Table of Content

15 August 2019, Volume 0 Issue 08

Previous Issue    Next Issue

Automatic Recognition of Dysarthria Based on Differential Evolution and Logistic Regression

LI Yu-xing, LIANG Zheng-you, SUN Yu

2019, 0(08):  1.  doi:10.3969/j.issn.1006-2475.2019.08.001

Asbtract ( 250 )   PDF (878KB) ( 113 )  

References | Related Articles | Metrics

Aiming at the problems of high time consuming and cost in traditional diagnosis of dysarthria speech, a computer automatic recognition method for dysarthria is proposed. Combining the Gammatone Frequency Cepstrum Coefficients (GFCC) with the common acoustic features to form a combined acoustic feature, a differential evolution algorithm is applied for feature selection, and a logistic regression classifier is used to identify the dysarthria speech. The Torgo database is divided into three subsets, which are non-words, short words, restricted sentence. 24-dimensional GFCC and 37-dimensional commonly used acoustic features are extracted to form combined acoustic features. Finally, differential evolution algorithm and logistic regression classifier are used for identificaiton of dysarthria. Experiments show that the differential evolution algorithm can effectively select feature subsets with better ability to distinguish dysarthria and healthy speech, which can significantly improve performance in the classification of dysarthria. The experiment on non-word subsets achieves 98.18% of accuracy, 98.3% of recall, and 98.3% of precision.

Detection and Recognition of Pipeline Point Cloud Based on Feedback Hough Transform

ZHANG Jing-rong, LU Zhu-heng

2019, 0(08):  6.  doi:10.3969/j.issn.1006-2475.2019.08.002

Asbtract ( 279 )   PDF (1905KB) ( 119 )  

References | Related Articles | Metrics

Pipeline is the main body of a processing factory, and its geometry is represented as a cylinder or a truncated cone. However, the existing methods of pipeline detection and identification are sensitive to noise and the accuracy of detection is also not high. Moreover, the detection results of the existing methods are difficult to verify. This paper proposes a 3D Hough transform algorithm with orientation feedback correction to detect and identify the pipelines of a processing factory. Firstly, the point cloud space is divided according to the Octree. The point cloud normal vector is calculated and the Gaussian map of a cylinder or a truncated cone is generated, and the initial orientation is estimated by the 3D Hough transform method. Then, the cylindrical cross section or the truncated cone projection profile is calculated based on the initial orientation. The orientation optimization objective function is established, and the final orientation is obtained through iterative optimization. Finally, the axis position of the cylinder or the truncated cone is fitted by Hough transform method and the radius value is computed. The experiments show that the proposed method can effectively improve the estimation accuracy of the parameters such as orientation and radius of the pipeline. At the same time, the optimization objective function presented in this paper also provides a new evaluation method for the detection results.

Exercise Training Assist System Based on Kinect

ZHENG Xuan-yu, SHI Chang, CUI Wen-cheng

2019, 0(08):  12.  doi:10.3969/j.issn.1006-2475.2019.08.003

Asbtract ( 309 )   PDF (2562KB) ( 183 )  

References | Related Articles | Metrics

According to the unsupervised and rudderless independent exercise training, also the nonstandard movements in it, an exercise training assist system based on Kinect is designed and realized. The coordinates of joint points in the human body are captured by Kinect, the features of movement are extracted, and the motion recognition algorithm is implemented by Dynamic Time Warping, then the template and the action are matched in real time, and evaluation is fed back in accordance with the matched degree. The experiment results show that the average correct recognition rate is 91.25%, the correct evaluation rate is 95.9% and the evaluation is fed back timely and effectively by the system, which indicates that the system meets the requirements of exercise training assist.

Real-time Semantic Segmentation Based on Multi-scale Fusion #br# and Its Application in Electric Power Scene

ZHOU Chen-yi, WANG Wen, LU Shan， XU Yi-bai

2019, 0(08):  17.  doi:10.3969/j.issn.1006-2475.2019.08.004

Asbtract ( 252 )   PDF (156KB) ( 194 )  

References | Related Articles | Metrics

Semantic segmentation is a basic work in computer vision. In this paper, a new upsampling structure combined point-wise convolution with dilation convolution is proposed and a real-time semantic segmentation model is designed based on this structure. The model can reach 72.1% mIoU and 125 fps running speed with the input of 640×360 on Cityscapes data set and has also good performance on a electric power scene data set. In addition, the paper transplants the model to the mobile terminal and implements an augmented reality application of electric power scene based on semantic segmentation.


Person Search System by Enhanced Deep Feature Fusion

MEI Wen-xin1,2,LIN Zhi-xian1,2,GUO Tai-liang1,2

2019, 0(08):  23.  doi:10.3969/j.issn.1006-2475.2019.08.005

Asbtract ( 250 )   PDF (6125KB) ( 144 )  

References | Related Articles | Metrics

The deep feature of pedestrian image lacks the description of local details, and it does not have the invariance of scale, rotation, translation and illumination changes fully, which leads to the low accuracy of person search. A pedestrian search system with enhanced depth feature fusion is proposed. The system integrates the pedestrian candidate network and the pedestrian identification network into a unified framework. Among them, the pedestrian candidate network realizes the acquisition and calibration of the pedestrian boxes, while the pedestrian recognition network integrates the traditional features with geometric invariance on the basis of acquiring the deep learning characteristics, which establishes a network model with enhanced deep feature fusion. The experimental results show that the network model with enhanced depth feature fusion detects and frames pedestrians in images on SSM dataset, and has a top rate of 80.7%, which is superior to the deep feature model.

Improved K-means Clustering Algorithm Based on MapReduce Framework

SONG Yang, SHI Hong-yan

2019, 0(08):  28.  doi:10.3969/j.issn.1006-2475.2019.08.006

Asbtract ( 353 )   PDF (809KB) ( 169 )  

References | Related Articles | Metrics

Aiming at the clustering effect and speed of K-means algorithm in processing massive data, a distributed parallel programming model of K-means clustering algorithm based on MapReduce framework is proposed. First, for the sensitive initialization problem of K-means clustering algorithm, a new dissimilarity function is given, according to the degree of dissimilarity between data, k value is determined, and the point with smaller dissimilarity is selected as the initial clustering center, then the K-means algorithm is deployed on the MapReduce programming model, K-means algorithm speeds up to deal with massive data by improving MapReduce programming model. Experiments show that both accuracy and convergence time of the improved K-means algorithm under MapReduce are improved compared with the traditional K-means algorithm, and the parallel clustering model has good expansivity in different data scales and the number of calculated nodes.

Improved WSN Coverage Algorithms Based on Differential Evolution #br#   and Particle Swarm Optimization

YI Wen-zhou

2019, 0(08):  33.  doi:10.3969/j.issn.1006-2475.2019.08.007

Asbtract ( 237 )   PDF (1799KB) ( 156 )  

References | Related Articles | Metrics

Dynamic deployment of sensor nodes is random, which can not guarantee the coverage quality of specific target areas. Intelligent optimization algorithm is introduced to effectively improve the quality of dynamic deployment of sensor nodes. However, the general intelligent optimization algorithm has some shortcomings such as “premature” in dynamic deployment. In order to further improve the quality of dynamic deployment of nodes, this paper studies the coverage problem of nodes, combines the advantages of particle swarm optimization and differential evolution, uses particle swarm optimization in the early stage to give full play to the characteristics of particle swarm optimization which is good at global search, and uses differential evolution algorithm in the later stage to give full play to the characteristics of differential evolution which is good at local search, so as to take the advantages of both and overcome the shortcomings of both. The algorithm has better search ability. The simulation results show that the new algorithm has better search ability and better network coverage than the improved inertia weight particle swarm optimization algorithm, the virtual force particle swarm optimization algorithm and the basic differential evolution algorithm.

Path Planning Based on Plane Unclosed Graphic Cutting

CHEN Ting， YU Xiao-ping， CHEN Jun-feng， DENG Peng， GONG Dai-ping, DENG Xiao-hong

2019, 0(08):  39.  doi:10.3969/j.issn.1006-2475.2019.08.008

Asbtract ( 273 )   PDF (891KB) ( 145 )  

References | Related Articles | Metrics

In the plane cutting process, how to confirm shorter cutting paths to reduce processing costs, reduce equipment wear and improve cutting quality is the focus of industrial applications and academic research.At present, the research on planar cutting path at home and abroad mainly focuses on closed graphics. For this reason, a greedy algorithm based on tabu search and   local   optimization of greedy criteria is proposed for the problem of unclosed graphics in the laser die-cutting industry. Firstly, the constructed greedy algorithm and the improved tabu search algorithm are combined to optimize the order of the primitives in the process, and then the local optimization coefficients of the greedy criterion are proposed to weaken the greedy algorithms “  greedy  ” idea to solve planning and optimization of processing paths. The experimental data show that the greedy algorithm and local optimization of tabu search have significant effects on the planning of the cutting path and the optimization of the idle path. The idle stroke is reduced by more than 50%, and its optimization performance is proportional to the number of primitives, which can   effectively   solve the cutting problem of unclosed primitives in the knife mold industry and other laser engraving industries.


Real-time Correction of Channel Amplitude and Phase Error in SAR System

LI Chang1,2,ZHANG Zhi-min1

2019, 0(08):  44.  doi:10.3969/j.issn.1006-2475.2019.08.009

Asbtract ( 172 )   PDF (1780KB) ( 113 )  

References | Related Articles | Metrics

The resolution of synthetic aperture radar is an important index to measure the performance of the system. Because of the amplitude and phase error in the transmitting and receiving channel of the system, the echo signal is distorted, and the corresponding pulse compression effect deteriorates, which affects the range resolution of SAR imaging. In order to correct the amplitude and phase error in real time, the correlation windowing method is used to extract the error, and a complex FIR filter is designed by complex Chebyshev approximation method. The results show that the corrected pulse compression effect is improved, and the corrected effect is related to the order of the filter. Finally, it is concluded that the PSLR and ISLR of the corrected pulse compression decrease with the increase of filter order, and gradually become stable.

Comparison and Analysis of Cache Replacement Algorithms for Network Traffic

CAO Zuo-wei1,2,CHEN Xiao1,NI Hong1

2019, 0(08):  50.  doi:10.3969/j.issn.1006-2475.2019.08.010

Asbtract ( 227 )   PDF (1365KB) ( 121 )  

References | Related Articles | Metrics

The cache replacement algorithm plays a key role in optimizing the performance of network processing applications. The research on cache replacement algorithms for network traffic is mainly concentrated on the design and application of cache replacement algorithm. Yet analysis and comparative study for the performance of the existing cache replacement algorithm in the network environment are fewer. This paper analyzes and compares six major cache replacement algorithms. By analyzing the recency and frequency characteristics of network traffic, the practical basis for the cache replacement algorithm based on Least Recently Used (LRU) and Least Frequently Used (LFU) is given. The experimental results of the simulation environment and the actual system show that the LRU-like algorithm is more suitable for network traffic than the LFU algorithm, and the random replacement algorithm is more suitable for the multi-core environment than the LRU algorithm.

Application of Single Bus Communication Technology in Handheld Sensor Calibration

WANG Bo-wen

2019, 0(08):  57.  doi:10.3969/j.issn.1006-2475.2019.08.011

Asbtract ( 216 )   PDF (1432KB) ( 106 )  

References | Related Articles | Metrics

Mine handheld sensors such as methane, carbon monoxide, hydrogen sulfide and other sensors must be regularly adjusted after they leave the factory for ensuring the accuracy of electrochemical components, otherwise affecting the test results. Traditional calibration is based on single sensor using button control. The calibration proposed in this paper is to use single bus technology to calibrate multiple sensors. Single bus communication technology is the most primitive form of digital communication. Using single communication as communication medium, it is the simplest realization, the least hardware resources occupation, and the lowest cost compared with RS485/RS232, SPI, I2C and other communication technologies. The application of single bus communication technology in the adjustment of multi-handheld sensors will greatly save manpower and material resources and improve labor productivity under the condition of saving hardware resources.

Review on Satellite Constellation Design and Simulation Software

ZHOU Yan-xi, FENG Xu-zhe, DAI Jian-zhong

2019, 0(08):  63.  doi:10.3969/j.issn.1006-2475.2019.08.012

Asbtract ( 1310 )   PDF (1872KB) ( 775 )  

References | Related Articles | Metrics

Satellite constellation is developing into vaster number of satellites, multiple network features and complicated composition, which leads to higher requirement on the design and simulation of satellite constellations. Because determining the constellation topology and designing the constellation network are the key steps during constellation design, this article gives an in-depth introduction and comparison on the latest constellation design softwares and network simulation softwares. Software functions, software characteristics and their support on satellite constellation are focused. Finally, two simulation schemes are showed to be suitable. One of them is ns-3(Network Simulator 3) and SaVi(Satellite Constellation Visualization), which are based on Linux with high flexibility and openness. The other one is commercial softwares OPNET Modeler and STK(Satellite Tool Kit). They have poweful functions, complete graphical interface and are easy to employ. Under current sofware development conditions, the two schemes can be better adapted to satellite constellation.

An Incremental Attribute Reduction Algorithm in Incomplete System

WANG Guang-qiong1,2

2019, 0(08):  69.  doi:10.3969/j.issn.1006-2475.2019.08.013

Asbtract ( 168 )   PDF (724KB) ( 124 )  

References | Related Articles | Metrics

In practical applications, the data of information system is often dynamic. When the object is increased, the original attribute reduction set is not necessarily effective. An incremental attribute reduction algorithm based on the conditional entropy is proposed for the situation of incomplete decision system object being increased. Firstly, the conditional entropy in an incomplete decision system is defined, then the change mechanism of the conditional entropy and the influence on the reduction set are analyzed when the object is increased. An incremental attribute reduction algorithm is proposed, when the object is increased, the algorithm can be more efficient to reduce the attribute. Finally, experiments verify the effectiveness and efficiency of the proposed algorithm.

Intelligent Prediction Model Based on Neural Network Algorithm #br#   and Mechanism Model for Rolling Force in Tandem Cold Rolling

CUI Chen-geng

2019, 0(08):  74.  doi:10.3969/j.issn.1006-2475.2019.08.014

Asbtract ( 280 )   PDF (1621KB) ( 251 )  

References | Related Articles | Metrics

The production mode of cold rolled strip is changing to multivarieties, small batch and low inventory, which directly results in the need to change the varieties specifications more frequently during rolling. Because changes in product specifications and rolling process status are large, the traditional mechanism model and conventional self-learning and adaptive method are difficult to ensure the setting accuracy of the first volume product. In order to improve the prediction accuracy of rolling force in the process of variable specification, this paper puts forward a composite model based on mechanism model and mass history data of rolling process. The new rolling force model, based on the theoretical model of rolling force, is corrected by BP neural network optimized by genetic algorithm (GA). The new method for prediction of the rolling force makes the average relative error of the first specification of rolled steel rolling force prediction be controlled within 5.5%. The setting accuracy is much higher than that of the conventional mechanism model.

Rapid Visualization Design for High Resolution Data of #br#   WSR-88D Dual Polarization Radar Based on Qt and GLSL

GUAN Li, DAI Jian-hua

2019, 0(08):  79.  doi:10.3969/j.issn.1006-2475.2019.08.015

Asbtract ( 260 )   PDF (3546KB) ( 113 )  

References | Related Articles | Metrics

In some provinces and cities in China, higher resolution data with more variables are available for weather warning and comprehensive judgment of synoptic situation due to the promotion of the Doppler weather radar upgrading to dual polarization radar. As it is difficult to deal with such data through the traditional GDI+ visualization method, the rapid visualization of high resolution radar data has become an urgent problem to be solved. Using the WSR-88D dual polarization radar imported from USA by the Shanghai Meteorological Bureau, this paper intents to improve the display efficiency and accuracy by choosing the Qt graphical interface library in the designation of the human-computer interaction interface and utilizing the OpenGL shading language (GLSL) in the integration of the core functions, such as the color mapping, transformation and scaling, into the fragment shader. Practical application proves that the above designation provides a friendly human-computer interaction experience and a completing complex graphics with high efficiency and accuracy. Thus, it provides essential technical support for the accuracy of weather prediction and warning.

Realization of Accelerating Gene Big Data Analysis by Grid Computing

YANG Shuang-hao1，2

2019, 0(08):  85.  doi:10.3969/j.issn.1006-2475.2019.08.016

Asbtract ( 265 )   PDF (1250KB) ( 104 )  

References | Related Articles | Metrics

In order to solve the problems of large amount of gene sequencing data, long time data analysis, high cost of building FPGA and GPU computing platform, and insufficient compatibility of computing software, the paper designs a high-throughput sequencing data analysis architecture called Sequence Grid(SeqGrid) by distributed computing ideas. The architecture installs the centos open source operating system, uses the grid engine Sun Grid Engine (SGE), an ordinary CPU, a mechanical hard disk, and a SSD hard disk, and concurrently dispatches bioinformatics software bwa, GATK, etc. to realize data analysis. The results show the 30 GB data analysis time of single whole exome sequence is shortened from 15 hours to 1 hour, and the computing speed is 15 times faster than that of the serial process, which effectively improves the efficiency of data analysis.


GQM-based Risk Assessment Method for Industrial Control Systems

YE Qian1,WANG Yu-fei2,FU Yi3,TANG Yu-lan1

2019, 0(08):  92.  doi:10.3969/j.issn.1006-2475.2019.08.017

Asbtract ( 259 )   PDF (1012KB) ( 149 )  

References | Related Articles | Metrics

Risk assessment is an essential component of safety and security assurance infrastructure mechanisms for industrial control systems. And safety and security attributes are tightly coupled. Information security assessment of industrial control systems should be coupled with the business goals. Based on Goal-Question-Metric (GQM) model, the industrial control systems risk assessment process is defined as identifying business goals, describing questions, and specification of metrics. The proposed risk assessment method is guided by the business goals, which are supported by the industrial control systems. The questions are raised on account of the scenario-based risk model. Information and data are collected concentrating on these questions. Then metrics are measured or evaluated using association analysis. Finally, a risk assessment instance of programmable logic controller (PLC) is described to specify the effectiveness of the proposed GQM-based risk assessment method for industrial control systems.

Adaptive Access Control Model Based on Hazardous Events

YANG Yang, CAO Yan

2019, 0(08):  98.  doi:10.3969/j.issn.1006-2475.2019.08.018

Asbtract ( 227 )   PDF (1032KB) ( 148 )  

References | Related Articles | Metrics

 Access control is a technology that restricts the use of resources. It is often used to provide certain resource protection capabilities in application systems. It restricts the behavior of resource requesters by restricting their own conditions, use conditions and use obligations. In complex systems, access control technology is usually used to manage the accessible resources. When there are hazardous events in the system, traditional access control model can not provide enough flexibility for the dynamic system environment. In this paper, an adaptive access control model based on hazardous events is proposed, then the dynamic adjustment method of this model is given. Finally, the possible policy conflicts in the process of dynamic adjustment of the model are analyzed based on a case, and the resolution rules of the policy conflicts are determined.

Evaluation of Surface Soil Heavy Metal Pollution in City

DAI Yong-qiang， WANG Lian-guo

2019, 0(08):  105.  doi: 10.3969/j.issn.1006-2475.2019.08.019

Asbtract ( 109 )   PDF (3966KB) ( 78 )  

References | Related Articles | Metrics

Aiming at the current situation that heavy metal pollution destroies the soil environment, and directly or indirectly harms human health, an evaluation of surface soil heavy metal pollution in city based on projection pursuit and geostatistics is proposed. Firstly, eight heavy metals in a city soil are analyzed, the pollution index of the urban surface soil is obtained by using the projection pursuit model, and the overall pollution levels of soil heavy metals in different regions of the city are evaluated. Finally, the spatial distribution characteristics of heavy metals are studied by using geostatistics and GIS, the corresponding pollution assessment is given, and the quantitative analysis of the heavy metal pollution is carried on based on the influence factors, the position of pollution sources is determined. The research results are of some reference value for the study of the evolution of urban geological environment and the development of preventive measures.

Image Annotation System Based on Crowdsourcing

CHEN Zhi-yu1,2,3,LYU Tan-yue2,WANG Fei2,DUAN Zhen-wei2

2019, 0(08):  112.  doi:10.3969/j.issn.1006-2475.2019.08.020

Asbtract ( 292 )   PDF (2623KB) ( 171 )  

References | Related Articles | Metrics

In the training process of machine vision system, a large number of labeled images are needed to enhance the recognition ability. The traditional method is to gather a group of people for independent annotation, which is inefficient and of poor quality. This paper designs an image labeling system based on crowdsourcing. The system uses collaborative filtering technology to push pictures to volunteers with corresponding majors or interests, and then sorts and classifies the tag set of the same picture through semantic processing algorithm, and finally obtains accurate and effective tags. Experimental results show that the system has better robustness and higher efficiency than traditional image labeling methods.

Design and Implementation of Provincial Meteorological Data Transmission Monitoring Platform

LI Tao1,LIU Huan1,LI Ya-ling2,XIANG Xiao-ming1

2019, 0(08):  117.  doi:10.3969/j.issn.1006-2475.2019.08.021

Asbtract ( 194 )   PDF (1550KB) ( 126 )  

References | Related Articles | Metrics

Meteorological data is the foundation for conducting weather forecasting and meteorological services. There is a lack of effective means of data monitoring at the city and county levels after the provincial meteorological data transmission is switched to CIMISS (China Integrated Meteorological Information Service System), this paper provides an overall design scheme, proposes a method of monitoring log collection based on time unit offset, and then completed database design scheme and platform display design. The monitoring platform uses MVC framework, HTML, AJAX, and other technologies to realize the monitoring and quality statistics of meteorological data. The platform aims at three levels of users in provinces, cities,and counties, and provides reliable technical support for improving the transmission quality of provincial meteorological data.


Movie Recommendation System Based on Multi-feature Fusion

LI Dan-yu

2019, 0(08):  121.  doi:10.3969/j.issn.1006-2475.2019.08.022

Asbtract ( 375 )   PDF (982KB) ( 185 )  

References | Related Articles | Metrics

Collaborative Filtering Algorithm(CF) makes recommendations based on the user-item scoring matrix, without considering the item’s own attributes. In this paper, the movie attributes on the MovieLens dataset are used as factors influencing the recommendation results, and are combined with various factors such as the introduction, comments, ratings, directors, and actors of the movie. CNN (Convolutional Neural Network) and Word2Vec (Word to Vector) word vector model are used to process the movie introduction; AFINN (Finn rup Nielsen Emotion Dictionary) is used to process the comments and the results are mapped; the director and actor data are modeled to get the prediction score under the factors, and finally the results under the various factors are weighted and combined, and the weight is adjusted to obtain the best effect. It is verified that the recommended performance of this method is better than the traditional CF algorithm.