Abstract:To improve the object detect precision of Convolutional Neural Network (CNN), we present a YOLOv3 network which based on improved loss function. This network model uses a new loss function Tan-Squared Error (TSE) which transferred from primary Sum Squared Error(SSE), and works better on continuous variable error computing. Meanwhile, the properties of TSE could decrease the impact of vanishing gradient problem in sigmoid function, and speed up model converging. The experiment results in Pascal VOC dataset show that TSE improves the detect precision effectively compared with the performance of primary network model, and the convergence is accelerated.

Abstract:This paper mainly presents the research of named entity recognition of medical texts generated by online inquiry. Using the data of online medical quiz website, we employ {B, I, O} annotation system to build data sets, and extract four medical entities of disease, treatment, examination, and symptom. Taking BiLSTM-CRF as the benchmark model, two deep learning models IndRNN-CRF and IDCNN-BiLSTM-CRF are proposed, and the validity of the model on the self built dataset is verified. The two new models are compared with the benchmark model by experiment. It is concluded that the model IDCNN-BiLSTM-CRF has an F1 value of 0.8165, which exceeds the BiLSTM-CRF's F1 value of 0.8009. The overall performance of IDCNN-BiLSTM-CRF is better than that of BiLSTM-CRF. The IndRNN-CRF model has a high precision rate of 0.8427, but its recall rate is lower than the benchmark model BiLSTM-CRF.

Abstract:As a typical short cycle and experiential product, Movie's box-office is influenced by many factors, so it is hard to forecast its box-office accurately. In this study, a box-office forecasting model based on weighted K-means and local BP Neural Network (BPNN) is constructed, with aims to improve the shortcomings of the current model and improve the accuracy of box office prediction:(1) Construct the factor influence measurement model based on Random Forest (RF) and simplify the box-office influence factors according to the value of variable importance, to achieve the purpose of simplifying the input of the following forecasting model. (2) In the traditional researches, the weight of each factor was equally allocated in sample classification, which without considering the question of different factor has different influence. So a box-office forecasting model based on weighted K-means and local BPNN is constructed, using weighted K-means clustering to classify the samples based on the value of factor influence, then build several local BPNN models based on each subsample. Experiments show that the Mean Absolute Percentage Error (MAPE) of this study's model is 8.49%, which is lower than 10.39% of the contrast experiment, which proves the superiority of the box-office forecasting model built in this study.

Abstract:In the field of target tracking, particle filter technology has the advantage of dealing with nonlinear non-Gaussian problems. However, when the standard particle filter solves the degradation phenomenon by using the resampling method, the particle depletion problem will occur, resulting in unstable filter precision. To solve this problem, the algorithm uses the differential evolution bat algorithm to improve the particle filter. In this study, the particle is characterized as a bat individual. The bat population adjusts the frequency, loudness, and pulse emissivity, and the current optimal bat individual searches in the target image area, and can dynamically decide whether to use global search or local search to improve the particle. The overall quality and reasonable distribution; the introduction of differential evolution strategies can enhance the ability of bat individuals to jump out of local optimum. In order to verify the optimization performance of the proposed algorithm, the performances of the proposed algorithm and the standard particle filter algorithm are compared. Experimental results show that the filter performance of the proposed algorithm is better than the standard particle filter algorithm.

Abstract:Aiming at the problem that the index is different due to basic data conditions and the index weight is different due to diversity of user experience in GEO satellite frequency and orbit selection risk evaluation, a flexible risk evaluation method for GEO satellite frequency and orbit selection is proposed. Based on the fuzzy comprehensive evaluation method, the evaluation method supports user to configure flexibly from two aspects:index calculation mapping and weight configuration. For the difference of the index calculation mapping and weight configuration among users, a flexible construction method based on cluster analysis and normal distribution model autonomous learning is proposed. The construction method integrates the user's experienced knowledge to form the index mapping relationship and weight which most users approve, making the risk evaluation more scientific and rational. Based on the flexible risk evaluation method, the flexible evaluation system under the B/S was developed and effectiveness of the algorithm and the system is verified by actual cases.

Abstract:The development of social networks and economic globalization gives logo a great commercial value, which makes logo detection have a good application prospect. In fact, logo objects typically occupy a small portion of the image, and the low resolution of the logo makes it difficult to further improve detection performance. Therefore, this study proposes an improved detection method based on Faster R-CNN. This approach combines the generative adversarial networks and a Faster R-CNN framework, uses the network to map lower resolution features to highly expressed high resolution features, and then sends them to fully connected layers for classification and regression. The outcomes of the experiment are evaluated on a publicly available logo dataset. The results show that the method can effectively improve the accuracy of logo object detection without affecting the detection speed of the basic network.

Abstract:High performance clusters are applied to the study of nuclear fusion. On one hand, the simulation of particles movement in the Tokamak is rely on high performance clusters heavily. On the other hand, valuable discharge data have been storaged for further researches. Keda Torus eXperiment reversed field pinch (KTX) device which is situated in University of Science and Technology of China works well, and the KTX laboratory highly demands for high performance computing and experimental data storage. Therefore, a high performance cluster is deployed and a security design for storage is proposed. Based on IOzone tests, the redundancy mechanism of the GPFS file system works, and the data read and write performance is stable.

Abstract:Because microservice fine-grained service splitting and decentralized architecture design is more suitable for current Internet agile development and rapid iteration than traditional SOA architecture, but traditional microservice service governance technology is difficult to realize the interconnection between services of different technical frameworks, and communication protocols, and there is a problem of high coupling between service governance and services. Based on the service mesh idea, this study implements a network proxy with service registration discovery, load balancing and protocol conversion as a service governance independent component of the microservice architecture, and maximize network proxy performance through Netty's framework, protobuf serialization, ETCD registry, and weighted polling load balancing algorithms. The experimental results show that the design of this study overcomes the problems of traditional microservice, and the network proxy has high availability, high concurrency, and high throughput performance.

Abstract:Aiming at the problem of the recognition of print content in the printing field, a visual identity system based on shallow features is designed. Shallow feature is applicable to the recognition of specific target. SIFT is a widely used and effective shallow feature. The system first establishes a sample image database, uses SIFT algorithm to extract features, finally identifies by matching with sample features. In addition, the system implements online learning functions. After experimental verification, the system can realize the identification of print content in real time and accurately.

Abstract:Coverage testing is often used in white box testing. The lightweight coverage testing tool GCOV has the disadvantage of operating complicatedly. The design described in this paper is based on the principle of GCOV coverage testing. According to the characteristics of batch processing of shell script, all parts of the operation are encapsulated as script tools. Through the Expect script tool asserting whether the last operation is successful or not, to achieve all script tools being excuted automatically. The output of coverage report by this design saves a lot of time than traditional operation. At the same time, this design has advantages of operating simplely and transplanting easily. It greatly reduces the repeated operation of programmers and improves the efficiency of software development and testing.

Abstract:ROS-IOT, as a new scheme of the Internet of Things system based on ROS and OpenWrt and Contiki, is put forward. It is divided into two parts:the construction of the system of the Internet of Things and the design of data acquisition robot under this system. In the perceptive layer, the sensor node uses Contiki protocol stack to realize the networking and data transmission of sensor nodes. In the access layer, access gateway adopts wireless routers running OpenWrt operating system, connection module of gateway realizes the dynamic transformation of the protocol, and it sets the functions of conversion address pool, data aggregation, and processing and forwarding to ROS network based on rosserial_embeddedLinux, which realizes the data circulation of each layer. In the application layer, a web service that interacts with ROS network data based on websocket technology is designed to achieve two-way interaction with the perception layer and the robot. The motor-driver board of the robot adopts STM32 singlechip, and the robot uses raspberry pi that running ROS environment as main control device. The software design of the robot adopts the communication mechanism based on ROS Topic and ROS_bridge to achieve the purpose of the robot are more easily integrated into the system of the Internet of Things,which is the base of expanding more services.

Abstract:In the administrative management of colleges and universities, the scheduling is a complex and critical task. The number of subjects and the limited teaching resources all restrict the complexity and results of class scheduling. The essence of class scheduling is to arrange the course and class to the appropriate teaching location at the appropriate time. It is a solution to the NP problem. As the scale continues expanding, the difficulty of solving problems increases exponentially. When the scale reaches a certain level, it is difficult to find the optimal solution in a short time. In view of this, this study proposes a genetic-ant colony hybrid algorithm, which uses a mixture of two algorithms, relies on genetic algorithm to generate pheromone distribution, and uses ant colony algorithm to find the optimal solution. The experimental results show that the hybrid algorithm improves the efficiency of class scheduling and the rationality of the class schedule.

Abstract:Mobile application technology has been gradually applied in the electric power industry. In recent years, a power generation group built a modular unified mobile platform in accordance with the mobile portal model in dual-network isolation environment. The standard functions are mainly used by the group headquarters, and some general functions can also be used by subcompanies. On this basis, this study, based on 4G and Wi-Fi wireless network, hybrid mobile development technology, Nginx reverse proxy technology, SSL VPN, gap and so on, designs a group level mobile platform with unified standard which can support two-level applications and can be flexiblely adapted to the local business needs of each subcompany. Seamless handover is implemented. It provides a reference for the construction of a group level mobile platform which supporting two-level applications in dual-network isolation environment.

Abstract:This study has improved Apriori algorithm and applied it to association analysis of diabetic risk factors. The improved algorithm transforms transaction database into a Boolean matrix in one scan time and compressing matrix according to its specific properties, it greatly reduces computation and redundancy through this way. Based on above, a diabetes prediagnosis system has been designed which adopts the latest technology stack such as micro-service framework Spring Boot and RPC framework Dubbo, RabbitMQ as message queue middleware, Redis as cache, and MySQL as database. The system assists medical staff in diagnosing diabetes and provides an effective basis for users' self-diagnosis.

Abstract:With the increasing diversity and complexity of the mission of general-purpose plane, airborne voice communication system is required to involve more members in a real-time call in missions. This study introduced a mixing method with 2 critical algorithms:mixing algorithm with real-time weights and dynamic clamping mechanism for single-output, and mixing organization logic for the customizable meeting. The mixing method brings a new design method of "customizable airborne conferences system" on the large-scale general mission aircraft.

Abstract:In order to solve the problem that teachers and students cannot communicate effectively in virtual experiment, this study uses VRML-JS-Java communication mechanism to realize a virtual physics laboratory. Firstly, the paper introduces the overall design and scene hierarchy of the laboratory. Secondly, it expounds the modeling process and process of three-dimensional scene, and focuses on the realization of the interaction and perception among server, client, and user of interactive function, as well as key technologies such as VRML virtual scene and Java interface. Finally, the experiments show that the virtual physics laboratory has good reusability and interoperability in university experimental teaching.

Abstract:In order to conceive and design the communication service subsystem of a DAS based on 4G broadband technology, the network access of FTU is realized by using the 4G LTE technology. For users' needs, in order to achieve the purpose of information transmission in distribution automation system, the VPDN services are adopted which provided by operators to group customers, and L2TP over IPSec technology is used to realize VPN network channel between DAS and FTU on the basis of the physical private line between the user and mobile operators. In the light of the constructed communication network, the DAS communication server software is designed, and the communicating function of data acquisition and output control is realized. The test results show that the requirements of communication function and network security have been realized at the communication subsystem.

Abstract:In order to meet people's needs of indoor fitness, a game system applied to the stepping machine is designed on the basis of a stepping machine in the market. It contains three parts:stepping machine, sensor equipment, and computer. The sensor equipment consists of two hall sensors, MAX232 serial port chip and STC89C52 single-chip microcomputer. The hall sensors find the pass of pedals and produce signals, single-chip microcomputer is used to calculate the step rates and send them, and serial port chip is used to transmit data to computer. The computer is used to run the game software and display the game images. The principle of the game software is that game protagonist's moving speed is determined by the average of step rate of left and right, and the directions of protagonist are determined by the difference of the step rate. The game of greed snake is adapted for the game system as test software, and functions of controlling snake movement and fitness are added in the test software. The game system can achieve the desired effect after repeated testing. The game system has good market potential because of its low cost of manufacture, easy manipulation and friendly universal property.

Abstract:In order to solve the problem of low coverage rate when nodes are deployed randomly in underwater three-dimensional (3D) sensor networks, an underwater 3D sensor network coverage algorithm based on vertical sampling is designed to improve the coverage and connectivity of underwater 3D sensor network. The coverage algorithm of underwater 3D sensor network based on vertical sampling firstly samples the 3D monitoring area in vertical plane, and then samples the plane in a straight line, which transforms the coverage problem of 3D space into the optimal problem of linear coverage. From the local to the whole, the goal of optimizing the coverage of the whole 3D network is achieved. The simulation results show that in the 100 m×100 m×100 m 3D monitoring area, the vertical sampling algorithm can increase the coverage by about 4%~28% compared with the 3D random deployment strategy, and the maximum degree of increase is obtained when the number of nodes is 40.

Abstract:Focusing on image splicing detection and splicing localization, we proposed an image splicing region detection method of noise level inconsistency. In the proposed method, we utilize the double enhancement effect of the improved Laplace operator on noise, and combine the singular value decomposition to extract the local image gradient matrix and noise features from non-overlapping image blocks. Then, we use a clustering-based threshold algorithm to classify the noise features and locate the tampered regions. Compared with the existing noise-based image splicing region detection method, the proposed method has superior performance, especially when the noise difference between the splicing region and the original region is less. In addition, the proposed method is robust to content maintenance operation such as JPEG compression, Gaussian blur, Gamma correction, down sampling, and so on.

Abstract:Aiming at the cloud resource provisioning problem, a stochastic programming based cloud resource allocation algorithm is proposed to reduce the total resource provisioning cost of the service provider. Combining the reservation plan and the on-demand plan, the cloud resource provisioning problem can be formulated as a two-stage stochastic programming problem using the probability distribution of the workload demand. Then the sample average approximation approach and a stage decomposition based hybrid evolutionary algorithm are applied to solve the cloud resource provisioning problem. The simulation results show that the proposed cloud resource allocation algorithm can obtain near optimal resource reservation scheme within a short time, reducing the total resource provisioning cost of the service provider significantly.

Abstract:Electronic voting is increasingly popular because of its convenience. However, the security problems exposed in electronic voting have become the focus of attention. How to ensure anonymity and verifiability in electronic voting has become a concern. Aiming at various problems in existing electronic voting, a multi-candidate electronic voting scheme is proposed based on digital signature algorithm and full homomorphic encryption. This scheme uses elliptic curve digital signature algorithm to solve the problem of identity authentication in electronic voting. The homomorphic encryption technology is used to realize the encryption of votes and homomorphism calculation of encrypted votes. To be able to batch votes, SIMD technology is used to packing votes. A homomorphic addition ticket counter was designed for the codec problem of encrypted votes counting. Finally, the security of the scheme is analyzed based on the eight security features of electronic voting, which shows that the scheme is safe and feasible.

Abstract:In order to optimize the delivery path of express delivery, a mathematical model based on time window is given. In this study, improved AHP-GA algorithm is used to optimize multi-target vehicle routing, and median Analytic Hierarchy Process (AHP) is used to compare the weight coefficients of multiple sub-targets, and it is not susceptible to extremes. Thus, the multi-objective optimization problem is transformed into a single objective optimization problem. The simple natural numbers are used to code the vehicle path to avoid duplication of the paths. The customer's requirement for arrival time window, including the opportunity cost of the vehicle to arrive before the agreed time, and the cost of the fine after the agreed time. Finally, this study takes 1 distribution center and 20 service customers for example, the mathematical model constructed in this study is optimized by using traditional GA algorithm and using improved AHP-GA algorithm respectively. The simulation results show that the optimal solution can be obtained efficiently by using improved AHP-GA algorithm in multi-objective distribution path optimization problem.

Abstract:With the increasing proportion and key degree of software in modern high reliability equipment, the traditional software safety testing methods cannot meet the requirement of current testing. This paper mainly introduces a new method of safety testing and analysis-the application of SFMEA technology in software safety testing, to make up for the deficiency of safety testing requirement analysis and test case design. The article first analyzes and introduces the contents, methods, and shortcomings of the conventional safety test. At the same time, it analyzes the necessity of introducing a new safety test method, and then introduces the definition, method, and process of SFMEA safety analysis technology. Finally, the software based on SFMEA safety analysis technology is described in detail. The safety testing method and workflow of the piece are introduced, and the merits and demerits of safety testing based on SFMEA safety analysis technology and the focus of attention are summarized.

Abstract:To overcome the shortcomings of the existing method in solving the 0-1 knapsack problems, an improved fireworks algorithm is proposed. After a mathematical model of the 0-1 knapsack problem is given, an improved fireworks algorithm is proposed to solve it. The main idea of the improved algorithm is as follows. The initialization of the basic fireworks algorithm is solved by using Kent chaotic map to make the initial position more uniform. The Sigmoid function is also introduced to get the gradual explosion radius to make the solution accurate and the search speed reach a certain balance. The resolving results of the typical test function and the 0-1 knapsack problem show that the improved fireworks algorithm has higher precision and more stable performance.

Abstract:Particle Swarm Optimization (PSO), as a group intelligence algorithm, effectively improves the practicability of the portfolio model, but it has the disadvantages of low search accuracy and easy to fall into local optimum. In order to overcome its shortcomings, this study proposes a particle swarm optimization algorithm based on the Beetle Antennae Search (Abbreviated as BAS), and applies it to the portfolio model with full cost. In the Optimization algorithm based on BAS (BSO), the update rule of each particle is derived from BAS. In each iteration, it has its own judgment on the environment space, and not only depends on the historical best solution in the PSO and the current global optimal solution of the particle individual, thereby reducing the number of iterations, improving search speed and accuracy. The empirical results show that the algorithm is more stable and effective.

Abstract:The resolution of the asphalt pavement disease image which collected by the CCD is too high and the area of the effective area containing information is small. A region segmentation algorithm for asphalt pavement disease image based on image content was proposed to eliminate the invalid region in the pavement image. Firstly, the original image was processed into a binary image containing the disease characteristics through a process of preprocessing and disease extraction. Then the initial traversal direction is obtained by calculating the up-to-down ratio and the left-to-right ratio of the pixels which contain information in the whole image, and counting the total number of information pixel of each row (or column). Finally, traversing from the initial traversal direction and discarding the row (or column) with the least amount of information in order to finally obtain the segmented image. In order to verify the validity and rationality of the algorithm, the image information entropy was used as the algorithm evaluation standard and compared with the traditional algorithm. The experimental results show that the proposed algorithm can keep the target information very well on the premise of effectively reducing the image resolution, and improve image information entropy.

Abstract:To meet the real-time demand of moving object detection in Unmanned Air Vehicle (UAV), and to cope with the problems of moving background and variable illumination, a novel moving object detection technique based on Single Gaussian Model (SGM) and optical flow is presented. First, an improved SGM is applied to model the background of the image captured by moving camera, and then the corresponding models of previous frame are fused to compensate the motion of camera. Second, the obtained foreground is used as a mask to extract feature points to calculate optical flow, and then these sparse points are clustered to detect the objects. Experimental results demonstrated the effectiveness of the proposed approach in preventing the background model of SGM from being contaminated by the foreground, as well as dealing with illumination changes. It can also update background model quickly and obtain moving objects precisely.

Abstract:Replica selection algorithm for streaming media in Cloud-P2P (C²P²RSA²) is proposed based on ant colony algorithm in this study. In replica selection model based on ant colony, a copy select metrics (copy node network bandwidth, network delay, etc.) is mapped by ant colony pheromone. The replica pheromone probability formula is proposed. Through constantly iterate the optimal replica resource is selected. The experimental results show the average access time of C²P²RSA² is increased by 2%-5% than that of PARSA (pheromone-base ant colony replica adaptive selection algorithm in cloud storage) and best copy selection algorithm, the cloud replica node load factor of C²P²RSA² is decreased by 15%-25%.

Abstract:In 3D facial expression recognition, operator algorithm based on LBP has the characteristics of accurate, precise, and invariable illumination, while it also has the disadvantages of high dimensions of histogram, poor discriminating ability, and large redundant information, comparing with traditional feature extraction method. In this study, a CPB algorithm is proposed based on multi-scaled block CBP feature extraction system for the classification of facial expressions that are represented in 3D, designating to extract features more efficiently. Then, sparse expression classifier is used to classify and identify the features. The experimental results demonstrate that the facial expression recognition rate has been greatly improved, comparing with traditional LBP algorithm and SVM algorithm.

Abstract:Aiming at the problem of weighted overlapping community detection in complex network, the DBGLLJ (modularity Density and Jaccard based BGLL) method for weighted network is proposed. The network is firstly reconstructed by the importance of node, and then the network is divided into a series of segment according to the modularity gain and the module density gain as the phase function. The overlapping detection method combined with the improved Jaccard index is also proposed. In order to verify the proposed method, three algorithms were selected for testing in LFR networks and real-life networks. The results show that DBGLLJ method is better than the others in standard LFR networks and real-life networks, and has higher overlapping modularity which shows the effectiveness and accuracy of the proposed method. The proposed method is also applied to the reality network of the complex electromechanical system. The overlapping detection result is better and has higher reference value.

Abstract:A health assessment model for power automation equipment based on fuzzy neural network is proposed to solve the problem that the evaluation method is simple, the evaluation method is weak, and the evaluation effect is inaccurate. Using fuzzy theory to analyze, use fuzzy sets to describe evaluation indicators, and use the degree of membership to describe equipment status, combining the adaptive function of neural network to provide more accurate and personalized health evaluation for individual devices.

Abstract:Aiming at the demand of marine ecological environmental data by marine aquaculture enterprises and the administrative department, a kind of mobile terminal software based on the Android operating system is designed. The APP displays the real-time data in the marine ecological environment for the users. The APP is simple, practical, and convenient to be operated, so as to better consider the needs of users. The Vitamio multimedia framework is adopted to realize the underwater video-live broadcast and video-on-demand. The APP can display the hydrologic data in a graphic form and show the location of delivery system by Baidu map, as well as using Aurora push platform to receive and push messages to remind users. The application results showed that the APP is highly practical and stable.

Abstract:Due to the strong mobility of nodes, higher requirements for multicast routing and multicast group partition were required.And the General Multicast Routing Protocol often failed to meet the requirements of the network.A distributed kernel stable routing algorithm based on ODMRP was proposed in this study. The algorithm combined the improved greedy mechanism and optimized the path according to the routing condition.In order to reduce the burden of data processing, a distributed kernel node selection mechanism was proposed. The information of multicast group was stored in multiple kernel nodes, and the storage space of forwarding nodes was also released. In this study, the routing repair mechanism was used to monitor the change of routing status and repair the broken routes in advance to ensure the validity of the path. The simulation shows that compared with the ODMRP algorithm and VCMP algorithm the algorithm can optimize the transmission path, reduce end-to-end delay, increase the rate of packet delivery and improve the performance of the network.

Abstract:In view of the water quality prediction problem, taking the surface water quality monitoring factors as the research object, a Long Short-Term Memory (LSTM) neural network based model is proposed for water quality multi-factor prediction. At the same time, the proposed K-Similarity method is used to denoise the input data of the model to improve the prediction performance of the model. Compared with BP neural network, RNN, and traditional LSTM neural network prediction model, the experiment shows that the proposed method has the least square error and the prediction result is more accurate.

Abstract:In the background of the application of UAV to the power line inspection, a new power line detection method based on UAV image is proposed in order to convenient analyzing the power line. First, an adaptive threshold Canny edge detection method based on Otsu is proposed, which is used to detect the edge of the power line. Then, the binary images obtained by the edge detection are processed by mathematical morphology, and the line segment detection is carried out by Hough transform improved by fractional look-up table method. Finally, a method based on line to line spatial information analysis is proposed to screen and fit the line segments. The experimental results on UAV images show that the proposed power line detection method is a well performance power lines detection method, which is based on UAV images.

Abstract:WebShell can be divided into various types according to its function and size; they have basic features and unique features. However, most existing WebShell detection only extracts features from single level, they cannot cover all the features of various types of WebShell in a more comprehensive way. These detections have problems such as kind bias, poor detection effect, weak generalization ability, etc. To solve these problems, a random forest based WenShell detection method is proposed. In the data preprocessing stage, this method extracts the statistical features of the text layer, and the sequence characteristics of the text layer sources and the compilation result layer opcode, to form a comprehensive combination features. Then, the feature set of the sample is formed by using Fisher feature selection to select important features with the appropriate proportion to reduce the feature dimension. Finally, the random forest classifier is used to train samples to get the detection model. The experiment shows that this detection method can detect WebShell effectively, and it is superior to the single level WebShell detection model in accuracy, recall, and false alarm rate.

Abstract:In order to solve the cold-start problem of the traditional collaborative filtering algorithm and to improve the performance of recommendation, this study focuses on the cold-start problem and proposes two algorithms. Cold-start problem of new users:user-based collaborative filtering algorithm integrated with user's information model, cold-start problem of new items:item-based collaborative filtering algorithm applying hierarchical clustering. After a series of experiments carried out on public data sets-MovieLens, comparing the difference between the precision and recall value of the improved algorithm and the traditional one, the results show that the new algorithm can effectively alleviate the cold start problem and improve the quality of recommendation.

Abstract:In the multi-source and multi-point environment, the path optimization involving dynamic loading and product distribution in dynamic logistics is a very complicated problem. Aiming at the diversification of goods demand in the actual distribution process and the path optimization problem of multi-vehicle delivery and high idling rate, this study proposes a new scheduling and distribution method. By establishing a vehicle loading and distribution path model, using the multi-source point multi-destination, weight correction, path optimization, etc. as constraints, a new way of simulating cell division is used to generate the next generation and improved the existing genetic algorithm to solve the problem. This method optimizes the generation of the initial population can quickly obtain the global optimal solution, jump out of the genetic premature convergence, get the best path, reduce the distribution cost and improve the distribution efficiency.

Abstract:The article first compares the two existing design methods of FPGA, concentrating on the advantages of graphical programming, then taking the Sobel edge detection algorithm as an example, a digital image processing system is built on the Matlab/Simulink platform based on the System Generator design tool. The graphical modeling and online simulation of the algorithm are completed and the design result is compiled into Hardware Description Language (HDL). Finally, the function verification and test result analysis are carried out, and the results of visual processing of graphical modeling, RTL view, actual consumption of resources, and running speed are discussed.

Abstract:Because of the limited number of meteorological observation points, uneven distribution, and inconsistency with the line corridor, it is unable to provide accurate data model for transmission line meteorological environment which is used to study line fault and defense technology caused by disasters. This study analyzes the principle of Kriging interpolation algorithm, realizes the grid modeling of meteorological environment for transmission lines by using Kriging interpolation method. In addition, this study formulates the evaluation criteria of the interpolation results, and selects the temperature data on one day in 2017 of 996 meteorological observation points in Jiangsu Province. Through the experiment simulation and the analysis of the Kriging interpolation results, the selection process of the kriging semi variant function model which is suitable for the grid interpolation of the temperature data in this area is verified.

Abstract:The spaceflight tracking and control system uses the special data protocol which makes higher demands on the algorithms of data processor. In order to complete the processing of mission data, the modularization processing model and aircraft telemetering processing model are designed based on the behaviour of the data formats. The selecting of processing model guarantees the correctness and reliability of data processing and makes the soft system more suitable and general.

Abstract:Big data services platform has important applications in the power distribution system. Massive quasi-real-time data service platform is a key data center in power grid, and a lot of valuable data is stored in the platform. The data in platform is unstructured, and cannot be selected by SQL statement. To maximize mining the data value, platform needs to integrate function that the accesses to real-time data, for this develop SQL engine based on PostgresSQL foreign data wrapper (Postgres_fdw). The SQL engine proposes a foreign table space estimation scheme and SQL resolution scheme for real-time databases in massive platforms. This access scheme provided good support for big data platform to get data from the underlying massive data service platform with an SQL way. Field function and performance tests validated the effectiveness of the engine.