Abstract:Object detection is the foundation of computer vision applications. Some object detection algorithms based on anchor boxes have been unable to meet the requirements for object processing efficiency and performance in object detection, and anchor free method is gradually widely used in object detection. This article firstly introduced a series of key-based anchor free object detection methods based on the CornerNet, CenterNet, and FCOS model, and summarized the algorithm ideas, their advantages and disadvantages. Then the performance comparison and analysis of the object detection algorithm based on anchor boxes and key points were performed on the same data set. Finally, the object detection based on key points was summarized, and the future development direction of object detection was prospected.

Abstract:The high bandwidth capacity, high reliability, and low latency communication requirements of 5G/B5G mobile communication systems require more novel technology support. Millimeter wave (mmWave) has become one of the research hotspots of 5G/B5G mobile communication systems due to its rich spectrum resources and extremely high bandwidth capacity. Different from the past Internet architecture dominated by wired networks, the mobile Internet has become a fusion of wireless access networks and high-speed core networks, so the network transfer performance is facing huge challenges. However, the research work on the transfer performance of mmWave end-to-end communication is still relatively small and simulation experiments are basically used. In this study, we use the actual network devices to carry out experiments under real network environment to analyze the link bottleneck in the transfer process of 5G/B5G mmWave mobile network by measuring the basic transfer performance of 5G/B5G mmWave link and the end-to-end TCP transfer performance in the 5G/B5G mmWave mobile network. The results lay the foundation for designing mmWave end-to-end network transmission protocols and improving network transmission throughput.

Abstract:This study designs and proposes a block chain architecture based on the domestic cipher algorithm — “DCchain”, and replaces ECC and SHA-256 of the international general cryptographic algorithm with SM2 and SM3, which realizes the autonomous control of the block chain architecture. At the same time, according to the current status of consensus algorithm compromise faced by block chain architecture, a “pluggable consensus” protocol is designed to solve the unalterable problem of consensus algorithm faced by block chain architecture. Experimental results show that “DCchain” and ordinary block chain architecture have higher consensus efficiency and lower resource cost under the condition of consistency and effectiveness.

Abstract:A wise cash forecasting method based on a dynamic weighted combination model is proposed in this study, to precisely predict the daily cash consumption of ATM equipments so as to make a better decision for daily cash transfer management. Different from single-algorithm prediction used in the past, with analyzing characteristics of banking business, transaction flow, and equipment, etc., an intelligent algorithm based on a dynamic weighted combination model that combining 4 single machine learning models, is proposed and implemented in this study. This algorithm provides a more intelligent, more precise, and more efficient forecasting method for the management of bank cash consumption, effectively reduces the total amount of cash inventory and the rate of cash return, and improves the utilization rate of cash. This method has been used in Guangdong, Chongqing, Jiangxi, Shanxi, Beijing, and other areas with sound results.

Abstract:Recently, the number of Chinese films and audience as well as film market have been increased rapidly, while various box office prediction approaches have been proposed and investigated. However, these approaches only utilized stable information of films for prediction without using any dynamic information, making them unable to adjust predictions in real-time. Meanwhile, current methods only predict total box office of films but ignore everyday’s revenue and other audience information such as attendance rate, the number of audiences. In order to accurately predict each day’s audience data during the screening period, this study proposes a prediction algorithm based on Generative Adversarial Networks (GAN). For a film, this algorithm firstly converts the time-series audience data of available dates to frequency domain using Fourier transform, allowing global feature to be extracted. Then, a novel dynamic encoding method is proposed to obtain the dynamic of recent days. Finally, the deep spectral feature is extracted and future audience data is predicted using GAN. The experimental results show that the proposed algorithm can accurately predict the daily audience data of films. Compared to only using stable information, the proposed dynamic method can improve the prediction accuracy. When combined both stable and dynamic information, the best results are achieved. Therefore, the algorithm can provide useful information for the film marketing industry.

Abstract:In order to motivate Internet Service Providers (ISPs) to deploy in-network caching, we need to develop a new pricing mechanism among ISPs, Content Provider (CP), and users. The existing pricing mechanism of Information Centric Networks (ICN) mainly considers paid content, but more practically, there is more free content on the network. Therefore, based on the free content, we first propose one pricing and caching mechanism with advertisers’ participation in ICN and use game theory to solve our model. The analysis shows that participants in the network have the best advertising charging strategy, content caching strategy, and the best profit point. The simulation results show that the model can maximize the participants’ profits in the competition process. At the same time, it can be seen that this model is more practical and more profitable than the traditional charging model, so it can achieve the purpose of motivating ISPs to deploy in-network caching.

Abstract:With a large number of publications published every year, academic data grows rapidly. Through published paper data to accurately and comprehensively present a scholar’s scientific research level and core competitiveness so as to provide assistant decision-making for managers, decision-makers or investors of large-scale scientific research institutions, big data visualization has become a research hotspot of literature. This study based on Web Of Science (WOS) paper data, (1) in order to improve the quality of data, a combination of algorithms and interactive visualization is used to design entity grouping algorithm and grouping visualization correction tool for the data characteristics of WOS papers, which can eliminate the difference between person name and affiliation name; (2) according to the commonly used academic competitiveness index, the visualization method of scholar’s profile is designed; (3) a set of visualization system of scholar’s profile based on thesis data is developed, and the visualization system of scholar’s profile based on publicated papers is developed. The real case of body proves the practicability and effectiveness of the system.

Abstract:Blockchain technology has been expanding in the field of digital rights management. Blockchain has the characteristics of openness and transparency. If the digital copyright is stored in the Blockchain as a whole, it is difficult to protect the privacy of the copyright. At the same time, due to the limitation of the block size, large copyright files are difficult to store. This study proposes a Blockchain video rights management framework based on deep learning. Under the condition of ensuring that the video copyright can be stored in the basic Blockchain, this architecture uses OpenPose to output the 18 key points for the pose of the characters in the character video, extracts the key information of the relevant digital copyright, and finally stores it in the Blockchain, which not only ensures the privacy of the information, but also ensures the efficient storage of the information.

Abstract:In order to make good use of laboratory data, a laboratory management model is proposed, and then a logical framework of management and data-driven intelligent process are put forward. A tagging system is used to sort management and storage data of laboratory and intelligent processing model such as the GBDT of laboratory evaluation, and an intelligent management platform based on the tagging system is constructed. Two modes of lab management, batch and scenario, are realized based on the platform. The intelligent platform has strong expansibility and can play a role in the whole life cycle of the laboratory.

Abstract:The ability to perform complex reasoning on semantic data streams generated by sensors has recently become an important research area in the Semantic Web community. Currently, most RDF stream processing systems are implemented based on SPARQL (W3C Standard Protocol and RDF Query Language), but these engines have limitations in capturing complex user requirements and processing complex reasoning tasks. In response to this problem, this study combines and extends Answer Set Programming (ASP) technology for continuous processing of RDF streams. In order to verify the effectiveness of this method, we firstly take the smart home ontology as the experimental object, and analyze the common characteristics and complex events between the sensor devices to build the ontology library; then generate instance objects based on the ontology library and generate RDF data stream through middleware. Next, through extending ASP, making full use of its expression, and reasoning capabilities and reducing the reasoning time, a window partitioning strategy for the RDF stream in this method is designed. The static knowledge base is selectively loaded according to the user’s request. Finally, the comparison with Sparkwave and Laser through experiments proves the performance advantage of this method in terms of latency and memory.

Abstract:Instant messaging system (IMS) has become an important communication method due to its real-time characteristics, which can improve work efficiency and reduce communication costs. It plays an increasingly important role in enterprises, schools, governments and other organizations. However, while instant communication brings convenience, its inherent security weaknesses hinder its further development. In order to ensure the security of the instant messaging system, some advanced security encryption algorithms are used in the communication system to prevent attacks and information leakage. However, these algorithms have their own shortcomings in terms of encryption strength or encryption speed. After understanding the limitations of these encryption algorithms, in this study, we proposed an alternative algorithm that aims to leverage and combine the best features of both these algorithms and provide much better security than either of them, namely 3DES-RC4 hybrid encryption algorithm, is an algorithm with a 256-byte key space. The complexity is increased from O(2¹⁶⁸) to O(2⁵¹⁰⁰) compared with 3DES algorithm. Based on this algorithm, an instant communication system is designed. The encryption and decryption functions of the system are tested, and the performance and strength of the proposed algorithm are analyzed. By comparing to the 3DES algorithm, it is proved that the algorithm proposed in this paper retains the characteristics of 3DES encryption strength and RC4 pseudo-randomness, and is superior to the constituent algorithms in terms of encryption strength and adaptability.

Abstract:As the information industry develops, data producers have generated great masses of valuable data. Data identifiers are assigned for data sharing to resolve where the data is located. In order to make the data accessible through the identifier for a long time, it is necessary to ensure that the identifier resolution service is available for a long time. Majority existing identifier systems use a semi-decentralized structure, while some of them have gradually lost their resolution capacity due to reliance on the final resolution service. Based on the consistency of distributed ledger data of Blockchain system, this paper proposes a persistent identifier system based on Blockchain. On the basis of compatibility with the access layer of the existing identifier system, a storage layer is provided to ensure the durability of the identifier resolution service and the long-term correct storage of data. Test results based on the Handle system and Hyperledger Fabric show that it can provide better data integrity and long-term availability of resolution service for persistent identifier service under the premise of providing acceptable request response speed and storage occupancy.

Abstract:In order to improve the efficiency, user experience, and utilization of the traceability system, this study puts forward the method of designing configurable dairy products traceability platform under specific risk control level. This method selects parameters related to food risk, including HACCP, risk levels, risk occurrence probability, traceability, and cost selection. Binary related mathematical derivation is used to generate the coding results of these parameters. The coding results are analyzed by combining low, medium and high intervals, and detailed design of the configurable function of the traceability system. Taking dairy products for example, Customized services are provided according to the enterprise demand system. The traceability system can be customized according to the personalized needs of the enterprise to solve the efficiency, user experience, utilization, and other problems of the traceability system. In the future, the retrofit and design of different food types traceability platforms will be implemented, so as to increase the enthusiasm of enterprises to apply traceability systems, the efficiency of government supervision, and the convenience of consumer inquiries.

Abstract:Compared with fossil energy power generation, the use of new energy such as wind power and photovoltaic power generation is conducive to energy security and social sustainable development. However, large-scale wind and photovoltaic power in national grid are seriously challenged by the overall dispatch of the power system, due to the regional, intermittent, random and unpredictable of wind and light. This study optimizes the objective function and the inter-regional tie-lines of the mid-to-long-term wind power acceptance capacity assessment model, and adds photovoltaic unit output constraints. The system has been practically applied in a province and a region, assisting the grid power system to determine power generation plans and unit maintenance arrangements. It can reduce the occurrence of new energy abandonment, and provide effective guidance for the actual dispatch and planning of power systems through visualized results.

Abstract:Gait-based feature recognition is an emerging biometric authentication technology, aiming at analyzing human characteristics such as identity through the walking posture of people. Compared with other biological recognition technologies, gait-based methods have the advantages of being difficult to hide, contactless, and remotely usable. This study designs a single-view gait-based human identity and attributes recognition system under video surveillance. The system uses image processing methods to detect a human gait in real-time from a complex surveillance video. After analyzing with the algorithm trained by deep learning, it can obtain the information of human's identity, gender, and age. Experiments show that the accuracy rate of the system is 98.1%, the accuracy of gender prediction is 97.1%, and the mean absolute error of the age prediction is 6.21, which are better than the traditional benchmark. The system is costless, supporting real-time detection, which can fully meet the needs of small and medium-scale gait research and analysis.

Abstract:In order to achieve real-time monitoring of container clouds, host devices, and business systems, a cloud platform monitoring and alarm system based on Kafka and Kubernetes is designed. Docker containers are managed through Kubernetes, and Kafka receives device operation information from different hosts in different regions. The business system is monitored through probes. By setting the alarm association rules, redundant alarms are reduced, alarm fault detection capabilities are enhanced, and alarm accuracy is improved.

Abstract:Effectively identifying the different group of human in an image or video is an important part of intelligent image analysis. It is how to obtain “effective features” in the image. Based on the convolution neural network model, this study proposes a multi-model fusion convolution neural network method. The model trained by ImageNet participates in the initialization of the weights of the neural network model, achieves more effective features on the premise of effectively saving time and resource calculating costs. Experiments prove that the model can maintain the recognition accuracy of adult males, adult females, and children in natural scenes at about 85%, which improves the accuracy and reliability of group classification.

Abstract:Giving or recommending appropriate content based on the quality of experience is the most important in recommender systems. This study proposes a new CBCF (Clustering-Based CF) method using an Incentivized/Penalized User (IPU) model, which is thus easy to implement. The purpose of this study is to improve recommendation performance of accuracy, recall and F1-score by studying the differences of users’ preferences through IPU model. This study formulates a constrained optimization problem in which we aim to maximize the recall (or equivalently F1-score) for a given precision. To this end, users are divided into several clusters based on the actual rating data and Pearson correlation coefficient. Afterward, we give each item an incentive/penalty according to the preference tendency by users within the same cluster. Experiments show that under the condition of given accuracy, the recall rate of the proposed algorithm can be improved by about 50%.

Abstract:SQL injection vulnerability has been the one of the most problems that threaten Web application security. Among them, second-order SQL injection vulnerabilities are more subtle and destructive than the first-order one, and the detection usually depends on the tester’s prior knowledge and experience. At present, in the Black-Box Testing scenario, there is no effective detection method for the second-order vulnerability yet. Utilizing the idea of model-based test case generation, in this study, a Test suite Generation method based on a Client Behavior Model (CBMTG) is proposed to get a test suite capable of detecting second-order SQL injection vulnerabilities in Web applications. In the CBMTG, firstly, the mapping relationship between transitions and SQL statements is established through the execution of the initial test suite. Then, the topological relationship between transitions is established through the field analysis of the SQL statements. Finally, the final test suite is generated under the guidance of the topological relationship. The experimental results show that the method in this study performs better in most Web application than the state-of-the-art second-order SQL injection vulnerability detection methods.

Abstract:With the rapid development of computer artificial intelligence, the number of cameras is increasing, and the amount of video data is also increasing rapidly. The security monitoring and tracking of humanoid trajectory in video is an important research direction of large-scale intelligent monitoring system. Considering that the difference of illumination and darkness of different cameras in different security control scenarios and the human angle and size of each frame will affect the accuracy of human tracking, Correct Single Shot multibox Detector (CSSD) network with advantage of fastness and associated analysis are proposed for human tracking. Based on the pedestrian multi-object tracking technology, this study proposes a CSSD network for model detection, and uses ordinary Kalman filter to track and predict the position of the target, predicts the position of the detection box, and uses IOU method and Hungarian algorithm to solve the problem of video frame target matching before and after. It has been proved that this method can effectively improve the accuracy of humanoid targets, alleviate the large changes caused by epigenetic mutation or partial occlusion, and adapt to the size, distance, and angle changes of targets to the greatest extent.

Abstract:The norm of opereation in electric power maintenance is related to the personal safety of the staff, and is very im-portant to the development of electric power industry. In order to detect the illegal operation behavior of power maintenance workers from the perspective of computer vision, a multi-tasking and multi-branch illegal behavior detection algorithm was designed based on the Mask RCNN algorithm. It integrates target detection, key point detection and instance segmentation tasks, and performs parallel target detection. Detect and obtain the frame coordinates, keypoints, and mask information of the target. The experimental result demonstrates that this algorithm has significantly improved the precision in instance segmentation and key point detection, has higher accuracy and robustness compared with Mask RCNN. And it meets the accuracy requirements of actual deployment in power maintenance violation detection.

Abstract:In order to solve the problem of large amount of data and slow retrieval speed in the existing audio retrieval, a fixed audio retrieval method is proposed in this study based on compressed sensing and audio fingerprint dimensionality reduction. In the training stage of audio retrieval, the sample audio signal is sparse processed, and the sparse audio data is compressed by the compression sensing algorithm, then the audio fingerprint is extracted, and then the audio fingerprint discrete Gini coefficient is introduced to reduce the dimension of the fingerprint by calculating the discrete Gini coefficient of each dimension of the audio fingerprint. In the recognition stage of audio retrieval, we use the same algorithm as in the training stage to process the audio to be tested and match with the sample audio fingerprint. The experimental results show that the proposed audio retrieval method greatly reduces the storage of the sample audio database and improves the audio retrieval speed on the basis of ensuring a better retrieval accuracy.

Abstract:In order to retain more characteristic information in the training process, this study uses pre-training word vector and fine-tuning word vector to extend Bi-directional Long Short-Term Memory network (Bi-LSTM), and combines the co-training semi-supervision method to deal with the feature of sparse annotated text in the medical field. An improved model of Co-Training Double word embedding conditioned Bi-LSTM (CTD-BLSTM) is further proposed for Chinese named entity recognition. Experiments show that compared with the original BLSTM and BLSTM-CRF, the CTD-BLSTM model has higher accuracy and recall rate in the absence of corpora, the proposed method can better support the construction of medical knowledge graph and the development of knowledge answering system.

Abstract:Traditional power grid industrial control systems are mainly isolated from external networks through tools such as firewalls, but with the application of new technologies such as cloud computing and the Internet of Things, the degree of interconnection between networks has continued to deepen, and the difficulty of security protection has greatly increased. How to effectively detect network intrusion behavior has become very important. Compared with traditional intrusion detection technology, convolutional neural networks have a better ability to extract intrusion features. This study proposes a power grid industrial control system intrusion detection algorithm based on convolutional neural networks. The KDD99 dataset is processed for model training, and a cascade convolution layer is added to optimize the network structure. Under the premise of small parameter scale, the real-time requirements of the model are guaranteed. Compared with the traditional SVM algorithm and the k-means algorithm, the intrusion detection accuracy of the proposed algorithm in this study is improved, the false detection rate is reduced, and the intrusion behavior to the power grid industrial control system can be effectively detected.

Abstract:Name entity recognition is an important part in the power knowledge map construction in power dispatching field. Currently, machine learning and deep learning models are used to name entity recognition in the general field or other professional fields. In order to solve the named entity recognition in the power dispatching field, the Transformer-BiGRU-CRF model is researched. The character vector of the corpus is obtained through the Transformer model, and the named entity recognition is performed through BiGRU-CRF. There are two training methods in the training process, the first method is only to train the parameters of the BiGRU-CRF part; the second method is to train the whole model parameters including the Transformer part. Finally, it is found that the first approach reaches the stationary state in less time, but the accuracy rate is about 5% higher for the second approach.

Abstract:Data is the key basis for the stable operation of the power grid dispatching control system, in the process of data collection, the lack of data due to hardware failure and other reasons will affect the integrity of the system data, which will have a corresponding impact on the intelligence and efficiency of power grid dispatching. Therefore, the accurate prediction of missing data is of great significance for the construction of smart grid dispatching system. In order to solve the problem of missing data prediction of electric energy collection system in the field of power grid, this study improves and optimizes the existing joint prediction method based on CNN and LSTM, adds a modified model on the basis of the joint prediction model, and uses CNN convolution neural network and the unique opposite side data scene modeling in the electric power data for different missing data segments. The experimental results show that this method reduces the average absolute error value to 0.142, which improves the accuracy of the existing prediction model and accuracy guarantee for the intelligence and efficiency of power grid dispatching system.

Abstract:Garbage classification, as one of the important links of resource recycling, can effectively improve the efficiency of resource recycling and further reduce the harm caused by environmental pollution. With the development of modern industry, traditional image classification algorithm cannot meet the requirements of garbage sorting equipment. This study proposes a garbage classification model based on convolutional neural networks (Garbage Classification Network, GCNet). By constructing the attention mechanism, the model completes extracting the local and global features and can obtain perfect and effective feature information. At the same time, the feature fusion mechanism is used to fuse features at different levels and sizes, which can effectively use features and prevent gradient from vanishing. The experimental results prove that GCNet has achieved excellent results on garbage classification datasets and can effectively improve the accuracy of garbage classification.

Abstract:Data resources are important production materials that make up a digital society, and the evaluation of the value of data resources is an important basis for data transactions, data circulation, and data appreciation. Based on the theoretical basis of the Lorentz transform and PageRank algorithm, firstly, we use the PageRank algorithm to calculate the weight coefficient in the data asset pricing system, and get the initial valuation of the data asset. Then, we use the data asset valuation model of the quality-speed relationship mapping to value the given data asset. The experimental results show that the proposed data asset valuation method is of certain efficiency and market reference utility.

Abstract:To solve the problem of path planning for parallel parking in narrow spaces, a two-piece five order polynomial equation based method was proposed for calculating approaching path and reverse path respectively, among which the approaching path was applied for adjusting the vehicle pose so that a curvature optimal path for reversing car into parking lot is available. Simulation was performed by considering the general size of parking lot and family car, as well as the kinematic model of vehicle motion, to find a collision free path in free space. The result shows that the two-piece five order polynomial equation-based method can realize intelligent parking in narrow spaces, and the steering angle of wheel changes continuously on the linked path.

Abstract:Real-Time Operating System (RTOS) is an important tool in embedded artificial intelligence and IoT terminals. RTOS developed by different institutions has slight differences in real-time properties, scheduling rules, and communication mechanisms between tasks, but the basic elements are the same. This work is based on the general embedded computer(GEC) architectures, studies the RTOS resident method in BIOS and the remapping mechanism of external interface function. Taking the KL36 chip of NXP as an example, the resident implementation of mbedOS in the BIOS is given, and living example of external interface function remapping is given. Practice shows that the RTOS resides in the BIOS, which can shorten the compilation and linking time. At the same time, by remapping external interface functions, the understanding of the RTOS scheduling mechanism is simplified, the programming difficulty is reduced, and the technical basis is provided for effectively implementing the portability of applications under different RTOS.

Abstract:In the traditional collaborative filtering recommendation algorithm, similarity calculation is the core of the algorithm. However, the previous calculation method is too dependent on the user’s score, does not consider the user’s own attributes and trust relationship, and does not distinguish malicious users. In order to solve the appeal problem, this study introduces an improved new trust relationship measurement method into similarity calculation. This new method not only considers the influence of malicious users, but also combines the properties of users effectively. In addition, the study also improves the similarity algorithm on the hot issues. The algorithm finally uses the initial user clustering to get the adjacent users, effectively eliminating the cold start and data sparsity. In the experimental part, it can be proved that the proposed algorithm can effectively improve the recommendation accuracy by comparing with other algorithms.

Abstract:In order to improve the operation efficiency of the three-dimensional warehouse, aiming at stacker path scheduling problem, a stacking machine scheduling optimization model is established based on the time, energy consumption, and operation efficiency, and an Improved Multi-Objective Genetic Algorithm (IMOGA) is proposed. In IMOGA, genetic operator is improved based on NSGA-Ⅱ, crossover and mutation operations are designed for this model, adaptive genetic operator is introduced, and a local random search strategy based on the simulated annealing is added. The IMOGA is validated through the stacker scheduling situation in a spandex factory warehouse. The results show that convergence speed of IMOGA is faster, the quality of the solution set is higher, and it has higher applicability in stacker scheduling.

Abstract:In some chaos-based image encryption algorithms, the key is not related to the plaintext and the chaotic sequence has periodicity. In order to solve these problems, a new image encryption method is proposed. First, based on the plaintext image and the hash function SHA-384, the initial value of the Lorenz is generated, and the chaotic system is controlled to generate chaotic sequences. Then, the artificial neural network is introduced to train the chaotic sequence to eliminate its chaotic periodicity and output a new sequence. The scrambling and diffusion operations are performed on the plaintext image to complete the encryption. The experimental results show that the proposed algorithm is able to enhance the security of the cipher-image, increase the size of the key space and resist various attacks.

Abstract:In order to reduce the impact fluctuation of high power fast-charging pile on the power grid, and considering the advantages of Distributed Generation (DG) and energy storage of typical fast-charging stations, an optimal operation configuration method for typical fast-charging stations of Electric Vehicles (EVs) is proposed. By analyzing the power output characteristics of the DG in the station and the charging behavior law of EVs, the optimal operation configuration model of typical fast-charging station is established taking the minimum operation cost of the charging station as the optimization objective. The optimal solution of the model is solved by genetic optimization algorithm with the constraints of the power balance in the station and the power output of the distributed power supply. Finally, the feasibility of the proposed method is verified by different configuration examples to provide technical support for the optimal operation of a typical fast-charging station.

Abstract:Urban transmission cables are the lifeline of urban power supply. It is important for grid corporations to ensure the safe and reliable operation of the cables in the daily work. It still remains to be a challenge and manual inspection is difficult to quickly and effectively find and eliminate hidden dangerous points. Therefore, this study proposes a method for intelligently detecting hidden dangerous points of cables based on infrared thermal images. First, using UAV to obtain the infrared map of the outdoor terminal of the cable. Next, the infrared thermal map is binarized by an improved Bernsen algorithm. Then, the projection method is used to extract the subject cables from the binary image in order to eliminate the influence of the background. Finally, according to the intensity chromatogram, the abnormal areas with bright colors in the subject cable image are determined as hidden dangerous points. By applying this method, grid corporations can achieve rapid defect identification, elimination, and fault judgment. It can comprehensively improve the ability to manage and control the status of urban transmission cables.

Abstract:In order to reach the destination efficiently and quickly, it is necessary to find a travel path with the best comprehensive weight, and then set a guide sign on it to guide the destination. Based on this, this paper first describes the traffic network model according to the characteristics of the road network. Then, it expounds the basic concept and algorithm idea of genetic algorithm, and defines the path with the minimum comprehensive index of the driving distance and the number of intersections as the optimal path with the number of driving distance and intersections as the factors of route selection. Finally, it takes Sun Yat-Sen University in Guangzhou University City as an example under the condition that the starting and ending points are clear, the optimal path to Sun Yat-Sen University is found by using the method of genetic algorithm, which verifies the effectiveness of genetic algorithm in path planning.

Abstract:After the occurrence of destructive earthquake, compared with various urban public facilities and residential buildings built in accordance with the relatively seismic fortification standards, the vast number of villages and towns without seismic fortification houses are more likely to collapse or even completely damage. In the past, earthquake disaster risk investigation and disaster assessment relied on the field survey of experts to determine the number and proportion of buildings of different structural types. In this study, brick (concrete) wood structure houses are identified by deep learning and photography technology. The Faster R-CNN model is trained for the data set of brick (mixed) wood houses in Shandong Province of Tan Lu fault zone, with an average accuracy of 91.868%. The results show that this method can effectively detect brick (concrete) wood houses, and can be applied to earthquake disaster pre-assessment, earthquake disaster risk investigation, earthquake key-risk area investigation, and other related work.

Abstract:The establishment of domestic database selection & evaluation index system and test contents is of great significance. The first is to provide a basis for database performance comparison for the business IT systems of the government, financial banking, and central enterprises. The second is to compare different databases to guide and promote the further performance improvement of domestic database industry. Based on the software quality characteristics and security test standards, this study proposes a database selection & evaluation index system model which analyzes database evaluation factors from the perspectives of performance efficiency, information security, and other evaluation factors. It gives general test contents for each index which can be applied in specific evaluation practice. The index system model and test contents are reasonable and operable in selection and evaluation of 5 types of commonly used domestic databases.

Abstract:In view of the handwritten equation with complex two-dimensional spatial structure in the mathematics class of primary and secondary schools, this study proposes a solution of off-line handwritten equation recognition based on multiple geometric features and Convolutional Neural Network (CNN). First, based on CNN classification algorithm, the single handwritten character is recognized after image preprocessing. Then, using geometric features, such as aspect ratio, center of mass coordinate, center of mass offset angle, center offset, horizontal overlap interval ratio, etc., to recognize common handwritten formulas such as decimal, fraction, index, and root formula with complex spatial structure, and using the divide-and-conquer algorithm to complete the recognition of composite formulas nested by the above formula combination. Finally, the off-line handwritten arithmetic recognition system is designed and implemented. The experimental results show that under certain illumination conditions, the recognition rate of handwritten equation of different resolutions and noisy images can reach 90.43%, which has certain application value.

Abstract:In view of the high communication volume of traditional contract network protocols, this study proposes the Intelligent Contract network Protocol (ICP). The protocol adds a friendliness factor to the initiator of the task, a trust factor to the bidder, while two parties choose each other, and dynamically updates the friendliness and trust factors in time. It also adds a time factor to each task, and uses the global clock to control the entire model. In order to reduce the volume of communication, prevent bidders from bidding without restrictions, the protocol sets a threshold of the number of bids for bidders; and sets the number of unbid tasks. Based on historical bid records, various parameters are intelligently adjusted. This study uses colored Petri net to model and simulate ICP. The experimental data proves that ICP greatly reduces the communication volume, shortens the running time, and improves the task hit rate.