Table of Content

24 March 2020, Volume 0 Issue 03

Previous Issue    Next Issue

Lateral Collision Risk Evaluation Between Unmanned Aerial Vehicle and Manned Aircraft in Controlled Airspace

PAN Wei-jun， CHEN Jia-yang， ZHANG Zhi-wei， ZHANG Xiao-lei, LIU Kai-yuan, WANG Si-yu

2020, 0(03):  1.  doi:10.3969/j.issn.1006-2475.2020.03.001

Asbtract ( 307 )   PDF (1050KB) ( 205 )  

References | Related Articles | Metrics

With the gradual development of unmanned aerial vehicle and the expansion of its airspace range, the air space is becoming more and more limited. Integrating UAV systems into the controlled airspace will be one of the most necessary ways to solve the air space congestion problem. Based on the performance of the unmanned aerial vehicle and the requirements of the onboard equipment and facilities, combined with the delay of manned aircraft and UAV, the collision interval between them can be deduced. According to the principle of equivalent safety level, the target safety level between them can be obtained. A lateral collision risk model between UAV and manned aircraft based on the Reich model is developed. Then according to simulation examples, we assess the collision risk between UAV and manned aircraft in different controlled areas, calculate the collision risk, and compare it with the value of target safety level. The simulation results show that the deduced anti-collision separation used in controlled areas above 1000 meters meets the demand of ICAO, so it is usable in the controlled area.

Real-time Debris Flow Warning System Based on Infrasound Monitoring

SHANG Dong-fang1, LIU Dun-long2, 3, HAN Xue1, WANG Rui-xi1

2020, 0(03):  6.  doi:10.3969/j.issn.1006-2475.2020.03.002

Asbtract ( 315 )   PDF (4891KB) ( 162 )  

References | Related Articles | Metrics

False alarm often arises from the current debris flow infrasound alarms under the environmental noise interference; therefore, it cannot be put into practical application. To improve the accuracy of debris flow infrasound monitoring and warning, based on the principle of characteristics difference of infrasound signals generated from debris flow and environmental interference, combining microcontroller with upper computer, an efficient and reliable real-time debris flow early-warning system based on infrasonic monitoring is designed and realized by means of hybrid programming, database and secondary development of GIS, etc. The recognition performance of the system has been verified by long-term in-situ monitoring of debris flows in Jiangjia Gully, Dongchuan, China. The verified results show that the system not only has the characteristics of stable operation, timely response and real-time recognition, but also has high warning accuracy and low false alarm rate, during two years of field monitoring, there were only 18 false positives and none was missed.

Fault Location Algorithm for Electric Power Marketing System Based on Causal Rules

YAN Yi, ZHOU Kai-dong, LIN Xi-jun, MAI Xiao-hui, XIAO Jian-yi, ZENG Zhao-lin

2020, 0(03):  13.  doi:10.3969/j.issn.1006-2475.2020.03.003

Asbtract ( 237 )   PDF (711KB) ( 145 )  

References | Related Articles | Metrics

Existing fault location algorithms cannot be effectively applied in complex systems with load balancing mechanisms in which causality changes frequently. To this end, a fault location algorithm based on causal rules (CRFLA) is proposed. Firstly, the causal rules between faults and events are adaptively learned by the improved causal association interesting measure method, then the root cause is inferred according to the influence degree of the fault cause set on the occurred event set. The method considers causality without the need to specify a specific causal network structure, and can flexibly combine prior knowledge. Using the data generated by the real production environment in the power marketing system to locate the fault, the experiment results show that CRFLA is superior to the traditional method, and can quickly and effectively locate the root cause.

Heterogeneous Distributed MTH1 Virtual Screening System Based on JPPF

CHEN Yun-xia， ZHANG Yang, CHEN Wen-bo

2020, 0(03):  19.  doi:10.3969/j.issn.1006-2475.2020.03.004

Asbtract ( 224 )   PDF (1345KB) ( 103 )  

References | Related Articles | Metrics

With the increasing of the cancer incidence, finding new target points for the treatment of cancer has become the research highlights of the world. According to the latest study, MTH1 protein is indispensable for the survival of cancer cells but not for survival of normal cells. Therefore, it is promising for curing cancer to design a selective MTH1 inhibitors. The screening of MTH1 needs large-scale high-performance computing resources, but there is no such an integrated distributed and cross-platform compatible system capable of both simulation and identifying potential candidate drugs from large scale database of potential molecules. Based on JPPF and Autodock Vina, this paper designs a MTH1 tumor drug screening system with good compatibility, cross-platform and high-performance. Through the virtual screening for one million set of target molecules, it is found that the screening results are directly targeted to the MTH1 drug molecules. The realization of this system provides a solution and new ideas for the rapid construction of large-scale drug molecular virtual screening technology.

Resource Location Technology for Software Ecosystem

LI Hua-ying, LIU Li, LIU Yi-jing

2020, 0(03):  24.  doi:10.3969/j.issn.1006-2475.2020.03.005

Asbtract ( 265 )   PDF (766KB) ( 117 )  

References | Related Articles | Metrics

In order to meet the customization requirements of large-scale complex software systems, the software ecosystem emerges as the times require, and gradually becomes a new development trend in the field of software engineering. How to accurately and quickly locate software resources becomes a key issue. Based on ontology, this paper gives a software ecosystem model to ensure that different software resources can be described in a unified way. On this basis, a software access method based on unified access engine is proposed to ensure accurate access to different software resources, and to meet the needs of the fine management of different organizations and a large number of software in the software ecosystem. The experiment shows that the resource management technology proposed in this paper can improve the accuracy and efficiency of software resource positioning.

Subway Vehicle Health Assessment Based on Decision Tree and Analytic Hierarchy Process

LYU Tan-yue1,2, LU Xiao-min1,2, WANG Jian1,2

2020, 0(03):  29.  doi:10.3969/j.issn.1006-2475.2020.03.006

Asbtract ( 243 )   PDF (910KB) ( 122 )  

References | Related Articles | Metrics

With the continuous growth of underground rail in large cities, subway vehicles have gradually entered a high-density and heavy-load operation state, which will reduce the safety and stability of subway vehicles and bring more subway accidents. The time and cost of subway maintenance have been increasing day by day. Traditional planned repair and fault repair cannot adapt to the current status of maintenance. In this paper, a method of health assessment of subway vehicles based on decision tree and hierarchical analysis is proposed. The soft decision tree distilled by neural network is used to qualitatively judge whether the subway vehicles are in a healthy state or not. Then, according to the expert scoring and weight calculation, the specific scores of the health status of subway vehicles are quantitatively evaluated by analytic hierarchy process. Experiments show that under the condition of optimal qualitative algorithm, quantitative evaluation of the health status of subway vehicles can effectively predict the health status of subway vehicles, and the method is robust.


Optimal Location of Urban Sports Facilities Based on Ant Colony Algorithm

LI Xian-liang1，2, ZHOU Qing-ping2, TAN Chang-geng3, TAN Yan-liang2, XU Ze-yang2

2020, 0(03):  33.  doi:10.3969/j.issn.1006-2475.2020.03.007

Asbtract ( 313 )   PDF (2138KB) ( 201 )  

References | Related Articles | Metrics

In the case of multi-objective and large-scale space constraints, the problem scale of the location of urban sports facilities is large, and it is difficult to obtain the ideal solution set. An improved ant colony intelligent algorithm model is proposed. The model mainly improves the convergence speed and accuracy of the algorithm by improving the original pheromone distribution and the evaporation coefficient of the ant colony, and calculates the ideal candidate solution. The method is applied in the site selection of sports facilities in Yuhua district of Changsha city, which gets good results. The experimental results show that the improved ant colony algorithm model designed in this paper is suitable for solving the problem of urban sports facilities location in large-scale space.

Discrimination of Injection and Production Connection Based on MLP and Sobol

WU Hai-yun

2020, 0(03):  40.  doi:10.3969/j.issn.1006-2475.2020.03.008

Asbtract ( 219 )   PDF (657KB) ( 137 )  

References | Related Articles | Metrics

In the actual production of oilfields, the connection of injection and production is a difficult but important issue. It is of great significance for the formulation of oilfield development plans and the description of remaining oil distribution. In this paper, the dynamic data of a reservoir in Dagang Oilfield is used to establish a MLP neural network model based on Bayesian optimization. The Sobol sensitivity analysis method is used to calculate the sensitivity coefficient. The sensitivity coefficient is used to quantitatively evaluate the connectivity of injection and production. The validity and reliability of the method are verified by comparison with the tracer interpretation results. The research shows that the established Bayesian optimization-based MLP neural network model achieves a good fitting effect, and the Sobol sensitivity coefficient can effectively evaluate the connection of injection and production. The result is consistent with the actual situation of the reservoir.

SVR Blood Pressure Prediction Method Based on Grid Search and Cross Validation

XI Xing-xing, LIU Yu-hong, ZHANG Rong-fen

2020, 0(03):  44.  doi:10.3969/j.issn.1006-2475.2020.03.009

Asbtract ( 341 )   PDF (788KB) ( 151 )  

References | Related Articles | Metrics

Aiming at the problems of nonstandard blood pressure measurement, large fluctuation range and low prediction accuracy, a support vector regression (SVR) blood pressure prediction algorithm based on grid search with cross validation is proposed. The algorithm first cleans the data and then finds the optimal parameter pair by combining grid search and cross validation. Then, it establishes the corresponding blood pressure prediction model by analyzing the implicit relationship between heart rate, blood oxygen and blood pressure of human physiological index data. Finally, the predicted results are compared with other several kinds of blood pressure prediction models. The results of classical machine learning model are compared, and the accuracy and rootmean square error are used to evaluate. The experimental results show that the prediction accuracy of this algorithm is about 71.39% and 81.69% respectively for high and low pressures, and the root mean square error is about 0.5349 and 0.4279, which are obviously superior to the traditional machine learning algorithms.

LSTM Rating Prediction Based on Fusing Features

ZHANG Shang-tian, CHEN Guang, QIU Tian

2020, 0(03):  49.  doi:10.3969/j.issn.1006-2475.2020.03.010

Asbtract ( 383 )   PDF (804KB) ( 243 )  

References | Related Articles | Metrics

The Latent Factor Model (LFM) can effectively extract the features of users and items. In this paper, based on the effective feature extracted by LFM, we propose a fusing-feature based Long Short Term Memory network (LSTM) prediction model (F-LFM-LSTM). Firstly, we employ the LFM model to extract the effective features of users and items. Then, the users occupation, age, gender label and item category label are fused. Finally, the prediction rating is obtained by training the LSTM. Experiments on MovieLens100k dataset show that, compared with several widely discussed algorithms, the F-LFM-LSTM model has higher rating prediction accuracy.

A New Algorithm for Rumor Source Detection Based on Information Transmission

LIU Che, LIU Zu-gen

2020, 0(03):  54.  doi:10.3969/j.issn.1006-2475.2020.03.011

Asbtract ( 346 )   PDF (951KB) ( 181 )  

References | Related Articles | Metrics

In recent years, rumor and rumor source detection has attracted wide attention from scholars in many fields. It is of great theoretical and practical significance to accurately and efficiently discover the sources of rumor propagation in social networks. Most traditional detection methods usually only detect the existence of rumors and rarely detect the source of rumors. MPA (Message-passing Algorithm) is a rumor source detection method based on rumor centrality. Based on this method, an IMPA algorithm (Improved Message-passing Algorithm) is proposed to improve the accuracy of relevant algorithm. Experimental results show that the new algorithm is more accurate in detecting rumor sources. In addition, the actual execution time is shorter for the same detection task.

Generation of Community Correctional Personnel Label Based on Stacking

WEN Jing， ZHENG Yang-fei

2020, 0(03):  60.  doi:10.3969/j.issn.1006-2475.2020.03.012

Asbtract ( 235 )   PDF (1448KB) ( 133 )  

References | Related Articles | Metrics

The standardized management technology platform for community correctional personnel is in the research stage. Due to the lack of actual data, community correctional personnel tag generation for building user portraits is not accurate enough. Therefore, based on the improved Stacking model fusion algorithm, this paper conducts modeling analysis after cleaning, sorting and feature selection for the community correctional personnel data of a citys judicial bureau. Furthermore, the prediction results of four labels, such as plead guilty attitude, mentality of society, mental health and corrective punishment case, are obtained. Comparing the prediction results with the experimental results, we can get the prediction accuracy. It shows the effectiveness and accuracy of the Stacking model fusion method for the generation of user labels of community correctional personnel.

Construction and Application of Intelligent Question Answering System#br# for Water Conservancy Information Resources

ZHANG Zi-xuan, LU Jia-min, JIANG Xiao, FENG Jun

2020, 0(03):  65.  doi:10.3969/j.issn.1006-2475.2020.03.013

Asbtract ( 386 )   PDF (1316KB) ( 205 )  

References | Related Articles | Metrics

Currently, the question-answering system in specific fields mainly adopts the method based on keyword matching to complete the question-answering, which is similar to the construction time and height of the reservoir dam, and cannot fully understand the retrieval intention of users natural language questions and give accurate answers. Therefore, this paper designs and develops an intelligent question answering system for water conservancy information resources based on knowledge mapping technology and semantic analysis. Aiming at the problem of semantic gap caused by the multi-step operation of transforming the natural language questions into structured query statements. This paper also proposes a corpus expansion method to build corpus in order to accumulate user corpus for the subsequent questions and answers based on knowledge representation.

Intelligent Evaluation Method for Loss in Postpartum Storage#br# of Grain Based on RDPSO-SVM Model

ZHENG Mo-li1, ZHAO Yan-ke1, YAN Min2, SUN Jun2, LIU Yong-rong1

2020, 0(03):  72.  doi:10.3969/j.issn.1006-2475.2020.03.014

Asbtract ( 221 )   PDF (1401KB) ( 127 )  

References | Related Articles | Metrics

The storage loss of grain during post-harvest stages is a major problem that plagues grain storage enterprises, and thus also it is an important factor affecting the economic benefits of enterprises. Therefore, the assessment of the storage loss of grain is of great significance for post-harvest loss reduction of grain. This paper investigates the factors influencing storage loss of grain through questionnaires, and models the data by the Support Vector Machine (SVM) model to intelligently evaluate the grain loss in storage stage. Meanwhile, in order to improve the accuracy of the model, this paper uses Random Drift Particle Swarm Optimization (RDPSO) algorithm to train the parameters of SVM, by making full use of the strong global search ability of the RDPSO algorithm to find the optimal solution of the model parameters. The experimental results show that the SVM model optimized by RDPSO algorithm can obtain more accurate grain loss prediction than basic SVM model and linear regression model.

Analysis of Text Sentiment Orientation Based on Machine Learning

CHEN Ping-ping, GENG Xiao-ran, ZOU Min, TAN Ding-ying

2020, 0(03):  77.  doi:10.3969/j.issn.1006-2475.2020.03.015

Asbtract ( 357 )   PDF (886KB) ( 200 )  

References | Related Articles | Metrics

In order to realize the emotional orientation analysis of movie reviews on the Internet, the film review data is crawled to obtain popular movie reviews, and text preprocessing and machine learning algorithms are used to complete the training and testing of the data, and finally the most superior sentiment classification model is constructed. The experimental results show that under the combination of all words and double words and the feature extraction of jieba’s TF-IDF and Chi-square statistics, the BP neural network and polynomial Bayesian algorithm are more suitable for the analysis of this kind of text, especially BP neural network is the best, the accuracy rate reaches 86.2%.

A Semi-supervised Learning Method for Operating State Identification of #br# Energy Measurement Devices

MA Ji-ke, YIN Fei, ZHU Yong-jin, DOU Long-long, LI Jian

2020, 0(03):  82.  doi:10.3969/j.issn.1006-2475.2020.03.016

Asbtract ( 175 )   PDF (593KB) ( 141 )  

References | Related Articles | Metrics

In view of the negative impact of abnormal operation of measurement devices on the steady economic growth and social stability of power supply enterprises, and the current situation of incomplete identification of power grid data, a semi-supervised learning method for identifying the running state of measurement device is proposed. By analyzing the data of power grid, the operation state of measurement device can be judged under the condition of incomplete identification.

Quantum Image Encryption Scheme Based on Chaotic Sequence

LU Ai-ping, LI Pan-chi

2020, 0(03):  86.  doi:10.3969/j.issn.1006-2475.2020.03.017

Asbtract ( 200 )   PDF (1753KB) ( 264 )  

References | Related Articles | Metrics

Aiming at the problem of quantum image encryption, this paper designs an encryption scheme based on chaotic sequence. Firstly, the NEQR (Novel Enhanced Quantum Representation) model is used to represent quantum images. Then, three chaotic sequences and controlled rotation gates are used to make each color qubit rotate randomly by ±π/4 radians to complete the encryption process. When decrypting, the chaotic sequence is first generated according to the secret key, and then each color qubit is rotated randomly by π/4 radians. The encrypted histogram is uniformly distributed, and the secret key space is large, the anti-attack ability is strong. The simulation results on the classical computer show that the method has better security.

Application of Fusion Model of GBDT and LR in Encrypted Traffic Identification

WANG Yao, LI Wei, WU Ke-he, CUI Wen-chao

2020, 0(03):  93.  doi:10.3969/j.issn.1006-2475.2020.03.018

Asbtract ( 274 )   PDF (859KB) ( 218 )  

References | Related Articles | Metrics

With the diversification of network application service types and the continuous development of traffic encryption technology, encrypted traffic identification has become a major challenge in the field of network security. Traditional traffic identification techniques, such as deep packet inspection, cannot effectively identify encrypted traffic, while the identification technology based on machine learning theory has shown good results. For this, an optimized encrypted traffic classification model based on the fusion of GBDT and LR is proposed, in which Bayesian optimization (BO) algorithm is used for hyperparameter tuning. By using the time-related flow features to identify common encrypted traffic and VPN encrypted traffic, it obtains an overall accuracy more than 90%, which gets better recognition effect than other common classification models.

An Anomaly Detection Algorithm for Smart Grid Time Series Data #br# Based on Combination of Isolation Forest and Random Forest

YANG Yong-jiao, XIAO Jian-yi, ZHAO Chuang-ye, ZHOU Kai-dong

2020, 0(03):  99.  doi:10.3969/j.issn.1006-2475.2020.03.019

Asbtract ( 333 )   PDF (663KB) ( 203 )  

References | Related Articles | Metrics

The information system of smart grid is the basis to ensure the normal operation of power industry, and the analysis results of various time series data in smart grid are the important basis to measure the stable operation of information system. Traditional time series data anomaly detection algorithm is difficult to take into account both accuracy and real-time. In this paper, an anomaly detection algorithm for smart grid time series data based on Isolation Forest and Random Forest is introduced. It combines the advantages of unsupervised learning algorithm and supervised learning algorithm, realizes automatic machine annotation and automatic learning threshold, labels a small number of eigenvalues manually, and improves the accuracy and real-time of time series data anomaly detection to a certain extent. The algorithm can meet the needs of anomaly detection of smart grid time series data, so as to improve the information security of smart grid.

Method of Small Sample Image Recognition Based on Prototype Network

FAN Di, JU Zhi-yong

2020, 0(03):  103.  doi:10.3969/j.issn.1006-2475.2020.03.020

Asbtract ( 351 )   PDF (963KB) ( 165 )  

References | Related Articles | Metrics

In the current image recognition field, most of the classification or recognition methods are built on the basis of existing large amounts of data, which are put into training and classified through sampling analysis and feature extraction. However, in the real world, most target classification problems do not have a large amount of annotated data. In order to solve the problem of image recognition based on small data sets, this paper uses the data augmentation to enhance data sets, and uses multi-layer CNN to map the image into high-dimensional space, then gets prototypes of each class by using the prototype network. Finally, the test image can be classified according to the distance among prototype points and test image in the embedded space. Experimental results show that this method has high recognition accuracy under the condition of small data set, and also has good stability and strong robustness.

An Improved YOLOv3-Tiny Traffic Detection Algorithm

LIU Li-ran 1, CAO Jie2, YANG Lei1, QIU Nan-hao1

2020, 0(03):  108.  doi:10.3969/j.issn.1006-2475.2020.03.021

Asbtract ( 296 )   PDF (4663KB) ( 192 )  

References | Related Articles | Metrics

The simplified version YOLOv3-Tiny of YOLO series algorithm has a relatively simple network framework and low requirement for GPU display and memory. Although the algorithm has high real-time performance but accuracy is low, it can not get accurate results in identifying driving targets. This paper first changes the size of the input pictures in order to obtain more lateral information of the pictures so that the network can easily learn the driving information. Secondly, the network structure of the algorithm is improved so as to improve the accuracy of the algorithm. Finally, the improved YOLOv3-Tiny algorithm is obtained. The experimental results show that the improved algorithm improves the accuracy while guaranteeing real-time performance.

Real-time Ship Monitoring and Recognition Based on YOLOv3

QU Wen-yi

2020, 0(03):  115.  doi:10.3969/j.issn.1006-2475.2020.03.022

Asbtract ( 389 )   PDF (2065KB) ( 181 )  

References | Related Articles | Metrics

Ship detection task faces some challenging problems, such as the changeable environment, long-distance small target, poor real-time performance. Compared with other algorithms, the advanced capability of YOLOv3 with backbone network Darknet-53 is analyzed, and a method of real-time ship monitoring and recognition based on YOLOv3 is put forward. Also for some difficult cases, the samples are further trained. So the mean average precision in these difficult cases are improved, and higher robustness is obtained. It is illustrated by the experimental data that the mean average precision of single class is up to 91.82%. The method can work as a support system for ship intelligent driving.


Human Behavior Recognition Based on Sparse Tensor Discriminant Analysis

LU Yu-tong, HAN Li-xin

2020, 0(03):  121.  doi:10.3969/j.issn.1006-2475.2020.03.023

Asbtract ( 215 )   PDF (1110KB) ( 159 )  

References | Related Articles | Metrics

In pattern recognition, how to reduce the dimension and identify the samples while extracting the key features is one of research hotspots. Based on local Fisher discriminant analysis (LFDA), this paper proposes a feature extraction method combining tensor representation with sparse analysis: Sparse Tensor Local Fishers Discriminant Analysis (STLFDA). This method transforms the feature decomposition problem in tensor local Fisher discriminant analysis (TLFDA) algorithm into linear regression problem, and solves the feature selection problem in linear regression with elastic network. It not only satisfies the goal of the Tensor Local Fisher Discriminant Analysis, but also guarantees the sparsity of the projection matrix. The validity of STLFDA algorithm is demonstrated by experiments on the Weizmann human behavior database.