ordinary oxidized pellets, indicating that the pellets containing waste catalysts can be used in blast furnace protection smelting. This study offers a new approach for recycling waste catalysts generated by flue gas denitrification in iron and steel enterprises.
Abstract: The development of deep learning techniques and support of big data computing power have revolutionized graph representation research by facilitating the implementation of the learning of different graph neural network structures. Existing methods, such as graph attention networks, mainly focus on global information propagation in graph neural networks, which have theoretically proven their strong representation capability. However, these general methods lack flexible representation mechanisms when facing graph data with local topology involving specific semantics, such as functional groups in the chemical reaction. Accordingly, it is of great importance to further exploit the local structure representations for graph-based tasks. Several existing methods either use domain expert knowledge or conduct subgraph isomorphism counting to learn local topology representations of graphs. However, there is no guarantee that these methods can easily be generalized to different domains without specific knowledge or complex substructure preprocessing. In this study, we propose a simple and automatic local topology inference method that uses variational convolutions to improve the local representation ability of graph attention networks. The proposed method not only considers the relationship reasoning and message passing on the global graph structure but also adaptively learns the graph’s local structure representations with the guidance of statistical priors that can be readily accessible. To be more specific, the variational inference is used to adaptively learn the convolutional template size, and the inference is conducted layer-by-layer with the guidance of the statistical priors to make the convolutional template size adaptable to multiple subgraphs with different structures in a self-supervised way. The variational convolution module is easily pluggable and can be concatenated with arbitrary hidden layers of any graph neural network. In contrast, due to the locality of the convolution operations, the relations between graph nodes can be further sparse to alleviate the over-squeezing problem in the global information propagation of the graph neural network. As a result, the proposed method can significantly improve the overall representation ability of the graph attention network using the variational inference of the convolutional operations for local topology representation. Experiments are conducted on three large-scale and publicly available datasets, i.e., the OGBG-MolHIV, USPTO, and Buchwald-Hartwig datasets. Experimental results show that exploiting various kinds of local topological information helps improve the performance of the graph attention network.
Abstract: At present, the flight safety work of civil aviation in China mainly investigates the probable causes of accidents and analyzes flight data after air accidents, causing numerous problems such as passive safety management and delayed risk control. To realize the early warning of flight risk during flight, a dynamic method for the evaluation of landing risk and early warning under the condition of future air–ground data real-time transmission was proposed. The landing stage, which has the most complex operation program and the highest accident rate during a flight, was taken as the research object, and future air-to-ground high-throughput interconnection scenarios comprising 5G and satellite networks were considered to solve the problem of advanced intelligent warnings and aircraft alarms in abnormal flights. First, according to the accident causation theory, the human factor reliability model, the system model, and other theories or models, a landing warning index system based on multisource real-time operation data and the integration of historical statistics and expert knowledge was established. Then, a grounding parameter prediction model was established to solve the problem of lag in the acquisition of four grounding parameters, namely ground pitch angle, ground speed, ground vertical rate, and 50 ft-ground horizontal flight distance in actual flight. This model classified the pilot’s landing operation mode by clustering ARJ21 historical landing data and determined the attribute mean value of the four parameters for each type of operation mode. Furthermore, according to decision field theory, the model discussed the landing mode selection of pilots with different personalities in different scenarios and calculated the selection probability of the pilot’s landing operation mode, thereby obtaining the predicted values of the four above-mentioned indicators. According to the above, aiming at the complexity and uncertainty of the landing risk early warning system, a reasoning method of the multilayer confidence rule base was proposed. This method improved the traditional reasoning method of the single-layer confidence rule base and adopted the bottom-up hierarchical reasoning method considering the complexity characteristics of the landing process, effectively integrating different sources and forms of qualitative or quantitative data. Thus, the dynamic assessment and reasoning of the landing risk were realized. Finally, using the reasoning-based calculation of the landing process for the “2020.10.16 Panzhihua runway grounding event” and “2010.8.2 Yichun air disaster,” the results verified the effectiveness of the method. It was found that the early warning time of the Panzhihua event can reach 13 s.
Abstract: Benefit maximization is the enduring objective of production and management for international oil companies. This objective can only be realized through oil and gas production; thus, enhancing production efficiency inevitably leads to maintaining and increasing the value of overseas assets. Given the current practices adopted by domestic companies to improve the quality and efficiency of overseas projects, it is imperative to establish a set of comprehensive benefit and output optimization methods that can withstand oil price shocks, adapt to overseas projects, and cater to various other requirements, thereby serving to improve the quality and efficiency of overseas projects. Taking into account the differences between overseas and domestic projects, this paper analyzes the characteristics and strategies for realizing benefits under different financial and tax regimes, such as mine tax contracts, output-sharing contracts, and service contracts. A framework for evaluating the benefits and outputs of overseas projects under different conditions (such as cost, output, and other floating indicators) is established based on research and analysis conducted at home and overseas. Overall marginal benefit, cash flow, and profit optimization objectives are used to guide benefit allocation to achieve asset appreciation and preservation. A multidimensional and multi-objective decision-making model and an algorithm for benefit maximization are developed by considering both profitability and risk. A Pareto solution set is provided for a comprehensive optimal decision-making interval of overseas oil and gas field project development by considering the constraints of investment, cost, and block in conjunction with multiple decision-making objectives such as production, profit, and risk. This model was applied to specific cases of overseas oil fields under certain decision objectives, and the Pareto optimal decisions were generated. An in-depth analysis and comparison of each solution in the solution set was conducted. Appropriate selection of different solutions is advocated for different decision preferences to ensure the objectivity and scientific nature of profit management decisions. Finally, considering the influence of uncertainty factors, the scenario-based uncertainty analysis of the project output, oil price, cost, and investment has proved effective and generally provides reliable support for decision making in the context of overseas oilfield efficiency and plans for production optimization and high-quality development. The benefit production model and solving algorithm for overseas oilfield projects developed in this paper can provide a theoretical basis and support for decision making by oil field companies to optimize the production benefits of overseas oil fields and improve profitability.
Abstract: The middle and lower reaches of the Yellow River are rich in silt. The Yellow River silt can be utilized as a subgrade filling material along the Yellow River expressway to enhance its resource utilization potential. However, research on the geotechnical mechanical properties of the Yellow River silt is limited. In this study, a series of triaxial shear tests were conducted using the global digital systems triaxial apparatus to examine the effects of initial conditions (confining pressure and relative density) and test conditions (shear rate and drainage conditions) on the static strength and deformation characteristics of the Yellow River silt. The stress–strain curve, volumetric strain curve, envelope of shear strength, stress ratio curve, and internal friction angle distribution under different characteristic states were obtained. The test results showed that the shear strength of the Yellow River silt was more sensitive to confining pressure, relative density, and drainage conditions. The stress–strain curves of the Yellow River silt samples under drained conditions showed a slight strain-softening phenomenon; therefore, there were three characteristic states: peak state, phase transformation state, and critical state. Moreover, the stress–strain curves of the Yellow River silt samples under undrained conditions showed strain hardening characteristics, and there existed three characteristic states: the peak state, critical state, and peak pore pressure state. Additionally, the Yellow River silt samples simultaneously reached the peak and critical states at the end of the shear procedure. Specifically, the strength at the peak and critical states increased with increasing confining pressure and relative density. The shear strength under the undrained conditions was greater than that under the drained conditions. The development of pore pressure under the undrained conditions was in contrast with the dilatancy characteristics under the drained conditions; however, the pore pressure developed more rapidly than that depicted by the dilatancy characteristics. The distribution interval of the friction angle at the characteristic states of the Yellow River silt was between 22.6° and 38.1°. The initial shear modulus and ultimate deviator stress of the Yellow River silt increased with increasing confining pressure and relative density but were not sensitive to the shear rate. The ultimate deviator stress under undrained conditions was greater than that under drained conditions, while the initial shear modulus under drained conditions was smaller than that under undrained conditions under medium–low confining pressure. To strengthen the shear resistance of Yellow River silt, more attention should be paid to improving the compaction degree when the Yellow River silt is used as the filling material of expressway subgrades. This study can provide data and theoretical references for the resource utilization of Yellow River silt in subgrade engineering.
Abstract: In the field of building energy conservation, solar energy is a highly favored clean energy source. However, the instability and discontinuity of solar energy greatly affect its application. Phase-change energy storage technology is widely used for solar energy storage because of its huge latent heat and constant temperature during phase change. To summarize the application effect and research status of phase-change energy storage technology in the field of solar energy storage, this paper reviews the research progress on solar energy storage tanks based on phase-change energy storage materials at home and abroad. This paper focuses on the research progress on phase-change material (PCM) packaging technology from the aspects of geometry packaging and microcapsule encapsulation. The improvements in material thermal conductivity, supercooling and phase separation problems, and material cycle durability are summarized and analyzed. Moreover, this paper summarizes and analyzes the existing research on the structural optimization design of solar thermal storage tanks, stratification of solar phase-change energy storage tanks, storage performance of solar phase-change energy storage tanks, operation strategy of solar phase-change energy storage systems, and performance improvement of the solar heating system by a phase-change energy storage tank. The advantages and disadvantages of solar energy storage tanks based on PCM energy storage in applications are summarized. Finally, the research idea of improving the performance of solar phase-change energy storage tanks is proposed. First, it is suggested that further research should be conducted on the encapsulation technology and heat transfer enhancement technology of composite PCMs, and the economic problems of PCM preparation should be fully considered. Second, the problem that PCM cannot completely melt or solidify during heat storage and release should be comprehensively studied to further improve the energy release performance of the heat storage tank. Third, the structural design and operation strategy of solar phase-change energy storage tanks should be optimized. Finally, to further explore the application potential of solar phase-change energy storage tanks, it is necessary to develop a multi-energy coupled heating system based on a solar phase-change energy storage tank, study the cascade utilization of various energy sources such as photothermal, photoelectric, and electromagnetic heat, and improve the stability and energy conversion efficiency of the multi-energy coupled heating system. This study aims to provide a reference for further research on and application of solar phase-change energy storage tanks.
Monthly, started in 1955 Supervising institution：Ministry of Education Sponsoring Institution：University of Science and Technology Beijing Editorial office：Editorial Department of Chinese Journal of Engineering Publisher：Science Press Chairperson：Ren-shu Yang Editor-in-Chief：Ai-xiang Wu ISSN 2095-9389CN 2095-9389
