Loading...

Table of Content

    23 January 2024, Volume 0 Issue 01
    Automatic Arrangement Method of Cloud Network Security Service Chain Based on SRv6 Technology
    WANG Hong-jie, XU Sheng-chao, YANG Bo, MAO Ming-yang, JIANG Jin-ling
    2024, 0(01):  1-5.  doi:10.3969/j.issn.1006-2475.2024.01.001
    Asbtract ( 78 )   PDF (1156KB) ( 116 )  
    References | Related Articles | Metrics
    Abstract: To improve the resource utilization rate of cloud network data centers and save communication costs, a cloud network security service chain automatic orchestration method is designed based on SRv6 (Segment Route IPv6) technology. The method assists and guides network data packets to pass through the cloud network along the specified path, determines the specific forwarding path of the message, and reduces dependence on service nodes; establishes an objective function to minimize the total bandwidth, combines with various constraints to meet the security requirements of automatic orchestration; defines local behavior message, constructs automatic arrangement framework of security service chain, establishes security service policy, solves policy conflict and flow network scheduling problem, and achieves security arrangement of service chain. Experimental results show that the proposed method can effectively implement the automatic scheduling of cloud service chain, reduce the average total bandwidth consumption of CPU, improve the success rate of user requests, reduce the load of edge device in the cloud, and save communication costs.
    A Universally-composable Secure Non-interactive Commitment Scheme
    CAI Si-mu, WANG Li-bin
    2024, 0(01):  6-12.  doi:10.3969/j.issn.1006-2475.2024.01.002
    Asbtract ( 51 )   PDF (1042KB) ( 56 )  
    References | Related Articles | Metrics

    Abstract: The commitment scheme is one of the most fundamental components in cryptography, and is the basis of many cryptographic protocols, such as zero-knowledge proof and secure multi-party computing protocols. Universally composability (UC) is of great significance in designing secure protocols, if a protocol is proven secure in the UC framework, it still maintains security even if it is executed concurrently with arbitrary (even insecure) protocols. Several current efficient UC commitment schemes are all interactive protocols, and non-interactive UC commitments have high computational cost and communication complexity of the protocol. Aiming at solving this problem, an efficient UC-secure non-interactive commitment scheme in the common reference string model is proposed. The key design idea of universally composable commitments are to achieve extractability and equivocability at the same time. A CCA2-secure encryption scheme is used to achieve extractability in the commitment phase. A non-interactive zero-knowledge proof is used in the decommitment phase, and a dual-model commitment scheme is utilized to maintain protocol equivocability. The proposed protocol reduces the multi-round communication to one round in the open phase, achieving the non-interactivity. Compared with the existing non-interactive commitment scheme, the cost of computation and communication are greatly reduced, and the efficiency of the protocol is improved.

    Overview of Data Processing Techniques for MIoT Based on Fog Computing
    HAN Kun, WANG Zheng, DUAN Jun-yong, YANG Hua-lin
    2024, 0(01):  13-20.  doi:10.3969/j.issn.1006-2475.2024.01.003
    Asbtract ( 43 )   PDF (1208KB) ( 126 )  
    References | Related Articles | Metrics
    Abstract:Manufacturing Internet of Things (MIoT) is a kind of technology that combines manufacturing production system with Internet connection. Data processing plays a crucial role in MIoT. With the continuous expansion of manufacturing scale, traditional cloud computing has gradually failed to meet the needs of data processing, while the development of fog computing can effectively reduce decision delay and improve system efficiency. This paper summarizes the MIoT data processing technology based on fog computing. Firstly, the generation and characteristics of MIoT data are introduced, as well as the challenges to be faced in the data processing process. Secondly, the MIoT data processing architecture based on fog computation is introduced. Then, the key techniques of data processing in fog computation are introduced. Finally, it introduces the challenges to be faced in the deployment of the architecture and the future direction of fog computing in MIoT.
    Computational Offloading Strategy Based on Multi-objective Optimization in D2D Network
    CHEN Qi, LI Jing-jing
    2024, 0(01):  21-28.  doi:10.3969/j.issn.1006-2475.2024.01.004
    Asbtract ( 39 )   PDF (1409KB) ( 74 )  
    References | Related Articles | Metrics
    Abstract: Focused on the high latency and energy consumption for computational offload in mobile edge computing scenarios with device-to-device (D2D) communication technology, a computational offloading strategy based on multi-objective optimization is proposed. The strategy is based on a computing offloading model with multi-objective optimization of delay and energy consumption, introduces the analysis of excessive offloading problem, improves the NSGA-II algorithm, including genetic encoding strategy, crossover and variation methods applicable to computing offloading, and minimizes task execution time and energy consumption by solving the Pareto optimum. In addition, a data routing algorithm is proposed, which balances the transmission energy consumption of routing devices and optimizes the routing paths. Through simulation experiments, the average boosting efficiency of the algorithm is up to 41.7% and the task retransmission rate is reduced to 7.8%. The experiment results show that the proposed algorithm can significantly reduce the execution delay, energy consumption, task retransmission rate and improve the task offload success rate.
    Visual Servo Based on Model-free Adaptive Control
    PENG Zong-yu, HUANG Kai-qi, SU Jian-hua, WANG Li-li
    2024, 0(01):  29-34.  doi:10.3969/j.issn.1006-2475.2024.01.005
    Asbtract ( 56 )   PDF (1958KB) ( 122 )  
    References | Related Articles | Metrics
    Abstract: The traditional robot visual servo control technology requires accurate dynamics and kinematics models of known robots and the calibration of camera. However, due to the errors in the robot modeling and camera calibration, it is difficult to accurately build the error model, which affects the positioning accuracy and convergence speed of the robot vision servo system. To solve this problem, this paper proposes a robot vision servo technology based on Model-free Adaptive Control (MFAC). Using the input and output data of the system, this paper realizes adaptive visual servo control. Namely by the Jacobian matrix in the MFAC online estimation robot servo controller and combining with sliding mode controller, this paper achieves the precise tracking task to targets. The results of simulation experiments show that the proposed method can ensure the smooth convergence of the servo controller under the unknown disturbance caused by the change of system parameters and reduce the system positioning error.
    ZigBee Indoor Location Algorithm Based on Dynamic Modification of RSSI Parameters
    LI Shi-bao, CONG Yu-jie
    2024, 0(01):  35-40.  doi:10.3969/j.issn.1006-2475.2024.01.006
    Asbtract ( 40 )   PDF (2045KB) ( 73 )  
    References | Related Articles | Metrics
    Abstract: ZigBee indoor positioning technology has developed rapidly in recent years, but the traditional algorithm using fixed path loss model has poor adaptability to the environment, resulting in large positioning errors and affecting positioning accuracy. This paper proposes an indoor location algorithm based on ZigBee platform with dynamic correction of logarithmic path loss model parameters. First, the RSSI value obtained is filtered and optimized by Gaussian filtering, and then the parameters of the logarithmic path loss model are dynamically modified according to the distance between anchor nodes and the RSSI value, including the path loss factor and the signal strength value from the node to be measured, so the specific logarithmic path loss model in the current environment is obtained; Then the Kalman filter is used to modify the existing positioning parameters twice, which can correct the positioning deviation caused by the environment change caused by the time change in the above algorithm. Experimental results show that the positioning performance of this algorithm is 46.8% higher than that of the fixed path loss model based on ZigBee, which can improve the positioning error caused by environmental changes.
    Scenes Text Modification Network for Uyghur Based on Generative Adversarial Network
    FU Hong-lin, ZHANG Tai-hong, YANG Ya-ting, Aizimaiti Aiwanier, MA Bo
    2024, 0(01):  41-46.  doi:10.3969/j.issn.1006-2475.2024.01.007
    Asbtract ( 44 )   PDF (2063KB) ( 84 )  
    References | Related Articles | Metrics
    Abstract: Through the study of scene text detection and recognition in Uyghur languages, it is found that manual acquisition of labeled natural scene text images is time-consuming and labor-intensive. Therefore, artificially synthesized data is used as the main source of training data. To obtain more realistic data,  a scenes text modification network for Uyghur based on generative adversarial network is proposed. The efficient Transformer module is used to construct the network for fully extracting the global and local features of the image to complete the modification of the Uyghur, and a fine-tuning module is added to fine-tune the final results. The model is trained with WGAN thought strategy, which can effectively cope with the problems of pattern collapse as well as gradient explosion. The generalization ability and robustness of the model are verified by text modification experiments in English-English and English-Virginia. Good results are achieved in both objective metrics (SSIM, PSNR) and visual effects, and are validated on real scene datasets SVT and ICDAR 2013.
    Improved RetinaNet Target Detection Method for Power Equipment
    WANG Qiu-yi, ZHOU Hao, ZHENG Ting-ting
    2024, 0(01):  47-52.  doi:10.3969/j.issn.1006-2475.2024.01.008
    Asbtract ( 40 )   PDF (1696KB) ( 45 )  
    References | Related Articles | Metrics
    Abstract: A RetinaNet-based target detection method for power equipment detection is proposed for the problem of low accuracy of small target recognition in power equipment detection. The anchor box size of original network is optimized by K-means clustering method firstly. Then shallow feature maps with higher resolution are added to feature fusion to solve the problem that the feature maps contain too little information after convolution through multiple layers. Based on this, ECA (Efficient Channel Attention) attention mechanism is introduced to enable the network to locate the effective features of power devices and suppress the useless feature information. The experimental results show that compared with the original method, the average recognition accuracy of the method in the paper is improved by 18.1 percentage points for five types of power equipment: electric towers, pins, construction vehicles, insulators and poles, which indicates that the improved method can significantly improve the detection level of power equipment.
    A Temperature Field Reconstruction Method of Furnace Tube Based on Bidirectional Multistep Prediction
    LIN Qi-zhao, PENG Zhi-ping, GUO Mian, CUI De-long
    2024, 0(01):  53-58.  doi:10.3969/j.issn.1006-2475.2024.01.009
    Asbtract ( 30 )   PDF (1682KB) ( 59 )  
    References | Related Articles | Metrics




    Abstract: Aiming at the difficulty of sensing the tube temperature in cracking furnace under high temperature closed ethylene cracking environment, a method of surface temperature field reconstruction of cracking furnace tube based on fusion mechanism and Long Short-Term Memory(LSTM) is proposed. Firstly, the mechanism model of ethylene cracking reaction is constructed based on fluent, a computational fluid dynamics simulation platform, which is used to describe the mathematical relationship between cracking reaction and furnace tube temperature. Then, the mechanism model is numerically corrected and the process parameters are solved using the industrial field data. Major process parameters with strong applicability are determined based on Pearson correlation coefficient. Based on this, a convolutional block attention module (CBAM) is designed to extract the characteristics of the main process parameters reflecting the relationship between the cracking reaction and the temperature of the furnace tube. Finally, a bidirectional multistep prediction model (GA-BMLSTM) is designed based on genetic algorithm and long and short memory neural network to predict the temperature distribution of furnace tubes. Experimental results show that this method has high accuracy and applicability to the reconstruction of temperature field of furnace tube.
    Key words: ethylene cracking furnace; temperature field reconstruction; computational fluid dynamics; attention mechanism; genetic algorithm
    Regformer: Hydraulic Prediction Model of Oil Pipeline Based on GS-XGBoost
    LI Ya-ping, WANG Jun-fang, YU Hong-mei, DOU Yi-min, XIAO Yuan, TIAN Ji-lin
    2024, 0(01):  59-66.  doi:10.3969/j.issn.1006-2475.2024.01.010
    Asbtract ( 26 )   PDF (5692KB) ( 58 )  
    References | Related Articles | Metrics
    Abstract: Hydraulic pressure drop prediction is very important for production regulation of oil pipelines, and current machine learning methods regard pressure drop prediction as a regression problem, however, pipeline hydraulic calculation is affected by many factors, and the fixed weights obtained from the training set by traditional machine learning methods are difficult to generalize to more test samples or real engineering scenarios. This paper proposes a hydraulic pressure drop regression prediction method, Regformer, which introduces a sparse attention mechanism into the regression task, designs a smoothing probability method based on multi-headed attention, and incorporates a feature projection mechanism. In a comparative experimental analysis with seven mainstream methods on 10 public data sets, qualitative experiments show that Regformer has good fitting ability for local mutations; experiments on hydraulic pressure drop prediction show that the self-attentive method has significant advantages for regression tasks with multivariate uncertainty, especially for extreme cases reflecting the importance of adaptive regression parameters, and Regformer achieves better performance than Transformer with less computation, verifying the superiority of the proposed sparse attention and adaptive feature projection for the hydraulic pressure drop prediction task.
    Two-stage Critical Disease Prediction Model Based on Heterogeneous Attribute Fusion
    ZHAN Shao-qiang, ZENG An, ZHANG Yi-qun, SUN Hong-tao, ZHANG Xiao-bo
    2024, 0(01):  67-73.  doi:10.3969/j.issn.1006-2475.2024.01.011
    Asbtract ( 27 )   PDF (1777KB) ( 41 )  
    References | Related Articles | Metrics
    Abstract: With the emergence and wide application of Electronic Health Record (EHR), the prediction model based on EHR data can be used for early detection and intervention of diseases. Heterogeneous attributes are ubiquitous in EHR data, but it is difficult to thoroughly exploit their information. Therefore, the method of heterogeneous attribute fusion provides an informative data representation basis for subsequent model training. This paper designs an efficient two-stage prediction model for solving the problems of time and cost in predicting critical illness. In the first stage of the model, coarse-grained prediction is performed on patient samples. Patients with low severity are initially screened out, which plays a key role in patient diversion. The second stage makes more fine-grained predictions of potentially critical patients based on the coarse filtering results of the first stage. The experimental verified that, after heterogeneous attribute fusion, when we select the first 6 time points to construct a non-temporal model, the two-stage model has better performance in both initial disease screening and disease prediction.
    Unsupervised Domain Adaptation for Outdoor Point Cloud Semantic Segmentation
    HU Chong-jia, LIU Jin-zhou, FANG Li
    2024, 0(01):  74-79.  doi:10.3969/j.issn.1006-2475.2024.01.012
    Asbtract ( 49 )   PDF (2419KB) ( 102 )  
    References | Related Articles | Metrics

    Abstract: An unsupervised domain adaptation for LiDAR semantic segmentation method is proposed to deal with the problem of excessive data required for semantic segmentation network training in outdoor large-scale scenes. The method uses a modified RandLA-Net for semantic segmentation using a small number of point clouds from the SPTLS3D’s real world data as target objects. The model finishes the pre-training of the segmentation network on SensatUrban, and completes the transfer task by minimizing the domain gap between the source and target domains. The RandLA-Net losses the global features of the original point cloud in the encoding process, so an additional method of obtaining global information to join the network decoding is proposed. In addition, for getting the differentiated information, the weights of the local attention module of RandLA-Net is changed to use the difference between the features of each point and the average features of its neighbors. The experiments show that the mean intersection over union  of the network are 54.3% on SemanticKITTI and 71.91% on Semantic3D. The mIoU of the pre-trained network after fine-tuning are 80.05%, which is 8.83  percentage points better than training directly.

    A DNN Compression Method for Environmental Sound Classification on Microcontroller Unit
    MENG Na, FANG Wei-wei, LU Hong-ying
    2024, 0(01):  80-86.  doi:10.3969/j.issn.1006-2475.2024.01.013
    Asbtract ( 13 )   PDF (1307KB) ( 68 )  
    References | Related Articles | Metrics
    Abstract: Environmental Sound Classification (ESC) is known as one of the most important topics of the non-speech audio classification task. In recent years, deep neural networks (DNNs) have made a lot of progress in ESC. However, DNNs are computationally and memory-intensive, and cannot be directly deployed on IoT devices based on microcontroller units (MCU). To address this problem, this paper proposes a DNN compression method for highly resource-constrained devices. Since DNNs have a large number of parameters, which cannot be directly deployed, so this paper proposes to use the pruning method for substantial compression. Afterwards, aiming at the problem of accuracy loss caused by this operation, we design a knowledge distillation based on the feature information of multiple intermediate layers. Tests are carried out on public datasets (UrbanSound8K, ESC-50) using the STM32F746ZG device. The experimental results demonstrate that proposed method can achieve up to 97% compression rate while maintaining good inference performance and speed.
    Named Entity Recognition in Electronic Medical Record Based on BERT
    ZHENG Li-rui, XIAO Xiao-xia, ZOU Bei-ji, LIU Bin, ZHOU Zhan
    2024, 0(01):  87-91.  doi:10.3969/j.issn.1006-2475.2024.01.014
    Asbtract ( 57 )   PDF (992KB) ( 147 )  
    References | Related Articles | Metrics
    Abstract:Electronic medical record is an important resource for the preservation, management and transmission of patients’medical records. It is also an important text record for doctors’ diagnosis and treatment of diseases. Through the electronic medical record named entity recognition (NER) technology, diagnosis and treatment information such as symptoms, diseases and drug names can be extracted from the electronic medical record efficiently and intelligently. It is helpful for structured electronic medical records to use machine learning and other technologies for diagnosis and treatment regularity mining. In order to efficiently identify named entities in electronic medical records, a named entity recognition method based on BERT and bidirectional long short-term memory network (BILSTM) with fusion adversarial training (FGM) is proposed, referred to as BERT-BILSTM-CRF-FGM (BBCF). After preprocessing by correcting the Chinese electronic medical record corpus provided by the 2017 National Knowledge Graph and Semantic Computing Conference (CCKS2017), the BERT-BILSTM-CRF-FGM model is used to recognize five types of entities in the corpus, with an average F1 score of 92.84%. Compared to the BERT model based on the inflated convolutional neural network (BERT-IDCNN-CRF) and the conditional random field model based on BILSTM (BILSTM-CRF), the proposed method has higher F1 score and faster convergence speed, which can more efficiently structure electronic medical record text.
    An Autonomous Navigation Method for Intelligent Vehicles in Urban Battlefield
    LI Peng, XU Luo
    2024, 0(01):  92-98.  doi:10.3969/j.issn.1006-2475.2024.01.015
    Asbtract ( 24 )   PDF (2793KB) ( 92 )  
    References | Related Articles | Metrics
    Abstract: The urban battlefield is the main position of conventional warfare and daily security, and excellent urban battlefield penetration capabilities can help our fighters better and faster complete reconnaissance, strike, rescue and other tasks. However, the complex street environment in the city, and the possibility of interception by enemy targets, make the urban battlefield environment complex and changeable, greatly increasing the difficulty of completing the mission. Traditional path planning methods rely on accurate static maps and rule constraints, and lack flexibility and adaptability. Therefore, this paper proposes an autonomous navigation method for intelligent vehicles in urban battlefield, and designs discrete action spaces and reward functions based on task completion. Firstly, this paper takes the urban battlefield penetration task as an example to design the state space and action space, and selects a suitable deep reinforcement learning algorithm. Then, based on Gazebo simulation platform and ROS, the algorithm flow framework and experimental scheme are designed. The experimental results show that the intelligent car using this method in the urban battlefield environment can effectively pass through obstacles and avoid enemy units to reach the designated place, which improves the success rate of penetration.
    Recommended Technology for Solder Paste Printing Process Parameters on Data Driven
    SU Xin
    2024, 0(01):  99-102.  doi:10.3969/j.issn.1006-2475.2024.01.016
    Asbtract ( 24 )   PDF (1566KB) ( 41 )  
    References | Related Articles | Metrics
    Abstract: Aiming at the problems of subjectivity and strong dependence on experience in the process of solder paste printing of printed boards, a way to recommend printing process parameters on data driven method is proposed. Firstly, the prediction models of component solder paste printing quality for each component is constructed, which concluded three sub models for each component: printing qualification rate prediction, solder paste relative area/volume prediction, and printing defect type prediction. Then, the recommended model of solder paste printing process parameters for printed board is structure aiming to optimize the printing quality of each component. Finally, the correctness of the models were verified based on the actual printing data, the average accuracy of the prediction of the printing qualification rate reached 99%, the deviation between the recommended process parameters and the empirical value was less than 10%. All these means the results of quality prediction and process parameter recommendation can meet the requirements of practical production and application.
    View Frustum Culling Algorithm for Scene Based on Optimized Octree
    LI Ying-ying, HUANG Wen-pei
    2024, 0(01):  103-108.  doi:10.3969/j.issn.1006-2475.2024.01.017
    Asbtract ( 29 )   PDF (1038KB) ( 74 )  
    References | Related Articles | Metrics
    Abstract:Large-volume 3D models are prone to low rendering frame rate, slow display and large resource consumption on the browser side. The reason is that such models usually contain hundreds of millions of triangular slices, which cannot be loaded and rendered quickly in a limited time. To address such problems, a scene view frustum culling algorithm based on an optimized octree is proposed. The algorithm adopts address code (Morton code), node view distance criterion and on-demand incremental division technique, which makes the octree adaptive with good compression efficiency; it adopts double bounding volume and base intersection test techniques to improve the accuracy of view frustum culling and achieves the overall goal of improving rendering frame rate and smooth display. The high-speed train example model study shows that the proposed algorithm improves the average rendering frame rate by about 14 frames and the spatial compression rate by about 37.8 percentage points compared with the traditional octree view frustum culling algorithm.
    Improved DOA Based on PWLCM and Bald Eagle’s Swooping Mechanism
    OU Ji-fa, CAI Mao-guo, HONG Guang-jie, ZHAN Kai-jie
    2024, 0(01):  109-116.  doi:10.3969/j.issn.1006-2475.2024.01.018
    Asbtract ( 26 )   PDF (1163KB) ( 70 )  
    References | Related Articles | Metrics
    Abstract: Aiming at the problems of slow convergence speed and low optimization accuracy of dingo optimization algorithm (DOA), an improved dingo optimization algorithm (IDOA) based on PWLCM and the bald eagle’s swooping mechanism is proposed. Firstly, a piecewise linear chaotic map with eriodicity is used to initialize the dingo population, effectively increasing the diversity of the dingo population. Secondly, the bald eagle’s swooping mechanism is introduced into the persecution strategy to  accelerate the speed of prey capture and strengthen the ability of the algorithm to explore local areas. Finally, the spiral search factor is introduced into the scavenger strategy to enhance the local development and exploration ability of the algorithm, so as to further improve the optimization speed and accuracy of the algorithm. Simulation experiment data, ablation experiment and Wilcoxon rank sum test all show that the proposed IDOA has better optimization speed and optimization accuracy than other comparison algorithms; Compared to other improved dingo optimization algorithms, the proposed IDOA shows better overall performance.
    Adaptive Bald Eagle Search Algorithm with Dynamic Learning and Gaussian Mutation
    XIA Huang-zhi, CHEN Li-min, MAO Xue-di,
    2024, 0(01):  117-126.  doi:10.3969/j.issn.1006-2475.2024.01.019
    Asbtract ( 41 )   PDF (6442KB) ( 61 )  
    References | Related Articles | Metrics
    Abstract: To address the problems of uneven initial population distribution, poor individual adaptability and the tendency to fall into local optimality in bald eagle search algorithm, an improved bald eagle search algorithm is proposed for solving function optimization problems. Firstly, the Circle chaos mapping strategy is introduced in the initialization phase to enrich the diversity of the initial bald eagle individuals. The nonlinear weights are introduced to break the inherent linear search pattern of bald eagle individuals in the selected search space phase, and adaptively adjust the ability of the algorithm to search and exploit. Secondly, the bald eagle leader learns dynamically from the representative bald eagle individuals in the best search position. The purpose is to update the individual adaptive bald eagles during the spiral search. Finally, the Gaussian variation strategy is executed for the bald eagle individuals in the best search position, and the bald eagle leader individuals in the curve swoop process are updated iteratively according to the size of individual fitness, and the fitness of most bald eagle individuals in the population is enhanced, which can avoid the stagnation situation of algorithm in the function search to a certain extent. Based on some benchmark test functions and comparative experiments of some CEC2017 functions, the superiority of the algorithm proposed in this paper is verified.