Table of Content

08 June 2022, Volume 0 Issue 05

Previous Issue    Next Issue

Deep Recommendation Algorithm Integrating Triple Attention Mechanism and Review Score

ZHANG Zong-hai, YU Yue-cheng, FENG Shen

2022, 0(05):  1-9. 

Asbtract ( 272 )   PDF (1385KB) ( 91 )  

References | Related Articles | Metrics

Many shopping websites have a large amount of review information written by users. Although most recommendation systems use review information, there is still much room for improvement. On the one hand, the information in the comments is uneven, mixed with a lot of useless information; on the other hand, most existing recommendation systems assume that a user’s attention to a certain product feature is the same for all products and cannot accurately reflect user preferences. This paper proposes an aspect-aware depth recommendation model ANAP that integrates triple attention and review score. Starting from the two levels of words and features, the important information in the review text is extracted by constructing two different attention networks to reduce the impact of useless information; in order to accurately reflect user preferences, the attention interaction network is constructed to capture the user’s different attention to various aspects of different items, and to achieve fine-grained modeling of aspect perception. This paper conducts experiments on 6 real data sets and designs an attention mechanism comparison experiment. The results show that the ANAP model effectively improves the score prediction accuracy, and the mean absolute error (MAE) is lower than the existing best algorithm by 4.86 percentage points.

 Traffic Accident Text Information Extraction Model Based on BERT and BiGRU-CRF Fusion

FAN Hai-wei, QIN Jia-jie, SUN Huan, ZHANG Li-miao, LU Xin-siyu

2022, 0(05):  10-15. 

Asbtract ( 437 )   PDF (936KB) ( 103 )  

References | Related Articles | Metrics

Aiming at existing traffic accident text data has difficulties in effectively extracting a large number of key heterogeneous data such as time, place and casualty loss, and the accuracy of traffic accident text information extraction methods based on static word vector deep learning model is low. The BERT (Bidirectional Encoder Representations from Transformers) is used for a dynamic vector mapping of the text characters in order to resolve the problem of ambiguity and context dependence insufficient from the source of data representation. The vectored features of text are extracted by using BiGRU(Bi-Gate Recurrent Unit) and text sequences with high features are output. Based on CRF (Conditional Random Fields), the probabilistic advantage of the global optimal output node is calculated to optimize the feature results of text sequence, and a BERT-BiGRU-CRF fusion model based on dynamic word vector is proposed forextracting the key information of traffic accident text. The comparison experiment shows that the average accuracy of the model in traffic accident text information extraction is 0.952 and F1 is 0.925, and 6.3 percentage points and 7.9 percentage points higher respectively than those of the model based on static word vector Word2Vec.

ECG Signal Classification Based on Deep Learning

YU Yan, QIU Lei,

2022, 0(05):  16-20. 

Asbtract ( 515 )   PDF (1182KB) ( 115 )  

References | Related Articles | Metrics

Electrocardiogram (ECG) can reflect the state of the heart in real time, and can be used for the accurate diagnosis of arrhythmias and other cardiovascular diseases. In view of the noise interference during ECG signal acquisition, we reconstruct the fourth-order components of Db6 wavelet, then use Butterworth low pass filter to realize double denoising. Then, from denoised ECG signals to extract the R-wave, and the P-QRS-T are intercepted and input into the one-dimensional improved GoogLeNet model for training. One-dimensional improved GoogLeNet is an improved structure of the original two-dimensional GoogLeNet, which reduces the network depth and adds the maximum pooled layer and dilated convolution in the sparse connection to increase the receptive field, so as to reduce the amount of calculation and improve the training performance. Experiments on the MIT-BIH data set show that the classification accuracy is 99.39%, which is 0.17 percentage points and 0.22 percentage points higher than the one-dimensional GoogLeNet and the original GoogLeNet respectively, and the training efficiency is improved. Signal classification has a marked improvement over other advanced techniques.

Occupational Competence Evaluation Model Based on Affinity Propagation Clustering

DUAN Gui-qin, ZOU Chen-song

2022, 0(05):  21-27. 

Asbtract ( 171 )   PDF (1229KB) ( 85 )  

References | Related Articles | Metrics

Aiming at the problem that the number of clusters in the application of clustering algorithm in educational big data depends on human experience, a new clustering effectiveness index is proposed. The sum of the distances between all samples in the cluster and the cluster center is used to represent the compact density in the cluster, and the minimum value of the sum of the distances between any two clusters is used to represent the degree of separation between clusters. By balancing the relationship between the compact density in the cluster and the degree of separation between clusters, the division of optimal clustering is realized. The test results on UCI and KDD CUP99 data sets show that the clustering quality evaluation results of the new index are effective and reliable. On this basis, a new clustering analysis model is designed by combining with the nearest neighbor propagation algorithm. The model is used to cluster the professional ability of college students. The results show that the new model can accurately give the number of clusters k, effectively excavate students’ career tendency, can provide basis and decision-making for college students’ career potential analysis and enterprises’ talent selection.

Tibetan-Chinese Bidirectional Machine Translation Based on VOLT

SUN Yi-dong, YONG Cuo, YANG Dan,

2022, 0(05):  28-32. 

Asbtract ( 291 )   PDF (845KB) ( 84 )  

References | Related Articles | Metrics

The generation of Tibetan-Chinese vocabulary is not only the first step of Tibetan-Chinese bi-directional machine translation task, but also affects the performance of Tibetan-Chinese bi-directional machine translation. This paper improves the performance of downstream Tibetan-Chinese bidirectional translation by improving the generation of Tibetan-Chinese word lists. On the one hand, it starts with word list splicing, using normal word lists for high frequencies and byte pair encoding word lists for low frequencies, and finding the optimal word frequency threshold through iterative training; On the other hand, according to the optimal transport theory proposed by vocabulary learning approach, the Tibetan-Chinese vocabulary is generated, which is improved according to the characteristics of Tibetan language and applied to Tibetan-Chinese bidirectional translation. The experimental results show that, it is demonstrated that the byte pair encoding plus optimal transmission lexical learning method proposed in this paper for Tibetan language characteristics works best, reaching a BLEU value of 37.35 for the Tibetan-Chinese translation task and 27.60 for the Chinese-Tibetan translation task.

A Financial News Sentiment Analysis Method Based on Graph Convolutional Neural Network and Dependency Analysis

YAO Chun-hua, ZHANG Xue-lei, SONG Xing-yu, ZHANG Ju, CAI Jia-zhi, FENG Ao

2022, 0(05):  33-39. 

Asbtract ( 318 )   PDF (1086KB) ( 71 )  

References | Related Articles | Metrics

Sentiment analysis of financial news helps enterprises and investors to determine investment risks and improves economic benefits, resulting in high application value. Graph neural networks have excellent performance in text classification, and have been applied to the field of sentiment analysis. In this paper, we propose a sentiment analysis method that uses dependency syntax analysis in graph convolutional neural networks (Dependency Analysis-based Graph Convolutional Network, DA-GCN) for financial news. This method obtains the word order information of the sentence and the syntactic in the document by analyzing the dependency of words in the document. It then implements information propagation and weight updates in the graph with co-occurrence information in each document. Experiments on a financial news dataset show that our model achieves significant performance improvements over traditional deep learning methods.

Co-fly Immune Optimization Algorithm Based on Innate Immune Mechanism of Fruit-fly

ZHANG Xiao-ru, WANG Dan, WANG Jun, ZHOU Jin-cheng,

2022, 0(05):  40-45. 

Asbtract ( 126 )   PDF (1196KB) ( 60 )  

References | Related Articles | Metrics

Based on the simple and effective innate immune response mechanism of fruit-fly, a collaborative optimization algorithm of fruit-fly’s immune is discussed. The synergistic immune response mechanism of melanization, cellular immunity and humoral immunity is used to design the population division and subgroup evolution in this algorithm. The algorithm has the merits of structural simplicity, the balance of internal and external circulation and the collaborative design are induced to ensure the optimization effect and efficiency of the algorithm. Comparative numerical results show that the algorithm has obvious advantages in search ability, search efficiency and high-dimensional functions optimizationand so on.

Artificial Fish Swarm Algorithm Based on PSO Adaptive Dual Strategy

LIU Zhi-feng, SHU Zhi-hao, XU Yue-feng, YANG Shu-yi, SHEN Wen-long

2022, 0(05):  46-53. 

Asbtract ( 230 )   PDF (1211KB) ( 66 )  

References | Related Articles | Metrics

In order to improve the traditional artificial fish swarm algorithm (AFSA), which is easy to fall into local optimal, poor robustness and low search accuracy, an adaptive dual-strategy artificial fish swarm algorithm based on particle swarm algorithm is proposed. Firstly, the algorithm simulates the moving operator of particle swarm optimization algorithm to adjust the moving direction and position of artificial fish, so that artificial fish has inertia mechanism and better expand new areas, so as to provide more opportunities for exploring potential better solutions and enhance its ability to jump out of local optimization. Then a strategy of adaptive field of view and inertia weight is used to balance the relationship between global search and local search. Finally, the opposition-based learning mechanism is introduced to design the random behavior of the two strategies to avoid the blindness of the original random behavior and increase the diversity of the fish. The simulation results show that the improved algorithm is better than other artificial fish swarm algorithms in terms of optimization accuracy, convergence speed and robustness, and has a good optimization effect in solving high-dimensional problems.

An improved Particle Swarm Optimization (PSO) Force Unloading Algorithm for Multi-task and Multi-resource Moving Edge Computing Environment

ZHANG Yan-hu, YAN Li-juan, MA Zhi-fen, ZHANG Yan-jun

2022, 0(05):  54-60. 

Asbtract ( 248 )   PDF (1473KB) ( 73 )  

References | Related Articles | Metrics

In order to study the energy consumption of mobile devices in multi-resource complex environment, an improved particle swarm optimization (PSO) algorithm for unloading calculation of mobile edge devices is proposed. Firstly, a computing model is proposed based on the multi-environment energy consumption of mobile devices.Secondly, a fitness algorithm is designed to measure the advantages and disadvantages of resource allocation schemes for computing resource problems.Finally, an improved particle swarm optimization (PSO) algorithm for energy allocation is presented for solving the optimal solution to further reduce the energy consumption scheduling and allocation scheme of mobile devices.The comparison of energy consumption system response time and other indicators of mobile devices under various unloading strategies by simulation software shows that the proposed algorithm has a better performance in solving the optimal solution to reduce the energy consumption scheduling and allocation scheme of mobile devices on the premise of satisfying the user response time.

An Edge Caching Strategy for Intelligent Manufacturing Supervision of Electrical Equipment

LI Ling, CHEN Xi, SHEN Wei-jie, XIONG Han-wu, CAI Ran-ran

2022, 0(05):  61-67. 

Asbtract ( 146 )   PDF (2035KB) ( 57 )  

References | Related Articles | Metrics

Intelligent IoT of electrical equipment is conducive to the realization of high-level development of the electric equipment industry, while the traditional information interaction architecture cannot meet the information interaction requirements of intelligent IoT. Aiming at the problem of information distribution efficiency of electrical equipment intelligent manufacturing monitoring platform, an information distribution framework of electrical equipment based on edge cache is designed. Considering the cost of network construction and the delay demand of business data distribution, and aiming at maximizing the benefit of service cache of power companies, an edge information cache strategy for intelligent supervision of electrical equipment is studied. To reduce the complexity of problem solving, a cache decision algorithm based on improved particle swarm optimization is proposed. The simulation results show that the proposed caching algorithm achieves the high caching service benefits of power companies while guaranteeing the data caching requirements with high partial delay priority.

On-line Detection System of CNC Tool Breakage Based on Attention-LSTM

WANG Ke-yang, ZHANG Yao, LI Ke-wen, ZHANG Bao-qian, LI Jiang, REN Jie-wen

2022, 0(05):  68-74. 

Asbtract ( 359 )   PDF (1265KB) ( 92 )  

References | Related Articles | Metrics

In order to detect tool damage during the batch processing of CNC machine tools to reduce defective products, an online monitoring method for tool damage in CNC production lines based on Attention-LSTM using machine tool spindle power information is proposed. The method uses the built-in sensor of the numerical control system as the data source to obtain the time series of the spindle power of the machine tool. In the data collection link, it is necessary to distinguish the different processes in the machining process and the tool number used in the process. Therefore, in the data acquisition link, the NC code and spindle power are collected at the same time, the collected data is processed by the NC code analysis method, the process identification of the processing process is completed, the Attention-LSTM algorithm is used to predict the spindle power data, and then the DTW algorithm is used to calculate the time series similarity. The degree of similarity between the processing power time series and the standard time series should be within a reasonable threshold range, otherwise it is considered that tool breakage occurred during the processing. Experiments were conducted on the FANUC CNC system to verify the accuracy of tool breakage recognition.

Hierarchical Representation of Power Text Named Entity Recognition and Project-expert Matching

YANG Zheng, CAI Di, LI Hui-bin

2022, 0(05):  75-81. 

Asbtract ( 195 )   PDF (1280KB) ( 73 )  

References | Related Articles | Metrics

To address the project-expert matching problem existing inthe evaluation work of the application for science and technology projects in the power field, this paper proposes a novel hierarchical word representation model (Attention-RoBerta-BiLSTM-CRF, ARBC) for power text named entity recognition. Moreover, a project-expert matching algorithm is also presented based on semantic and pictorial double feature space mapping strategy. ARBC model consists of a word embedding module, a Bi-directional Long Short-Term Memory (BiLSTM) module and a Conditional Random Field (CRF) module. The hierarchical word embedding module utilizes the information of word, sentence and document of the power text. Specifically the word embedding vector based on the pre-trained RoBerta model is extracted firstly. Then, the contextual representation of any sentence is enhanced by introducing an attention mechanism based on word frequency-inverse document frequency values at the document level. Finally, the word embedding and sentence embedding are linearly weighted and fused to form a hierarchical representation vector of a given word. Once the named entities of power texts are recognized by ARBC model, the task of entity effetive accurate matching between power projects and experts is achieved by the semantic and pictorial double feature space mapping strategy. Experimental results demonstrated on a set of 2000 power project abstract texts for the task of named entities recognition, and a F1 value of 83% is achieved based on the ARBC model, which is significantly higher than the widely used pre-trained models such as Bert and RoBerta. In addition, the entity matching strategy based on double feature space mapping achieves 85% accuracy for the power text-expert matching task.

Design of Structured Data Registration Engine Based on Data Architecture

HUANG AN-qi, MIAO Fang, YANG Wen-hui, NI Ya-ting, JIANG Yuan

2022, 0(05):  82-89. 

Asbtract ( 143 )   PDF (1727KB) ( 66 )  

References | Related Articles | Metrics

With the advent of the big data era, the types and scale of structured data continue to increase, but there are currently no research results aimed at the registration of structured data. In order to solve the problem of structured data registration, adopting the idea of data architecture (DA) and related technologies, combined with the data registration center (DRC),a registration engine for structured data is designed, anda unified registration standard for structured data and registration method are proposed to realize the automatic collection and registration of structured data. Through experiments and analysis, the registration engine can accurately and efficiently collect and write structured data registration information into the DRC, and propose solutions to the registration problems of commonly used databases at home and abroad. It lays a solid foundation for the structured data registration information management and application of the DRC data registration center.

A Point Cloud Registration Algorithm Combining Improved PSO Algorithm and TrICP Algorithm

LIANG Zheng-you , , WANG Lu , LI Xuan-ang , YANG Feng ,

2022, 0(05):  90-95. 

Asbtract ( 531 )   PDF (1109KB) ( 123 )  

References | Related Articles | Metrics

Aiming at the problem that the traditional iterative closest point (ICP) algorithm is easy to fall into the problem of local optimality when the initial spatial position deviation is large, a point cloud registration method combining improved PSO-TrICP algorithm is proposed. Firstly, the traditional particle swarm optimization (PSO) algorithm is improved by introducing similarity measurement criterion of fitness to adjust the updating mode of particles. Then, the mean value of the historical global optimal solution of each iteration is added as a new learning factor to avoid the phenomenon of “precocity”; Secondly, the rigid transformation parameters and the overlap rate between the point clouds are used to form the particles, and the improved PSO algorithm is used to provide a good initial relative position; Finally, the space transformation between point clouds is estimated with trimmed iterative closest point (TrICP) algorithm. Experimental results show that the improved PSO-TRICP algorithm has better registration accuracy and operation efficiency than the similar registration algorithms proposed in recent years, and has better robustness.

An Improved Multispectral Reconstruction for Sparse-view CB-XLCT Imaging

ZHANG Wen-yuan, HAI Lin-qi, LIU Ying-jie, ZHANG Hai-bo

2022, 0(05):  96-101. 

Asbtract ( 223 )   PDF (3827KB) ( 67 )  

References | Related Articles | Metrics

Sparse-view Cone-beam X-ray luminescence computed tomography (CB-XLCT) is a novel multimode optical tomography technique, which has shown good potential for real-time detection of early tumor. However, the inverse problem of sparse-view CB-XLCT is more serious than that of traditional multi-angle CB-XLCT, because of the limited projective data. Aiming at above problem, this paper proposes a spectral regression (SR) and preconditioning method to improve multispectral reconstruction for sparse-view CB-XLCT. Firstly, the multispectral strategy is used to construct the system matrix and model the inverse problem; Then, the spectral regression method is used to learn the features of the high-dimensional system matrix in the previous inverse problem; After that, a preconditioning approach is used to reduce the coherence of the new system matrix, and further form a new inverse problem. Numerical simulations and robustness tests were performed to verify the effectiveness and robustness of the proposed method. The experimental results indicated that our proposed method not only can significantly improve the imaging quality of multispectral reconstruction for sparse-view CB-XLCT, but also has good robutsness.

A Seal Image Verification System for Identification and Traceability

CAI Yi-ni, CHEN Zheng-ming, NI Jia-jia

2022, 0(05):  102-107. 

Asbtract ( 371 )   PDF (2470KB) ( 79 )  

References | Related Articles | Metrics

Aiming at the problems existing in the management of modern corporate seals and the verification of seal images, a management and verification system for corporate seal images is designed, which provides multi-faceted, full-process auxiliary management and identification of seals. The seal verification system uses a seal type recognition method based on ResNet, which simplifies the process of feature extraction and has a good recognition accuracy. Then a method of tracing seal images based on siamese neural network is introduced, which can be traced back to a specific seal image, so as to determine the authenticity of the seal. Also, a management and control system for smart seals has been established, through which the smart seals are controlled and the stamping behaviors are recorded. Experiments have proved that the system can meet the mobile terminalv’s requirements for management, control, identification and traceability of smart seals, and provides good user experience.

Night Vehicle Detection Algorithm Based on CNN

ZHANG Wen-li, XU Li, LIU Xing-xing

2022, 0(05):  108-113. 

Asbtract ( 326 )   PDF (2091KB) ( 75 )  

References | Related Articles | Metrics

Aiming at the accuracy requirements of the night vehicle detection model, this paper proposes the night vehicle as the reasearch object and uses convolution neural network in deep learning to construct the detection model. Firstly, the data set is processed with white balance to reduce the interference of street lamp color and enhance the image quality, and mosaic data enhancement is used to enrich the detection data set and improve the detection effect of the model for small target vehicles; Secondly, K-means+〖KG-*3〗+ algorithm is used to select the prior box, and the intersection and union ratio distance is used to cluster the prior box; Then the attention mechanism module is added to the backbone feature extraction network to enhance the channel and spatial feature information of the target in the residual structure feature map; Finally, the gradient equilibrium mechanism is introduced into the original confidence cross entropy loss of the loss function to make the model attenuate the hard and easy samples effectively. Through the experiments and comparative analysis on UA-DETRAC data set, it can be seen that the accuracy of the proposed algorithm can reach 99.24%, and the number of frames per second to process image can reach 19.

Multi-step Prediction Method of Vehicle Trajectory for Eliminating Communication Delay

QI Zhan-shuo, GAO Yan-dong

2022, 0(05):  114-118. 

Asbtract ( 185 )   PDF (1192KB) ( 71 )  

References | Related Articles | Metrics

In the current vehicle-road collaborative test environment, systems with real-time characteristics are usually used. In order to solve the communication delay problem that easily occurs in the real-time system of vehicle-road collaborative test, this paper proposes a multi-step predicion method of vehicle trajectory oriented to eliminate communication delay. Through the construction of LSTM neural network model, the high-frequency sampling sequence is split and reorganized to establish a new sequence, and the difference sequence of different intervals is input one by one. After single-point prediction of trajectory points under each sequence, the vehicle trajectory for a certain distance in the future is formed, and then the multi-step prediction of the vehicle trajectory is realized. The experimental results show that the multi-step trajectory forecasting method proposed in this paper can eliminate 93.94% of the communication and system delays, and the multi-step trajectory prediction reduces the MSE growth rate by 7.47 percentage points at medium and long distances compared to the single-step trajectory prediction, which has good time delay elimination characteristic and error control ability.

Reverse Overtaking Control Algorithm for Autonomous Vehicles

RUAN Shi-feng, HUI Fei, YU Jian-you, ZHANG Zhi-gang, DU Yi-ru, GUO Xing

2022, 0(05):  119-126. 

Asbtract ( 237 )   PDF (2351KB) ( 62 )  

References | Related Articles | Metrics

In order to solve the problem that two-way lanes are limited by road conditions and traffic characteristics, a reverse overtaking control strategy based on graph search and model predictive control (MPC) is proposed. The strategy obtains global information such as speed and acceleration of the environment vehicles with the help of telematics and in-vehicle sensors, and incorporate the impact of each entity in the multi-vehicle scenario into the overtaking decision. Firstly, based on the global information obtained from vehicle-vehicle communication, combined with a non-cooperative game, each vehicle is predicted within the action of the entire time period, and each area of the road is evaluated for safety based on the prediction, and the evaluation is based on the probability of a vehicle appearing in that area at the next moment. After completing the assessment of the road, the collision probability hot zone map is obtained, and then the safe path is searched by the A* algorithm, and the trajectory planning of the main vehicle is completed according to the safe path. After that, the model prediction controller is designed to control the main vehicle in real time so that the vehicle follows the established trajectory. Finally, the proposed algorithm is verified by building a joint simulation platform with the help of Carsim and MATLAB/Simulink. The simulation test results show that the maximum control error of the model does not exceed 0.15 m, and the average error rate is about 1.7%, which can realize the accurate control of the vehicle and ensure the controlled vehicle to complete the reverse overtaking safely.