Computer and Modernization

LC-Raft: A Consensus Algorithm Based on Calculated Value of Historical Log

MA Bo-tao, NI Hong, ZHU Xiao-yong

2020, 0(12): 1-8.

Asbtract ( 265 )

PDF (1393KB) ( 230 )

References | Related Articles | Metrics

Raft consensus algorithm has been widely used in the industry because of its clear principle and easy implementation. However, as a simplified version of the Paxos-like solution, the Raft algorithm sacrifices part of its performance. In different specific application scenarios, some aspects can be improved according to the actual application. For a distributed system composed by devices with poor stability, it will go through multiple election processes in the working cycle. In each election process, there is a small probability that the election operation based on Raft algorithm will experience multiple timeout elections. In order to reduce the time-consuming of the election process in extreme cases, this paper designs an improved version of the consensus algorithm LC-Raft. By counting the number of failures in the history log, a set of node stability evaluation indicators are constructed, and the election process is modified. Based on these above steps, the algorithm is able to realize that the election process can be completed before timeout when the system network is unblocked. At the same time, based on the Docker container engine, a series of simulation experiments are designed to realize the election process of different node sizes. Multiple statistical data can verify the good performance of the algorithm in the election process.

Design and Analysis of Reliable Link Scheduling Algorithm in Wireless Network

ZHANG Xin, ZHANG Xu, ZHANG Ru-wen, YU Qi-qi, WANG Yu-xia, WEI Ying-xin, HUANG Bao-gui

2020, 0(12): 9-12.

Asbtract ( 163 )

PDF (793KB) ( 112 )

References | Related Articles | Metrics

There is much uncertain interference in the wireless signal transmission process, which causes the quality of the signal to seriously degrade when it reaches the receiver. The receiver cannot correctly decode the received signal and communication error occurs. At present, low-latency link scheduling algorithms based on the SINR (Signal to Interference plus Noise Ratio) interference model are effective methods to improve the reliability and communication capacity of the wireless network. In this paper, a shortest link scheduling algorithm with an approximate ratio of O(log Δ) is proposed (Δ is the ratio of the longest link length to the shortest link length). All links adopt the uniform power assignment. First, the link set is divided into subsets according to the link length. Then, using TDMA operation mechanism, the interfering links in each subset is assigned different transmission time slots so that the links in each time slot can communicate simultaneously. The theory proves the correctness and effectiveness of the algorithm in this paper.

Recommendation System Based on Heterogeneous Information Network

CUI Xin

2020, 0(12): 13-19.

Asbtract ( 151 )

PDF (946KB) ( 131 )

References | Related Articles | Metrics

With the development of Internet, computer and other technologies, Internet has brought various network services for users to enhance communication among users. Among them, the community question answering provides users with a communication platform for questions and answers, the purpose of which is to achieve knowledge and experience sharing and information dissemination among users through Internet. However, there are still some problems that limit the development of the community question answering. For example, as the number of users continues to increase, a large number of questions cannot be answered in time and the questioners are not satisfied with the answers to existing questions. Therefore, for the question and answer community, how to find experts from a large number of users is very important. In response to the above problems, this paper proposed a recommendation method based on heterogeneous information network. Firstly, a heterogeneous information network was established for the question and user attributes in the question and answer community, and meta-paths were used to capture the rich semantic information in the heterogeneous information network. Secondly, the similarity calculation method based on the meta paths was used to calculate the similarity matrix between the question and the user, and three methods were used to fuse the obtained similarity matrix with the question-user scoring matrix. In the end, the matrix decomposition was used to obtain the potential features of the question and the user. Factorization machine is used for training and recommendation. The method proposed in this paper was compared with various advanced recommendation algorithms on Haichuan chemical community question answering data set, and the evaluation index was used to evaluate the model. Experimental results show that the algorithm proposed in this paper has certain advantages over the previous algorithms in terms of relevant evaluation indicators.

Web Service Composition Based on K-medoids Point Optimization Algorithm in Big Data Environment

LIU Feng, ZOU Chen-song, CUI Wei

2020, 0(12): 20-24.

Asbtract ( 105 )

PDF (728KB) ( 92 )

References | Related Articles | Metrics

Based on the current increase in the mass of Web services, the inefficiency of existing Web service selection algorithms, and poor user matching, this paper proposes a solution to the problems of particle shift, low accuracy, and easy distortion in the K-medoids point algorithm. Research on Web service composition based on K-medoids point optimization algorithm is given in big data environment. The method is based on the study of Web service selection and optimal Web service combination based on the K-medoids point algorithm that optimizes the initial clustering center based on the satisfaction of different user needs and the QoS parameters of Web services in a big data environment. At the same time, the accuracy of dynamic selection and combination of services, the number of iteration updates, the selection time of candidate sets and the total selection time are experimentally analyzed for different selection methods, which verifies the effectiveness and reliability of the method in this paper.

Using Genetic Algorithm for Virtual Machine Placement Optimization

XU Sheng-chao

2020, 0(12): 25-31.

Asbtract ( 173 )

PDF (1190KB) ( 141 )

References | Related Articles | Metrics

A genetic algorithm based approach for virtual machine placement called GA-VMP was proposed. In GA-VMP, local regression robust (LRR) algorithm was adopted to identify critical hosts in the physical host status detection procedure; minimum migration time (MMT) policy was also used for selecting VMs on critical hosts to be migrated. In the virtual placement, genetic algorithm was used to find a near-optimal solution and thus formed a new virtual machine migration model called LRR-MMT-GA. The mathematic model for energy consumption in GA-VMP was also designed and the minimized energy consumption was used as the objective function in genetic algorithm. The experimental results and performance analysis show our strategy leads to a further improvement in energy consumption and virtual machine migration. Our strategy is helpful for other cloud providers to build a low energy consumption cloud data center.

Dynamic Routing of Routing Frequency Based on Ant Colony Algorithm Optimization

GAO Yue-ren, WU Tao, GAO Xia

2020, 0(12): 32-37.

Asbtract ( 111 )

PDF (1890KB) ( 76 )

References | Related Articles | Metrics

In the conventional routing frequency dynamic path selection method, the set obstacle avoidance rules are missing, resulting in the route transmission path is not optimal, and the path selection time is long. Therefore, based on ant colony algorithm, the dynamic path selection method of routing frequency is optimized. The grid simulation transmission environment is adopted to set key information selection rules. According to the rule of route frequency change and input guiding factors, the setting of route movement rules is realized. The overall situation is taken as a whole, ant perception is calculated according to ant colony algorithm, and the global obstacle avoidance rules are set based on the local area obstacle avoidance rules. The initial pheromone and pheromone during ant search are calculated by compensating the pheromone concentration, the optimal solution of the route is obtained. The experimental results show that compared with the conventional path selection method, the proposed path selection method takes into account the overall situation, and the route obtained is superior to the conventional method, and the path selection time is the shortest. It can be seen that the path selection method based on ant colony algorithm has achieved the purpose of this study.

Research and Practice of Constructing Smart Hospital Based on Improved Clustering Algorithm

CAO Lei, LIU Qiang, YAO Hui

2020, 0(12): 38-42.

Asbtract ( 111 )

PDF (1686KB) ( 105 )

References | Related Articles | Metrics

In order to solve the problem of the lack of intelligent application in hospital, this paper puts forward a kind of intelligent application related to hospital aided diagnosis and treatment. This algorithm combines an improved genetic algorithm and network center mathematical model to optimize the initial center. Firstly, an improved genetic algorithm is used to obtain the approximate optimal clustering number k of document set, and then the network center and center of gravity mathematical model are used to obtain the optimized initial clustering center, which effectively solves the sensitivity of the algorithm to the initial clustering center, and achieves good experiments result. In the practical application stage, different application rule models are designed and formulated in combination with different medical business scenarios, and the operation effect of the algorithm is tested through practical applications such as intelligent inspection appointment timeliness analysis, intelligent evaluation of blood transfusion quality, surgical risk prediction analysis and auxiliary diagnosis recommendation, and good application results are achieved.

A Supporting Platform of Board Game Teaching for Artificial Intelligence Courses

ZHAO Cai-rong, FU Jia-yue, WEI Zhi-hua, DING Zhi-jun, MIAO Duo-qian

2020, 0(12): 43-48.

Asbtract ( 255 )

PDF (2029KB) ( 86 )

References | Related Articles | Metrics

Aiming at the problems such as too theoretical, complex and abstract of game algorithms in artificial intelligence courses, this paper designs and implements a board game teaching platform with relatively complete functions and user-interface friendly to assist in the learning of game algorithms in AI courses. The platform allows users to upload the game programs, and supports a wide range of games between game programs, between humans and game programs, and between humans. This paper expounds the overall design, the main function modules and the core building technologies of the board game teaching platform. The design of a board game teaching platform helps to develop students’ ability to solve practical problems based on their theoretical knowledge, and to encourage students to continuously improve computer game programs and AI methods, so as to achieve better teaching results. This paper further evaluates the actual use and practical effect of board game teaching platform in artificial intelligence courses.

Train Equipment Fault Prediction Model Based on Machine Learning

YUAN Jiao, WANG Xun, PAN Zhao-ma, YANG Xue-feng, ZOU Wen-lu

2020, 0(12): 49-54.

Asbtract ( 201 )

PDF (1108KB) ( 108 )

References | Related Articles | Metrics

Decision trees are widely used as predictive models in the field of machine learning and data mining, and their output is easy to understand and explain. The onboard equipment of high-speed railway has problems such as large streaming data, complicated equipment failure and low diagnostic efficiency. According to the characteristics, the CVFDT decision tree algorithm is proposed to build an intelligent fault prediction model for vehicle equipment (low probability, high probability and failure) by machine learning of the normalized column control device stream data. It becomes “pre-exclusion” of potential equipment failures, improving fault classification accuracy, positioning and diagnostic accuracy, and ensuring high-speed railway operation safety and transportation efficiency.

Analysis on Block-chain and Its Application in Energy Industry Based on Knowledge Map

ZHANG Yan, YANG Fang, YANG Lei, HAN Kui-guo, LI Hui

2020, 0(12): 55-60.

Asbtract ( 160 )

PDF (3470KB) ( 90 )

References | Related Articles | Metrics

The paper conducts an objective and comprehensive analysis of the international research status， research hotspots and development trends of block-chain based on visual analysis software， and provides useful guidance and reference for the research of block-chain. The Web of Science database was used as the search object, and the articles related to the block-chain published from 2000 to 2020 were included. Visualizing analysis of the authors, institutions, countries, key words and cited documents was conducted based on bibliometric analysis and information visualization software CiteSpace. A total of 4349 articles were eventually included, and the number of a publication increased very rapidly after 2016. China, USA, England, Australia and many other countries are building their global technology research collaboration network. The Chinese researchers are active in this field and have concurred several high-productive authors, whom entered the worldwide top 10 based on the numbers of articles. The application of block-chain in energy industry is in early stage and is heating up quickly. There are 3 articles related with application in energy industry out of the top 10 citation records. The hotspots include the application of block-chain in electric vehicles, micro-grid, energy internet and electricity market. The study of blockchain is still under development, which has strong potential. The global collaboration is also under development and China needs to enhance the influence.

Analysis and Prediction of Training Effects of National Disaster Life Support Course with Machine Learning Methods

GUO Xin, CHEN Ying, ZHANG Ming-huan, ZHANG Xuan, PAN Shu-ming, TANG Lu-jia

2020, 0(12): 61-66.

Asbtract ( 105 )

PDF (1638KB) ( 79 )

References | Related Articles | Metrics

The outbreak of novel coronavirus (2019-nCoV) pneumonia occurred at the end of 2019. The epidemic situation makes us more aware of the importance of emergency rescue team construction and related personnel training. On the basis of participants in national disaster life support (NDLS) training data from 2018 to 2019, this paper is to study influence factors of NDLS training effects with machine learning method with a view to assisting organizations in evaluating the training effects， using Apriori and decision tree algorithm to build the model. Through data preprocessing, 14 fields are selected as model input fields, and test scores are selected as model output field. Firstly, Apriori algorithm is used to find out several factors that have a great influence on the training effect. Then the decision tree model is used to predict the training effects, and the results of decision tree are used to verify the conclusion of Apriori algorithm. The parameters of confidence, support and promotion are used as the evaluation indexes of Apriori algorithm. Ten fold cross validation is used as the evaluation method of decision tree model. The results show that the model is effective. Conclusion of this study can help the training organizations to effectively predict the learning effects of the trainees; in addition, it can monitor and ensure the quality of training.

Subspace Clustering Method for Analysis of Four-diagnoses of Traditional Chinese Medicine

XU Li-hui, CHEN Min, WANG Chi-she

2020, 0(12): 67-71.

Asbtract ( 104 )

PDF (801KB) ( 77 )

References | Related Articles | Metrics

The analysis of the four-diagnosis of traditional Chinese medicine is an important part of the analysis of TCM syndromes based on the information of the four-diagnosis. Constructing an effective analysis model of the four-diagnosis of TCM can better mine the correlation between TCM syndromes and provide decision support for the clinical of TCM. In this paper, through the analysis of the CLIQUE algorithm of subspace clustering, combined with the data characteristics of the four-diagnosis information, an improved CLIQUE algorithm (ChM-CLIQUE) based on limited space search strategy is proposed. By optimizing the search strategy of the CLIQUE algorithm and performing a depth-first search centered on the cell with the largest grid density among dense cells, the cluster clusters are generated to improve the performance of the algorithm, and introducing grid adaptation density based on the characteristics of the sample Gaussian distribution in the cluster clusters, the recognition accuracy of cluster boundaries is enhanced. In the experiment, multiple sets of comparative experiments were carried out on the data set collected in the clinical medicine of traditional Chinese medicine. The experimental results show that the contour coefficients of the algorithm in this paper are significantly improved by 12.6% and 19.3% respectively compared with the CLIQUE algorithm.

An Efficient Entity Identification Method for Electric Bidding Documents Based on Conditional Random Field

SHAO Shi-yun, ZHOU Yu, YANG Lei, ZHONG Mao-sheng, DAI Rui, ZHAO Jia-le

2020, 0(12): 72-77.

Asbtract ( 125 )

PDF (862KB) ( 89 )

References | Related Articles | Metrics

In recent years, with the rapid development of the national economy, the investment in power construction projects has increased rapidly. Both the number of the associated tenders and the corresponding workload of evaluation have soared. The conventional manual method for assessment is time-consuming, costly, and inefficient. For improving the efficiency of the bid review and reducing the related costs, it is ideal to take advantage of automatic or semi-automatic analysis. Among the adoption of machine-assisted, the entity identification in the tender text, definitely, plays an essential role in information extraction and text summarization. Since there are many complex and a hybrid combination of words in text like location names, the existing recognition technology does not perform well. In this paper, we propose an application-friendly critical information extraction method based on the conditional random fields (CRF), which realizes the automatic and rapid processing of tenders and accelerates the re-assessment of various engineering construction projects and data sharing. Our proposed mechanism has got an experimental verification of efficiency. It has been employed to the automatic transactions in the power sector.

Application of Improved Spark System Based on ANN in Big Data Processing of Air Traffic Management#br#

PAN Wei-jun, LIU Hao-chen, WANG Run-dong, HU Bo-wen

2020, 0(12): 78-82.

Asbtract ( 168 )

PDF (1002KB) ( 73 )

References | Related Articles | Metrics

A new method based on artificial neural network (ANN) was proposed to automatically adjust the configuration parameters of the Spark system to improve the performance of the Spark system in processing ATM big data and solving time-consuming and inefficient issues. The Dell PowerEdge T430 server was used to test 5 commonly used datasets of different sizes in air traffic control big data to verify the method. Research shows that compared with the default parameter configuration, this method can improve the performance of the Spark system by about 35% on average. As the size of the dataset increases, the performance shows a trend of further improvement. This method can effectively guarantee the parameter adjustment efficiency of the Spark system and achieve the purpose of efficiently processing the ATM big data.

Multi-modal Music Emotion Classification Based on Optimized Residual Network

LI Xiao-shuang, HAN Li-xin, LI Jing-xian, ZHOU Jing-wei

2020, 0(12): 83-89.

Asbtract ( 173 )

PDF (2199KB) ( 324 )

References | Related Articles | Metrics

Aiming at the problems of traditional music sentiment classification due to the difficulty of feature extraction, the model classification accuracy is not high and the manual workload is large, this paper proposes a multi-modal music sentiment classification method based on an optimized deep residual network. This method first uses multi-modal translation to convert difficult-to-extract feature music audio modalities into easy-to-operate image modalities; at the same time, based on the deep residual network, the convolution kernel size of the network input layer and the speed of the residual block, the connection has been optimized and improved, which reduces the information loss and shortens the calculation time. In addition, in order to alleviate the shortcomings of Softmax classifiers such as intra-class dispersion and inter-class aggregation, this paper introduces a variant of the Center loss function to improve the Softmax classification function performance. The experimental results prove the effectiveness and robustness of the optimized residual network model in this paper. Compared with the original residual network, the accuracy rate of music emotion classification is improved by 4.27 percentage points.

Graph Algorithm Virtual Simulation System Based on Lushan 3D Scene

LIU Jia-xin, YOU Zhen, HUANG Jie-wen, CHEN Jia-xiang, HU Hong-wen,

2020, 0(12): 90-98.

Asbtract ( 112 )

PDF (4240KB) ( 78 )

References | Related Articles | Metrics

In order to deal with the problems of strong logic, high abstraction, disconnection between teaching and experiment, and lack of interactivity in traditional computer algorithm teaching, based on the three-dimensional scene of a well-known Lushan tourist attraction in Jiangxi province, a graph algorithm virtual simulation system is designed using Unity3D software engine by using virtual reality technology. The system realizes simulation experiment of 5 kinds of graph algorithms. Each graph-algorithm experiment provides 2 operating modes, including “automatic display” and “user interaction”. It also offers the function of entering scenic spots and viewing the Lushan landscape from the user-controlled perspective. Meanwhile, some theoretical problems of the virtual simulation system are discussed, and the key technology and implementation plan to solve these problems are given. Finally, the practicability and flexibility of the virtual simulation system were verified by a case study of Prim minimum spanning tree algorithm. Compared with traditional algorithm teaching and personalized problem-driven teaching methods, the graph algorithm virtual simulation system designed in this paper is more interesting, interactive, and immersive. It not only stimulates students’ exploration and initiative in learning, but also provides a new teaching and experimental method for algorithm and data structure course.

A Packing Counting Method of Medical Plastic Bottles Based on Improved RetinaNet

QIU Lyu, REN De-jun, GAO Ming, FU Lei, WU Hua-yun, HU Yun-qi

2020, 0(12): 99-103.

Asbtract ( 116 )

PDF (3087KB) ( 89 )

References | Related Articles | Metrics

In order to improve the efficiency, accuracy and stability of packing counting in medical plastic packaging production line, this paper proposes a packing counting detection algorithm based on deep learning which can realize automatic online counting. Firstly, an improved RetinaNet network was constructed with ResNet as the framework, the feature pyramid network was used to generate multi-scale feature maps, and the convolution layers were cut appropriately. Then, clustering algorithm is used to optimize the Anchor size, so that the algorithm can adapt to count detection under abnormal conditions such as crooked bottle and inverted bottle, so as to reduce the missed detection rate and improve the positioning accuracy. Finally, the experimental evaluation of the algorithm on the actual packing data set shows that the algorithm is robust and reliable, and can quickly and accurately count and detect the packing plastic bottles under the production conditions. The counting accuracy can reach more than 99.98%, and the single detection time is 33 ms, which meets the real-time detection requirements of the production line.

Optimization Reconstruction of EPMA Image Based on SWOMP Algorithm

JIN An-an, DING Shuang-shuang, XIONG Qing-zhi, XU Ting-ting, XI Shu-ping, MA Ji-zhong

2020, 0(12): 104-111.

Asbtract ( 153 )

PDF (5523KB) ( 112 )

References | Related Articles | Metrics

Compressed sensing has been developed for many years, and there are many reconstruction algorithms. The stagewise weak orthogonal matching pursuit (SWOMP) algorithm, which does not require sparsity, is an improved algorithm. The measurement matrix selects Gaussian matrix, but its reconstruction effect is not ideal. Aiming at the shortcomings of the algorithm, the algorithm is optimized by combining the electron probe image. This optimization takes full advantage of the Fourier matrix and adjusts the number of iterations and threshold parameters. Firstly, the commonly used matrix is tested several times to find the best quality measurement matrix—Fourier orthogonal matrix. Secondly, the iteration number and threshold are modified to find the best parameter matching to improve the reconstruction quality of the algorithm. The experimental results show that the proposed method has better reconstruction effect on the electron probe image and achieves the super-resolution recovery requirement. The reconstructed image quality is higher than the original one.

Infrared Dim Small Target Detection Based on Wavelet Packet Transform

FENG Yang

2020, 0(12): 112-115.

Asbtract ( 187 )

PDF (1889KB) ( 369 )

References | Related Articles | Metrics

For the detection of infrared dim small targets under complex background, an infrared dim small targets detection algorithm based on wavelet packet transform is proposed. Firstly, the wavelet packet transform is used to decompose the infrared image with small and weak targets to obtain the high and low frequency node coefficients at different scales. Secondly, according to the different contributions to the target energy during the reconstruction of different node coefficients, the frequency band node coefficients in the middle of the energy distribution in the high frequency band are selected to reconstruct the image to complete the background suppression. Finally, the target image after reconstruction is segmented by adaptive threshold segmentation method, and the result of target detection is obtained. In the experiment, multiple infrared sequence images were used to verify the results. The simulation results show that the algorithm can suppress the background and cloud edge well, detect the target signal accurately, and improve the target’s signal-to-clutter ratio and contrast.

Detection Method of One-shot Legend Based on Siamese Neural Networks

WANG Chao-qi, GONG Fa-ming

2020, 0(12): 116-122.

Asbtract ( 134 )

PDF (1625KB) ( 98 )

References | Related Articles | Metrics

In view of the problems such as the difficulty of training, the slow detection and the difficulty of obtaining the training data in the existing deep learning methods, a new solution is proposed for the single sample learning problem. Based on the structure of convolutional neural network, combined with the characteristics of fixed aspect ratio and independence of legend, a new SiameseSSD detection frame is used for target detection. The framework includes a siamese subnet for feature extraction and an improved SSD subnet for classification and regression. At the same time, we use the data enhancement technology to expand the sample, then make the data set and train the model and adjust network structure and detection method to detect large-resolution construction drawings. The experimental results of this method on the construction drawing data set show that this method is a new method to solve the single sample learning task, with an accuracy of 91.3%, the detection speed reached 61 fps. Compared with the existing top level, it has certain advantages and meets the actual work needs.

Table of Content