Computer and Modernization

Information Service Performance Modeling Based on Laguerre FNN

SONG Xin , FAN Zhi-qiang ,

2021, 0(03): 1-6.

Asbtract ( 179 )

PDF (960KB) ( 143 )

References | Related Articles | Metrics

System architecture is the design and blueprint of information system. System architecture simulation can estimate in advance whether the built system can meet the design expectation, but the current mainstream architecture simulation method lacks the means to accurately predict the service performance attribute in the early stage of the system life cycle. Aiming at this problem, this paper takes information service performance modeling as the research goal. Firstly, it classifies information service at three levels according to its functions, and provides the factors that define and influence the performance attributes of each type of information service. Furthermore, an information service performance modeling method based on Laguerre forward neural network is proposed, and the advantages and reasons of Laguerre forward neural network are analysised. Finally, taking information retrieval service as an example, six factors affecting the number of machine cycles required for the execution of information retrieval service are proposed. Ablation experiments and comparative experiments are used to verify the feasibility of the information service performance modeling method proposed in this paper.

Adaptive Traffic Control Method of IDC Network Based on Envelope Feature

LIANG Yun-de, CHEN Shou-ming, LU Yan-qian, LI Xue-wu, YU Shun-huai

2021, 0(03): 7-11.

Asbtract ( 177 )

PDF (1237KB) ( 102 )

References | Related Articles | Metrics

The traditional traffic control method can not automatically adjust the constraint rules, which leads to a large bandwidth loss rate when facing different traffic density. Therefore, based on the envelope feature, a new adaptive traffic control method for IDC network is studied. This method constructs the IDC network flow energy consumption model, extracts the envelope characteristics, obtains the flow spatial distribution state and sets the regulation adaptive constraint rules according to the network internal information flow direction, and applies the BP neural network to improve realization of the IDC network adaptive flow regulation. Experimental results show that under the conditions of random20, random40 and random60, the throughput and bandwidth loss rate of the proposed method are high, and it is suitable to control the adaptive traffic of IDC network.

Load Forecasting Method of Ultra Short-Term Based on Integrated Model

WEI Jian , ZHAO Hong-tao , LIU Dun-nan , JIA He-ping

2021, 0(03): 12-17.

Asbtract ( 154 )

PDF (1222KB) ( 75 )

References | Related Articles | Metrics

Accurate short-term load forecasting is the key to ensure the smooth operation of power system. After the popularity of machine learning algorithm, it provides algorithm support for short-term and ultra short-term load forecasting which is difficult to solve before. In view of the good effect of the Catboost, the convolutional neural network-long short term memory, extratrees and other integrated models in processing nonlinear correlation data, this paper combines the above three methods to build an integrated prediction model, uses BP neural network to determine the weight coefficient, and combines the advantages of various single prediction models through the weight, so as to achieve a better prediction effect. In order to better illustrate the advantages of the method used in this paper, Mean Absolute Percentage Error (MAPE), root mean square error, mean square error, goodness of fit are used as measurement indexes. Compared with each single prediction model, the integrated model decreased by 1.01 percentage points, 0.94 percentage points and 1.19 percentage points respectively compared with Catboost, CNN-LSTM and Extratrees models in MAPE standard.

Multi-node Failure Repair Algorithm Based on Erasure Code

XU Jia-bing, ZHU Hao-chen, YANG Li

2021, 0(03): 18-23.

Asbtract ( 142 )

PDF (799KB) ( 84 )

References | Related Articles | Metrics

As an important data fault-tolerance technology in a distributed system, the correction and deletion of code are widely used in the field of repair of invalid data. However, most of the existing correction and deletion algorithm aim at single-node repair, which has a high repair cost and does not consider the information transmission between the new nodes, bringing inconvenience to the repair of multiple-fail nodes. Based on this, a multi-node failure repair algorithm based on correction code is proposed, which uses node selection strategy to select the central node as the root node in the new node. The maximum repair tree is constructed with the supply node and the remaining new node according to the link bandwidth, so as to reduce the data repair time. The experimental results show that, compared with the existing BHS and SSR serial repair methods, the algorithm can effectively improve the repair efficiency of multi-failure nodes, which verifies the validity of the proposed algorithm.

Short Term Load Forecasting Algorithm Based on Improved Grey Theory

XU Hui-jun , WANG Zong-yao , LI Zhong-qing , LI Peng-fei , ZHOU Xin-kang

2021, 0(03): 24-27.

Asbtract ( 119 )

PDF (667KB) ( 81 )

References | Related Articles | Metrics

Based on the analysis of the traditional grey load forecasting theory and GM (1, n) model, the GM (1, n) model is improved again to solve the problem of large error of forecasting results due to the strict requirements of the traditional grey theory on the original data. The experiment shows that the load forecasting model established by this method has a great improvement in the forecasting accuracy, which will be helpful for the future grey theory.

Optimal Algorithm of Remaining Space for 3D Packing Problem of Paperboard

WANG Cheng, CHEN Zheng-ming, LYU Jia

2021, 0(03): 28-34.

Asbtract ( 207 )

PDF (1331KB) ( 84 )

References | Related Articles | Metrics

Aiming at the various practical constraints encountered in the packing process of corrugated board, a fast algorithm based on the optimal remaining space and various practical constraints is proposed. Firstly, according to the first in, last out and combination of loading constraints, the packing sequence of paperboard is determined. Then, the three-dimensional packing problem is transformed into a two-dimensional packing problem with height constraints. Based on the optimal strategy of remaining space, the space partition mode and cardboard placement mode are selected, and the remaining space is merged and repartitioned, so as to obtain the result of cardboard loading and placement and achive the goal of the highest utilization rate of container space and the minimum number of container space. Through the calculation of random and practical examples, as well as the three-dimensional visualization of the results, it is proved that the algorithm can achieve a variety of constraints, high space utilization and high operation efficiency. The effectiveness and practicability of the method are verified.

Optimization and Analysis of Underwater LEER Protocol Based on NS3

MA Yuan , DU Xiu-juan

2021, 0(03): 35-40.

Asbtract ( 196 )

PDF (1403KB) ( 87 )

References | Related Articles | Metrics

Underwater wireless sensor network, as the application extension of wireless sensor network in underwater, has become a hot research topic. The problems of difficult energy replenishment, poor energy balance and short network lifetime in the underwater communications have not been solved. In order to improve the data delivery rate and prolong the lifetime of underwater wireless sensor network, it is very important to design a routing protocol suitable for dynamic changes of underwater topology and high energy balance. The Layer-based and Energy-Efficient Routing (LEER) protocol effectively solves the problem of routing packets to a void area. The optimized LEER Protocol adopts multi-sink topology. From the aspects of single-hop delay, residual energy, node density, and subsequent updating of the level of underwater nodes when they move, the energy balance optimization strategy of the underwater LEER protocol based on NS3 is proposed. Simulation results show that the optimized LEER protocol is superior to the LEER protocol in terms of energy balance, data delivery rate and adaptive dynamic topology.

Single Bid Trading Mode Based on Security Performance

CHE Jia-li, REN Jun-ling

2021, 0(03): 41-45.

Asbtract ( 132 )

PDF (738KB) ( 120 )

References | Related Articles | Metrics

This paper analyzes the data security and privacy protection in commodity trading from the perspective of security performance. Taking the single bidding transaction model as an example, the blind signature technology is used to authenticate the users identity without relying on the credibility of the third party authentication agency, it realized the secret of the users identity information. The Bit commitment agreement is used during the transaction to guarantee the non-repudiation of both parties. The result of security analysis shows that the scheme can effectively protect the security of the transaction process even if the third party is not trusted.

Two-way Security Authentication RFID Protocol Based on Random Hash Chain

ZHOU Jing , , DONG Guo-chao , , DENG Zu-qiang , , ZHANG Jin-luan , , LIU Chao , , NIU Yong-liang

2021, 0(03): 46-50.

Asbtract ( 433 )

PDF (694KB) ( 235 )

References | Related Articles | Metrics

Radio Frequency IDentification (RFID) technology is widely used in the field of Internet of Things. But the RFID system is easy to be stolen and forged during the information transmission process, RFID system has exposed serious security problems. Aiming at the problem of counterfeit attacks in the current mainstream Hash chain protocol cluster, a two-way security authentication protocol based on random Hash chain is proposed. In the process of mutual authentication, the reader verifies the validity of the response message by identifying the value of the random number K, the tag detects the legitimacy of the reader by detecting the tag attributes of the returned message, the reader prevents the label from being counterfeited by detecting the preset identity of the label and ensure the authenticity of data in the system. It can effectively improve the systems forward security, anti-counterfeiting and anti-location tracking and other security features. Finally, this paper uses BAN logic to perform security verification on the proposed protocol, the experimental results show that there is the security of mutual authentication between the tag and the reader.

KM-SSD Method for Vehicle and Pedestrian Detection

ZHENG Qin-hao , YANG Zhen , YANG Zhen

2021, 0(03): 51-56.

Asbtract ( 318 )

PDF (3090KB) ( 140 )

References | Related Articles | Metrics

In view of the fact that the conventional improved SSD methods improve the object detection accuracy of SSD while reducing its detection speed, this paper proposes an improved KM-SSD method based on SSD. Firstly, K-means+〖KG-*3〗+ clustering algorithm is used to adaptively learn the ratio of width to height in prior boxes; Secondly, an efficient feature mergence module is designed to achieve high and low level feature information fusion; Finally, the KM-SSD method is verified on the challenging KITTI dataset. The experimental results show that the mAP of SSD is 62.7%, and its average detection time is 0.162 s; the mAP of KM-SSD is 69.8%, and its average detection time is 0.133 s. Consequently, KM-SSD method not only improves the accuracy of SSD in vehicle and pedestrian detection, but also improves the detection speed of SSD, which proves the effectiveness of K-means+〖KG-*3〗+ clustering algorithm and the efficiency of feature fusion method used in this paper.

Social Service Mechanism of Intelligent Family Based on Artificial Intelligence Recommendation Algorithm#br#

WANG Shi-qi

2021, 0(03): 57-62.

Asbtract ( 178 )

PDF (1700KB) ( 80 )

References | Related Articles | Metrics

With the advent of the era of social network, the types of social network gradually cover different user groups. The potential mining of social network needs of different user groups is the current research hotspot. Based on the artificial intelligence recommendation algorithm, this paper constructs the multi-level circle structure of family social network, and proposes an intelligent recommendation algorithm of family and relatives. Based on the intelligent algorithm and the social mode of family, a family social network system that is composed of personal information module, basic function module, family information module, application service module and open API module is designed. The results show that when the amount of data is more than 10000, the execution efficiency of the intelligent recommendation algorithm is about three times that of the non intelligent recommendation algorithm, and the number of cluster nodes has a significant impact on the execution efficiency. In addition, the social system under the intelligent recommendation algorithm is of good effect, which shows that the intelligent recommendation algorithm has high practicability. This research provides the data support of social network service mechanism and verifies the market potential of family social service through the research of data-based recommendation algorithm for family social service and the quantitative verification of algorithm execution efficiency.

Construction Method of Isomorphic Surface Subsidence Based on Akima

PENG Cheng , ZHANG Jin-cai , , CHEN Yong , , LI Bin

2021, 0(03): 63-69.

Asbtract ( 222 )

PDF (2525KB) ( 84 )

References | Related Articles | Metrics

Traditional data analysis platforms are mostly developed based on ideal conditions with sufficient data, which are often far from the reality of incomplete data and are difficult to meet the objective requirements of practical applications. Therefore, effective processing of data missing becomes the key to informatization work. Aiming at the problem of missing data in the development and application of the “Hebei Land Subsidence System”, this paper applies the mature interpolation technology to the construction of the equivalent surface of land subsidence, which is an exploration and answer to the actual problems in the information management and control work. The key to data curve interpolation is to select a suitable interpolation method. Based on the principle of simplicity and practicability, this paper proposes a method for modeling the land subsidence isosurface combining Akima interpolation and GP service. This method is composed of two parts. One is that Akima interpolation is used to effectively solve the problem of discontinuity of the benchmark observation sequence. The second is to relize the automatic construction of the ground subsidence isosurface based on GP service. The system deployed in Hebei and the results of its operation show that the proposed method not only fills in the missing data quickly and effectively, and the isosurface drawn is consistent with the actual production situation, but also achieves the goal of low coupling and easy to use and flexible method based on the micro-service programming idea.

Classification of Motor EEG Signals Based on PCA and PSO-SVM

HUANG Xu-bin, LIANG Shu-jie

2021, 0(03): 70-76.

Asbtract ( 511 )

PDF (4542KB) ( 347 )

References | Related Articles | Metrics

The feature extraction, classification and recognition of electroencephalogram signals of motor imagination are the difficult problems faced by the current Brain Computer Interface (BCI) technology. Aiming at this problem, this paper proposes a classification method of motor imaging EEG signals combining Principal Component Analysis (PCA) and Particle Swarm Optimization optimized-Support Vector Machine (PSO-SVM). Firstly, PCA is used to reduce the dimension of the collected high-dimensional electroencephalogram signal, eliminating the noise components and extracting the feature vectors reflecting the different characteristics of three-dimensional EEG signals. Then SVM is used to classify the feature vectors. In view of the problem that the SVM classification performance is greatly affected by the kernel parameters, the global optimization ability of PSO algorithm is used to optimize the SVM classification performance so as to improve the SVM classification performance. Finally, the Graz data used in the BCI competition is used for experiments. The results show that the proposed PCA fusion PSO-SVM method can obtain 95.3% classification performance, and has a high application prospect.

Extracting Key Elements of Traffic Accident Litigation Cases Based on CRF

GUO Fan-sha, YANG Feng-bao

2021, 0(03): 77-81.

Asbtract ( 184 )

PDF (721KB) ( 91 )

References | Related Articles | Metrics

In order to solve the problems of the relevant personnel in the case, such as miscellaneous handling of litigation cases, scattered information collection, and long time for handling the case, a model for extracting key elements of traffic accident litigation cases is proposed based on Conditional Random Fields (CRF). The model uses information extraction technology to design different feature templates by constructing key element tagging set and building corpus. Fully combining the text characteristics of litigation cases in the field of traffic accidents, considering the window length and the selection and combination of different features, the traffic accident litigation case is trained and tested based on the PyCharm platform. The experimental results show that the optimal feature template can extract the key elements in traffic accident litigation cases with an F1 value of 80.15%, and different word segmentation tools have an impact on the key element identification results. The proposed model is an effective exploration and attempt to give a fair and just judgment result quickly and correctly.

Video Click-through Rate Prediction Model Based on Users Dynamic Interests

YANG Jia-xue, PENG Guo-zheng, HAN Li-xin

2021, 0(03): 82-87.

Asbtract ( 150 )

PDF (1075KB) ( 194 )

References | Related Articles | Metrics

Aiming at the problem that the classic click prediction model cannot capture the users dynamic interest and analyze the characteristics of low-level and high-level interaction, this paper proposes a video click prediction model based on the users dynamic interest. The model first maps the discrete data into low-dimensional continuous vectors that are easy to operate after embedding. In order to capture the users dynamic interest changes, the transformer model is introduced to analyze the video sequence clicked by users and the candidate video to be predicted, and the interaction between the video in the behavior sequence and the video to be recommended is extracted. In order to dig deeper into the implicit feature interaction behind the users click behavior, the DeepFM network is introduced and the network structure is optimized and improved to make the model more suitable for sequence-dependent click data. The experimental results show that the prediction accuracy extension of the model proposed and improved in this paper is better than that of the typical deep decomposition model in click rate prediction, and the release of the transformer mechanism can significantly improve the accuracy of click rate prediction.

A Screening Method of Machine Learning Model for Auxiliary Diagnosis

DENG Zi-yun ,

2021, 0(03): 88-93.

Asbtract ( 123 )

PDF (1629KB) ( 84 )

References | Related Articles | Metrics

To automatically recommend machine learning models for auxiliary diagnosis of diseases to doctors, a screening method of machine learning model is proposed. The screening method consists of three steps including screening machine learning models with training accuracy and testing accuracy, screening machine learning models with precision rates, recall rates and F1 scores, recommending the optimal machine learning model using the total score formula with weights. Taking the auxiliary diagnosis of breast cancer as an example, the support vector machine model （γ=0.5） based on Gaussian RBF (Radial Basis Function) is finally selected among 8 machine learning models and recommended for doctors to use, because in addition to meeting the three conditions of the screening method, the model achieves the highest total score of 0.985.

AC-Rec: Academic Collaborators Recommendation Method Based on Multi-features

SHAO Meng-qiao, JI Shun-hui, ZHANG Peng-cheng

2021, 0(03): 94-100.

Asbtract ( 555 )

PDF (1294KB) ( 218 )

References | Related Articles | Metrics

The rapid increase in the number of researchers on social network platform makes it very difficult to find academic collaborators with similar interests. This paper proposes a recommendation method for academic collaborators based on multi-features under ResearchGate platform. This method combines the three features of paper text similarity, social relevance and self-activity to measure the relationship, and uses the multi-layer perception mechanism to build a recommendation model for top-N recommendation. The text similarity is calculated by the Doc2Vec text depth representation model, and the social relevance is calculated by graph-based random walk algorithm. The experimental results show that AC-Rec is better than the existing academic collaborators recommendation methods based on ResearchGate platform. When N is 30, the hit rate reaches 53.90%, which can effectively recommend potential academic collaborators.

Grid Optimization Algorithm Based on Point Cloud Enhancement

YANG Yu-hang, YANG Yao-quan, LIU Hong-fei

2021, 0(03): 101-107.

Asbtract ( 337 )

PDF (4395KB) ( 327 )

References | Related Articles | Metrics

Aiming at the problems of 3D point cloud using traditional Poisson algorithm for grid reconstruction, the reconstruction time is long and the final model has holes and local details missing, a grid optimization algorithm based on point cloud enhancement is proposed. Firstly, the initial point cloud is denoised by statistical filtering. In order to improve the reconstruction efficiency on the basis of guaranteeing the detailed features, the point cloud hole repair is performed using bicubic spline interpolation while performing appropriate point cloud down sampling through voxel filtering. Then, the moving least squares error function is introduced into the point cloud normal calculation to optimize the quality of the point cloud normal vector. The experimental results show that the optimized gridding algorithm takes less time than the traditional Poisson reconstruction algorithm and improves the accuracy of the reconstructed model to a certain extent.

An Image Description Algorithm Based on Object Detection and Part of Speech Analysis

GAO Yi-fan, WANG Yong

2021, 0(03): 108-114.

Asbtract ( 176 )

PDF (3865KB) ( 98 )

References | Related Articles | Metrics

In this paper, an image description algorithm based on object detection and part of speech analysis is proposed to solve the problem of low correlation between the description content and image in the existing image description method that based on attention mechanism. Based on the attention mechanism, this method extracts information from the picture by the target detection algorithm, and processes it by the recurrent neural network with the attention mechanism to generate the image description statement. In the process of generating words, the algorithm predicts the part of speech of each word, and then selects different neural networks according to the prediction results, so it improves the correlation between the description statement and the original image. The experimental results show that in many objective description evaluation criteria, the description statements generated by the algorithm in this paper have different degrees of improvement compared with the existing algorithms, at the same time, the content of the picture can be more accurately and smoothly described in the subjective evaluation.

Moving Target Tracking Based on Improved Optical Flow Characteristics

LIU Hong-fei, YANG Yao-quan, YANG Yu-hang

2021, 0(03): 115-121.

Asbtract ( 635 )

PDF (3616KB) ( 352 )

References | Related Articles | Metrics

In the city intelligent video monitoring, it is necessary to track the moving object in real time. Aiming at the problems of the traditional moving object detection, for example, the target is easy to lose, low tracking rate and poor real-time performance, a moving object tracking detection method based on the improved optical flow characteristics is proposed to track the moving pedestrian object. Firstly, the improved Vibe moving background modeling method is used to detect the moving pedestrian in the video. Then the Shi-Tomasi corner detection and LK optical flow method are combined. The corner detection results are integrated into LK optical flow method, and the moving optical flow features of the detected corner points are extracted. Finally, Kalman filter is used to predict and track the pedestrians, and the Hungarian optimal matching algorithm is used to achieve continuous matching of moving objects and tracking of moving objects. The simulation results show that the proposed method can detect and track the moving objects in the video, and it has better recognition effect, and the detection efficiency is improved.

A Quickly Mapping Method of Low Altitude UAV Images Based on SSEQ Algorithm

WANG Jin , CHEN Xiao-xuan , XU Jia-li , WANG De-juan , YUAN Huan-huan

2021, 0(03): 122-126.

Asbtract ( 208 )

PDF (2067KB) ( 142 )

References | Related Articles | Metrics

Aiming at the problems that the number of image processing has increased sharply and some images have poor quality due to UAVs low flight height, small field of view and small image size, this paper proposes a quick mapping method of UAV images based on dubbed spatial-spectral entropy-based quality algorithm. Firstly, the unmanned aerial vehicle image is sampled and segmented to extract the spatial entropy and spectral entropy information of the local image blocks. Then, SVM and SVR are used for training、prediction and regression to get the final image quality score. Finally, based on the evaluation results, the higher quality image is selected to complete stitching experiment, and the coordinates of the control point are extracted to record the time of the experiment. The experimental results show that the orthophoto generated based on the SSEQ algorithm is close to the accuracy of the orthophoto generated by the original image stitching, and the stitching efficiency is increased by 39.27%.

Table of Content