Computer and Modernization

Analysis of Alarm Data Based on Improved Association Rules Algorithm

WANG Yun, LI Cong

2019, 0(12): 1. doi:10.3969/j.issn.1006-2475.2019.12.001

Asbtract ( 225 )

PDF (826KB) ( 150 )

References | Related Articles | Metrics

Aiming at the shortcomings of traditional Apriori algorithm for mining alarm data, an improved Apriori algorithm is proposed. Firstly, the algorithm introduces the weight parameter in the association rule discovery stage, and designs the support degree threshold function to mine the abnormal case occurrence law. Then a compression matrix optimization algorithm is proposed to store the compressed data in only 0 or 1. In the matrix, two arrays are used to record the total number of 1 for each row and each column in the matrix. The matrix can be compressed multiple times to improve the mining efficiency. Finally, the improved algorithm is applied to the actual police data mining analysis, and the association rules are given from mining results. Experiments show that the improved algorithm not only improves the execution efficiency compared with the traditional algorithm, but also improves the accuracy of the mining results for the police data.

Failure Prediction of Railway Scaffolding Structure Based on Kriging Model

QIN Peng-xiao

2019, 0(12): 6. doi:10.3969/j.issn.1006-2475.2019.12.002

Asbtract ( 148 )

PDF (1758KB) ( 380 )

References | Related Articles | Metrics

In order to study whether the railway scaffolding structure fails under the impact of the broken conductor, this paper proposes a Kriging model-based railway scaffolding structure failure prediction method. Firstly, the finite element model of the scaffolding structure is established to simulate the stress of the scaffold under the ultimate load. Then, the Kriging model is used to predict the damage area of the scaffolding structure under the ultimate load. Finally, the K-folding cross is used. The verification method verifies the accuracy of the established Kriging model and compares it with the radial basis function (RBF) model prediction accuracy. The results show that the proposed Kriging model-based railway scaffolding structure failure prediction method has higher prediction accuracy and can be applied to guide the structural design of railway scaffolding.

A POI Recommendation Algorithm Combined with Expert Trust

GUI Yi-qi, TIAN Xing-chen

2019, 0(12): 10. doi:10.3969/j.issn.1006-2475.2019.12.003

Asbtract ( 187 )

PDF (1301KB) ( 121 )

References | Related Articles | Metrics

Aiming at the recommendation problem that the current POI (point-of-interest) recommendation algorithm fails to deal with the sparse users of the check-in data, this paper proposes a POI recommendation algorithm combined with expert trust. According to the user’s check-in information, this algorithm selects the user check-in data in a certain space-time range, selects the expert user and makes the recommendation by combining the influence of check-in number and check-in range, uses the kernel function to optimize the recommendation result, and gets the final Top-N recommendation list. Experimental results show that the algorithm is improved in both recall rate and accuracy rate.

New Chaotic Simplified Particle Swarm Optimization Algorithm Based on Logistic Mapping

YANG Wan-li1,2, ZHOU Xue-ting1, CHEN Meng-na1

2019, 0(12): 15. doi:10.3969/j.issn.1006-2475.2019.12.004

Asbtract ( 255 )

PDF (1243KB) ( 221 )

References | Related Articles | Metrics

An new chaotic simplified particle swarm optimization algorithm based on logistic mapping (CIW-SPSO) is proposed to tackle the problems of basic PSO algorithm, such as easy to fall into local optimum, slow convergence, and low accuracy. The algorithm introduces chaos theory make inertia weight with chaotic search ability, and make learning factor changing with sine function optimization process, reduce the probability of algorithm falling into local optimum. Six classical test functions are used for simulation. The results show that the CIW-SPSO algorithm has faster convergence speed and higher accuracy, and can avoid local optimum and improve the algorithm optimization performance effectively.

biRNN-based Method for Processing Unbalanced Text Data Sets of Naval Ordnance

QI Yu-dong1, DING Hai-qiang1, ZHAO Jin-chao2, SUN Ming-wei1

2019, 0(12): 21. doi:10.3969/j.issn.1006-2475.2019.12.005

Asbtract ( 207 )

PDF (900KB) ( 161 )

References | Related Articles | Metrics

Traditional unbalanced data sets processing methods are characterized by complicated artificial settings and poor universality, which are difficult to be applied to naval ordnance unbalanced text data sets processing. Aiming at this problem, this paper proposes a method of processing unbalanced text data sets of naval ordnance based on biRNN model. The biRNN model is used to automatically learn the features of text sequences and expand a few types of texts by two-way text sequence prediction to achieve the goal of text data balancing. The whole text data set is expanded on the basis of balanced data set. Text classification experiments are carried out on the original data set, the balanced data set and the extended data set. The experimental results show that the unbalanced data set expansion method based on biRNN can effectively improve the performance of text classification by balancing and extending the original data set.

Optimization of Area Target Aiming Point Based on Differential Evolution Algorithm

E Xiang-nan1, LIU Ze-ping2, ZHANG Zi-ye3

2019, 0(12): 27. doi:10.3969/j.issn.1006-2475.2019.12.006

Asbtract ( 203 )

PDF (648KB) ( 133 )

References | Related Articles | Metrics

The optimization of aiming point is one of the core theoretical problems in missile firepower planning. In order to solve this problem, this paper designs an evaluation function based on the damage area calculation of rectangular area target, uses the principle of differential evolution algorithm to encode the coordinates of target aiming point, designs a differential evolution operator, and establishes an optimization model of target aiming point. The model is verified by designing an area target calculation instance. The experimental results show that the differential evolution algorithm has strong stability and good operability. The aiming point obtained by the model is highly reliable, and it can improve the missile strike effect and reduce the strike cost. It provides a new method for the optimization of the target point in firepower planning.

Distributed Stage Adaptive Association Rules Mining Algorithm Based on Spark

SHI Hui1, CHEN En2

2019, 0(12): 31. doi:10.3969/j.issn.1006-2475.2019.12.007

Asbtract ( 198 )

PDF (1277KB) ( 131 )

References | Related Articles | Metrics

In order to meet the growing demand for massive data mining, it is urgent to design a distributed association rule mining algorithm that can run on multiple machines. Apriori is a highly iterative algorithm that performs a large number of disk I/O operations per iteration when running on the Hadoop platform, greatly affecting and limiting the efficiency of the algorithm. This paper uses Spark to support the characteristics of memory distribution calculation and designs and implements a distributed association rule mining algorithm called Staged Adaptive Apriori on the Spark platform. The algorithm uses the adaptive data set partial processing strategy to efficiently mine frequent itemsets. The algorithm initially evaluates the execution time before each iteration, and adopts a more appropriate method to reduce the complexity of time and space. It is an adaptive association rule mining algorithm based on the nature of data sets. The experimental results demonstrate the effectiveness of the algorithm.

Fault Injection Technique Based on Vulnerability Analysis

WEI Jin-yi, XU Luo, DAI Wen-bo

2019, 0(12): 39. doi:10.3969/j.issn.1006-2475.2019.12.008

Asbtract ( 193 )

PDF (1131KB) ( 117 )

References | Related Articles | Metrics

Aiming at the increasing scale of distributed systems, this paper proposes a fault search strategy—a vulnerability-assisted recognition algorithm. Combined with the fault injection model of common fault modes, we study the fault injection technology based on vulnerability analysis for searching fault combination within the specified fault range, which enables fault injection quickly and efficiently, improving test efficiency.

#br# Image Denoising Algorithm Based on Stochastic Resonance of Saturating System

YIN Xue-jie, MA Yu-mei, PAN Zhen-kuan

2019, 0(12): 44. doi:10.3969/j.issn.1006-2475.2019.12.009

Asbtract ( 177 )

PDF (3501KB) ( 150 )

References | Related Articles | Metrics

Stochastic resonance (SR) is a mechanism that noise can be added to the nonlinear system for transmitting the certain signals, and the output can be enhanced. In this paper, we propose an image denoising algorithm based on stochastic resonance of dynamical saturating nonlinear systems. Firstly, the image is resampled into one-dimensional signals, and the parameters of the saturating nonlinear system can be tuned optimally, then the SR effect can be obtained in the saturating nonlinear system. Comparing with the effect of the one-dimensional SR, the effect of image restoration of the two-dimensional SR is closer to the original image and it is showed that the effect of image enhancement of the two-dimensional SR is superior by the output histogram and peak signal-to-noise ratio (PSNR). Compared with traditional filtering methods, the denoising effect of the saturating nonlinear system is better and it is more robust to the change of noise intensity.

Vehicle Shadow Detection and Removal Based on Multi-feature Fusion

WANG Wei, LI Zhi-hua, WU Shi-yu

2019, 0(12): 49. doi:10.3969/j.issn.1006-2475.2019.12.010

Asbtract ( 155 )

PDF (1801KB) ( 135 )

References | Related Articles | Metrics

Based on the ViBE algorithm and the edge and color characteristics of vehicles in traffic surveillance video, a multi-feature fusion algorithm is proposed to detect and remove vehicle shadows in complex traffic scenes. The method uses ViBE to extract foreground targets, and detects shadow by serial fusion strategy. Firstly, on the basis of the traditional edge-based shadow detection, the level set method is used to achieve rapid filling of inner edges of multiple foreground targets instead of horizontal and vertical operation. After obtaining candidate shadow regions, the HSV color feature and the morphological processing are combined to remove shadow perfectly. Tests with different video image sequences show that the proposed multi-feature fusion algorithm can effectively remove cast shadows and is superior to methods based on a single feature, which is suitable for complex traffic scenes.

A Two-phase Face Recognition Algorithm Based on Discriminative Low-rank Representation

CUI Juan-juan1, ZHANG Lei1, HOU Xie-lian2, CHEN Cai-kou2, ZHANG Hai-yan1

2019, 0(12): 55. doi:10.3969/j.issn.1006-2475.2019.12.011

Asbtract ( 188 )

PDF (929KB) ( 111 )

References | Related Articles | Metrics

A two-phase face recognition algorithm based on discriminative low-rank representation is proposed to deal with the noise in image training samples. In the first stage, all the training samples are processed by low-rank representation, and the M nearest neighbors of test sample are selected for rough classification. In the second stage, the samples screened in the first stage are used for discriminative low-rank representation, and sparse linear representation is used for fine classification, so as to determine the most suitable class labels for test samples. This algorithm combines the advantages of low-rank algorithm and sparse algorithm. The performance of this algorithm is proved by experiments on standard face database.

Student Position Detection and Face Image Capture Algorithm #br# Based on Classroom Monitoring Video

HU Qian-he1, FANG Shu-ya1, LIU Shou-yin1, LI Ji-ping2

2019, 0(12): 60. doi:10.3969/j.issn.1006-2475.2019.12.012

Asbtract ( 177 )

PDF (2846KB) ( 138 )

References | Related Articles | Metrics

This paper implements a student position detection and face image acquisition system based on classroom monitoring video. The system consists of a fixed focus panoramic camera and a PTZ camera. At first, the fixed camera obtains a panoramic image of the actual classroom environment. An abnormal light elimination algorithm based on the frame difference is proposed to realize the abnormal light and empty classroom detection and background image storage. The face detection is realized by HR network structure, which obtains a set of face detection boxes. As the lack of face information caused by the change of posture, occlusion and low resolution, this paper proposes a weighted moving target detection algorithm based on human head and shoulder feature to get target detection boxes, improving the position detection rate of students without face information. Aiming at the large amount of redundancy of multiple detection frames, this paper proposes multiple detection frame weighted fusion algorithms to effectively reduce the detection boxes. After the repetition of the test frame, the set of student character test frame is obtained. Then, the 〖JP2〗position information contained by the position detection is transmitted to the PTZ camera control subsystem, so that the PTZ focuses on the target students one by one, capturing a clear student face image, and providing a high quality image for face recognition.

Traffic Sign Recognition Based on Convolutional Neural Network and Ensemble Learning

LIU Shu-yi, LI Jing, HU Chun, WANG Wei

2019, 0(12): 67. doi:10.3969/j.issn.1006-2475.2019.12.013

Asbtract ( 280 )

PDF (2182KB) ( 197 )

References | Related Articles | Metrics

In order to improve the accuracy of traffic sign recognition under complex conditions (such as occlusion, perspective distortion, etc.), the paper presents an effective traffic sign recognition method based on convolutional neural network and ensemble learning. The proposed method firstly splits out the traffic signs by incorporating color segmentation, morphology processing and shape detection, and then identifies them using SVM and Softmax classifier based on the features extracted by CNN, respectively. Finally, the two kinds of classification results are integrated under the ensemble learning framework. Experimental results show that the proposed method can effectively improve the accuracy of traffic sign recognition under complex conditions, and has high overall performance.

A Ship Detection and Plate Recognition System Based on FCN

LI Zhao-tong, SUN Hao-yun

2019, 0(12): 72. doi:10.3969/j.issn.1006-2475.2019.12.014

Asbtract ( 358 )

PDF (3550KB) ( 193 )

References | Related Articles | Metrics

Ship detection and recognition are important for smart monitoring of ships in order to manage port resources effectively. However, this is challenging due to complex ship profile, ship license background and object occlusion, variations of ship license plate locations and text types. This paper proposes an efficient method based on fully convolutional neural network for ship detection and recognition named SDR-FCN. SDR-FCN, which uses a tiny fully convolutional neural network named SDNet to locate ships, then detects text of plate with PDNet designed in this paper, at last, recognizes the plate with an online adaptive classifier named OA-Classifier. The recognition accuracy of the classifier is improved with integrating the AIS (Automatic Identification System) information. The actual SDR-FCN deployment demonstrates that it can work reliably with a high accuracy for satisfying practical usages.

Simulation of Acid Water Extraction Based on Aspen Plus

XIE Ya-nan1, LI Fu2

2019, 0(12): 78. doi:10.3969/j.issn.1006-2475.2019.12.015

Asbtract ( 219 )

PDF (3751KB) ( 83 )

References | Related Articles | Metrics

The syngas produced by coal gasification process can produce waste water containing acidic gas through the carbon monoxide transformation device, which will pollute the environment and cause resource wastes if emited directly. An acid water extraction plant is simulated by Aspen Plus, and the simulation results are compared with the design values. The process is discussed depending on the process simulation, and the influences of different operating conditions are studied.

Application Research and Realization of Character Recognition in Virtual Tour Guide System

LI Fan-ruo, HAN Ying, DAI Guang-bin, FENG Tian-ge, HU Lin

2019, 0(12): 83. doi:10.3969/j.issn.1006-2475.2019.12.016

Asbtract ( 152 )

PDF (1036KB) ( 127 )

References | Related Articles | Metrics

With the continuous development of virtual reality and augmented reality technology, its application in the tourism industry is becoming more and more wide. Under the condition of the progressive expansion of the virtual tour guide application, the use of character recognition technology instead of the traditional GPS positioning and wireless radio frequency technology to realize the user positioning has the advantages of strong anti-interference, accurate positioning, high positioning efficiency and wide range of use. Taking the virtual tour guide system on campus as an example, the Unity 3D as the development platform and C# as the development language, this paper applies the algorithm of object cutting based on appearance information to realize the character recognition of natural scene through the prefabricated character recognition library and the design of character recognition filter, identifies the name and position of the current place through the recognized characters, and loads the introduction and other virtual information of corresponding place quickly and accurately. Through the experimental verification, the text recognition technology in the natural scene used in this system not only has the characteristics of fast recognition speed and high accuracy, but also can provide a user positioning method for the virtual tour guide system.

Multi-channel Heart Sound Feature Characterization Method #br# Based on PCA Serial-merged Fusion

CHENG Yu-han1, ZHANG You-xun2, SUN ke-xue2

2019, 0(12): 88. doi:10.3969/j.issn.1006-2475.2019.12.017

Asbtract ( 138 )

PDF (1708KB) ( 117 )

References | Related Articles | Metrics

The multi-channel heart sound signal not only covers more general characteristics than the single-channel heart sound signals, but also can make up for the defect that the information carried by the single-channel heart sound data. Using a four-way heart sound sensor, a small four-way heart sound database was established. Based on it, firstly, the characteristics of multi-channel heart sound signals are clarified, and the relationship between heart murmur and auscultation position is discussed. Then, the single-channel and four-channel energy entropy are extracted as effective feature data by using PCA pair. The energy entropy feature is dimension-reduced to obtain serial features. The correlation features and mutual information features are extended from real vector space to complex vector space. Finally, serial parallel features are re-converged into multi-optimal combinations feature. The simulation results show that the feature representation of multivariate optimal combination obtained by multi-channel heart sound signals is better than that of single-channel heart sound signals, which is not only beneficial to the construction of classification models, but also to quickly screen for congenital heart disease and improve classification. The recognition rate has a positive meaning.

A Causal Feature Selection Algorithm for Feedback Networks and Its Applications

YAN Yi, LI Bo, CHEN Shou-ming, LIN Qiang, HUANG Ju-tao, Wen Bo-jian

2019, 0(12): 95. doi:10.3969/j.issn.1006-2475.2019.12.018

Asbtract ( 227 )

PDF (930KB) ( 197 )

References | Related Articles | Metrics

As the two feature selection strategies of Markov blanket based methods and information-theoretic based methods often fail to solve the feature selection problem under the multi-layer network with feedback mechanism, a causal feature selection method for feedback multi-layer networks is proposed. The method first uses the D-separation method to find the neighbor node of the target node T, that is, the neighbor feature Ne(T). Then, the mutual information of the target node and the remaining features is obtained, and the feature set R of the element D-separation of the mutual information and not the set Ne(T) is found, and finally the Ne(T) and R are merged as the target node. The method effectively avoids the problem of feature selection error based on Markov blanket under feedback network and feature selection of maximum mutual information under multi-layer network. Compared with the two classic methods in the typical warning of power marketing system, the experimental results show that the method is more effective.

On-line Monitoring of Electric Power Communication #br# Transmission Equipment Based on Analytic Hierarchy Process

LI Jian-lu1, XU Li-kun1, CHEN Hai-ping1, WANG Lin1, ZHU Peng-yu2

2019, 0(12): 101. doi:10.3969/j.issn.1006-2475.2019.12.019

Asbtract ( 117 )

PDF (1389KB) ( 96 )

References | Related Articles | Metrics

Based on state perception, considering the dynamic and static data of communication equipment, combined with the experience of operation and maintenance management, using historical and real-time data such as historical defects and maintenance, current performance value and state value, this paper focuses on the research on the influence of operation environment and equipment state key parameters on the life of transmission equipment. We construct the health state evaluation model and its algorithm in order to perceive the current state of communication transmission equipment and realize the health status evaluation and the influence of various parameters on life. The research results can guide the operation, maintenance and technical reform of equipment, reasonably build and adjust the communication network, provide support for the whole life cycle management of power communication transmission equipment, and improve the management level of communication operation.

An Algorithm for Mining Key Nodes of Directed Networks Based on Contribution Matrix

ZHUANG Tian-yi, XU Guo-yan, SUN Jie, ZHOU Xing-yi, ZHU Jin

2019, 0(12): 108. doi:10.3969/j.issn.1006-2475.2019.12.020

Asbtract ( 227 )

PDF (1220KB) ( 104 )

References | Related Articles | Metrics

Critical nodes in complex networks are generally more important than non-critical nodes. Existing methods to calculate the importance of nodes are mostly based on different measurement criteria. At present, in the case of less calculation methods for node criticality applicable to directed networks and less rigorous combination of different measurement indexes in the methods, a new method for calculating node criticality of directed network is proposed. The algorithm calculates the node key value by combining the node relation and the node position with the contribution matrix. The propagation experiment on the experimental network shows that, compared with the classical algorithm that evaluates the importance of key nodes by node degree centrality and other methods, this algorithm is more efficient in the process of mining key nodes, and the mined key nodes spread information more widely in the network.

Electric Power Industrial Control Network Anomaly Detection #br# System Based on Deep Q Network

WANG Zhu-xiao, ZHANG Peng-peng, LI Wei, WU Ke-he, CUI Wen-chao, CHENG Rui

2019, 0(12): 114. doi:10.3969/j.issn.1006-2475.2019.12.021

Asbtract ( 185 )

PDF (991KB) ( 95 )

References | Related Articles | Metrics

Electricity refers to energy powered by electrical energy. The complete power system includes power generation, transmission, substation, power distribution and power consumption. Electricity is a basic industry that affects the national economy and the people’s livelihood. Power supply and security are related to national security strategies and are related to the overall situation of economic and social development. Industrial automation and control systems (referred to as “industrial control”) as the sensory and central nervous system of electricity, to ensure their network security, so that it is always in a stable and reliable state of operation, is essential to ensure safe operation of electricity. Because most networks are highly interconnected, they are vulnerable to cyber attacks. Although network-based intrusion detection systems can combine intrusion warnings and security responses well, as technology continues to evolve, attacks become more common and difficult to detect, and escape technology is a representative of such technologies. It can evade detection by the intrusion detection system by masquerading the network data stream. Combining with the knowledge and the characteristics of the power industrial control network, a power industrial network intrusion detection system based on deep reinforcement learning is proposed. The deep reinforcement learning algorithm combines the neural network and Q-learning methods into the network. The anomaly is trained to enable the system to detect intrusions and issue warnings in a timely manner.

Image Encryption Based on Fractional Rossler Chaotic Sequence

ZHANG Yi, WANG Bo

2019, 0(12): 119.

Asbtract ( 151 )

PDF (1734KB) ( 82 )

References | Related Articles | Metrics

Image encryption is very important under the background of information security. The traditional encryption method uses integer-order chaotic sequence or one or more coefficients of other low-dimensional discrete sequence as the key to encrypt the image information. Because the encryption sequence is relatively simple and the key space is small, the security is not good. In this paper, an image encryption algorithm based on fractional Rossler chaotic sequence is proposed, which takes the order and system parameters of fractional Rossler chaotic system as the key to increase the key space, while the memory characteristic of fractional chaotic system effectively increases the complexity of chaotic sequence and makes it more secure in image encryption.

Table of Content