Abstract:Vascular stenosis caused by atherosclerotic lesions is the biggest cause of coronary heart disease, and its incidence is high, and the mortality rate is high. Therefore, it is very important to study the degree of coronary artery stenosis for early diagnosis and evaluation of coronary heart disease. Digital Subtraction Angiography (DSA) images are the “gold standard” for the diagnosis of coronary heart disease. Medical aids in the treatment of DSA angiography to assess the degree of stenosis, first need to segment the blood vessels in order to carry out subsequent analysis of stenosis. The segmentation and extraction of blood vessels is an important premise for quantitative description of diseases and three-dimensional reconstruction of blood vessels, and is also an important means to assist doctors in clinical diagnosis and treatment. In this paper, we study the digital subtraction angiography (DSA) images of cardiovascular, and summarize the methods of segmentation of blood vessels in angiocardiography in recent years from three aspects: preprocessing, segmentation method and evaluation criteria.

Abstract:The development of neural network compression relieves the difficulty of deep neural networks running on resource-restricted devices, such as mobile or embedded devices. However, neural network compression encounters challenges in automation of compression, conflict of the sparsity and hardware deployment, avoidance of retraining compressed networks and other issues. This paper firstly reviews classic neural network models and current compression toolkits. Secondly, this paper summarizes advantages and weaknesses of representative compression methods of parameter pruning, quantization, low-rank factorization and distillation. This paper lists evaluating indicators and common datasets for the performance evaluation and then analyzes compression performance in different tasks and resource constraints. Finally, promising development trends are stated in this paper as references for promoting the neural network compression technique.

Abstract:At present, complex network has quickly formed a cross-disciplinary discipline across many fields, and its relevant theories have been applied to many fields. To understand the research status of complex networks, starting from the definition and statistical characteristics perspectives of complex network, we introduce the basic concept of complex network, then enumerate several typical complex networks model, and on this basis, we set up the improved models and discuss their advantages and disadvantages. Focusing two aspects of structural characteristics of complex networks and network dynamics, we further analyze the current status of research on complex networks, and the research achievements of recent years are listed. Finally, the future research direction of complex network is summarized and forecasted.

Abstract:Understanding the spatial correlation of places plays an important role in geographic information retrieval and recommendation systems, urban traffic management, and resident travel pattern exploration. In order to represent the places and their spatial relationships specifically, we propose a deep learning-based vectorization method for places. The correlation between places can be calculated by the place vectors. Firstly, the trajectories of long-distance and short-distance are matched and connected to build a large-scale traffic network, which could cover multiple travel modes and obtain a complete cognition of spatial relations. Then we propose a spatial vectorization method which is based on graph neural network and combines place features and trajectory information. Besides, we improve the representation ability of latent representations for places by optimizing a node sampling method. Finally, the empirical analysis is performed on the shared bicycle track data and public traffic data in Beijing. The result demonstrates that the proposed method outperforms the existing methods such as DeepMove on place correlation analysis and cluster analysis.

Abstract:In the software testing, coverage of test cases is one of the important prerequisites to find software defects. In this study, the experiment method in software engineering was used to analyze the coverage of program modules and program branches. Based on the Defects4J dataset, Evosuite and Randoop tools were used to generate test cases under different generating time limits. When the generation time exceeded 20 s, the numbers of test cases produced by Randoop was more than that produced by Evosuite, but the coverage of Evosuite test cases was significantly higher than the coverage of Randoop. At the same time, this study also analyzed the factors affecting the coverage. It is a good reference for how to use these tools to generate high coverage test case and for the improvement of tools.

Abstract:How to effectively identify the intrusion attack behavior of the Industrial Internet of Things (IIOT) is a new challenge. Aiming at the problems of low intrusion detection feature extraction, low detection efficiency, and poor adaptability in IIOT, an intelligent intrusion detection method based on deep learning is proposed. First, improve the sampling algorithm in data processing for adjusting the number of samples in a few categories to improve the detection accuracy. Second, build a stacked denoising convolutional self-encoding network to extract key features. Combine the convolutional neural network and the denoising self-encoder to enhance feature recognition ability. In order to avoid information loss and information ambiguity, improve the pooling operation to increase its adaptive processing ability, and use Adam algorithm to obtain the optimal parameters during model training. Finally, use the NSL-KDD dataset to test the performance of the proposed method. Experimental results show that the accuracy of the method is 3.66%, 4.93%, and 0.04% higher than the existing RNN, DBN, and IDMBCNN, respectively. Compared with the SDCAENN test without sampling algorithm, the detection accuracy of U2R and R2L is improved by 17.57 % and 3.28%.

Abstract:In this study, the adaptive low rank denoising based Magnetic Resonance Imaging (MRI) reconstruction method is proposed. This method uses denoising-based approximate message passing algorithm to reconstruct MR images. The adaptive Weighted Schatten p-Norm Minimization (WSNM) method is used as its noise reduction model to study the reconstruction performance of the MR images. And the image block size and the number of similar blocks of WSNM are set adaptively according to the noise standard deviation estimated during the algorithm iteration process. Compared with the MR image reconstruction algorithms proposed in recent years, the experimental results show that the proposed method can get higher Peak Signal-to-Noise Ratio (PSNR) and lower Relative L₂ Norm Error (RLNE) and have the best reconstruction performance.

Abstract:Jobs’ temperature will decrease as their starting times delay in heat treating flow shop. In order to process the jobs normally, the worker has to reheat them or keep them in holding furnace. In response to this situation, this study considers a parallel machine scheduling problem with step-deteriorating jobs for minimizing the total tardiness and energy consumption. Firstly, a mixed integer programming model is proposed for the problem under study. Due to the intractability of the problem, a genetic -variable neighborhood search hybrid algorithm is designed to solve it. The algorithm use genetic operations to generate the solution, and then use the variable neighborhood search to improve the solution. The numerical tests show the proposed algorithm can efficiently reduce the total tardiness and energy consumption compared with standard genetic algorithm and mathematical model with standard solver Gurobi.

Abstract:According to the actual demand of provincial operation service, the provincial environment meteorological operation system was designed and developed. This paper described the system objectives, overall architecture, functional composition, and main key technologies. Based on the standardized message middleware technology, the key technologies such as unified grid data environment and intelligent grid forecast analysis service were constructed, and the operation system composed of integrated comprehensive monitor, forecast interactive analysis, product production and release was established. The operation system is stable and effective. The realization of the system promotes the standardization level of operation services and improves the guarantee ability of ecological environment meteorological services.

Abstract:In order to solve the problem of power grid supply area analysis and efficiency of large-scale data sets in the dispatch and control cloud platform, a method of power grid supply area analysis based on the dispatch and control cloud model is designed in this study. Firstly, this study defines the data model of dispatch and control cloud power grid supply area. Secondly, on the basis of the data model, a method of power grid supply area analysis is designed and applied. Finally, the experiments based on the real model and topological data of the dispatch and control cloud in a regional power grid is verified. The experimental results show that the computing method of power grid supply area proposed in this study can analyze the power grid supply area in a better way, and meet the needs of the actual power grid dispatching.

Abstract:In recent years, how to deliver reliable products to users quickly has become a hot issue of continuous delivery research and application. Traditional agile software methods lack of team cooperation and standardized construction process in the delivery process. Due to the complexity of the system, the DevOps framework of large companies will produce the contradiction between iteration speed and product quality when applied in small and medium-sized enterprises. This study proposes a method based on DevOps, which is a lightweight continuous delivery framework. In the project scenario with overlapping roles and frequent iterations, it automatically achieves project code acquisition, testing, construction, and deployment in the form of script, and completes the continuous delivery of the project. Through industry survey and enterprise practice, the scheme can not only shorten the project cycle and improve the quality of delivery, but also realize the visualization of delivery process and promote the continuous improvement of software quality.

Abstract:The Stewart platform is widely used in motion simulator, optics, precision positioning, and other fields. However, due to the complex multivariate nonlinearity, it is difficult to accurately obtain the forward kinematics solution. Aiming at the problem of forward kinematics solution of Stewart platform, conventional methods such as iterative method and numerical method have problems such as difficulty in selecting initial values and slow calculation speed, a forward kinematics solution method based on Elman neural network is proposed. First, the kinematics model of the leg length and platform kinematics of the Stewart platform is established, and then the Elman neural network is used to solve the forward kinematics solution and experimentally verify it. This method has sound dynamic characteristics, high accuracy, and can quickly and accurately solve the forward kinematics solution of Stewart platform. Experiments prove the effectiveness of the method.

Abstract:The spare items trade among college students is one of the main activities of college extracurricular life. Due to the convenience of in-person trading, high reuse rate, low price, environmental conservation, and other reasons, the spare goods trading on campus is favored by college students. The information of spare items in traditional schools is released through QQ group, which has no security guarantee, information cannot be integrated, and information within the group is easy to be ignored. In order to bring better spare items trading user experience, the platform builds the front end of the spare item information platform based on the QQ mini-program, and builds the back end of the platform based on the Flask+MySQL frame, which has fully realized all kinds of functions of second-hand trading information platform, including multi-dimensional commodity classification, commodity recommendation, message reminder, price reminder, one-click to add friends, and so on. The platform achieves the front end and the back end encryption interactions through HTTPS and RESTful API, and introduces the function of school identity authentication, which perfectly meets the various functional requirements while ensuring the security of campus spare items trade.

Abstract:Focused on the issue of the complexity and uncertainty of high-altitude inspection environment, a semantic interaction method of UAV navigation for transmission line inspection is proposed. Firstly, a semantic topology of human-UVA interaction based on ontology is structed to form a navigation framework with perceptual motion data processing. Then, an algorithm of structured route navigation is proposed by combining the background knowledge of predicated logic form and semantic topology based. Lastly, a semantic interaction demonstration system is developed and the ideal results of the experimental are obtained. As a case study of semantic interaction, this research proves that the entity noun extraction satisfies actual expectation estimation and has high precision control, which can provide high position navigation control UAV transmission line inspection. It also can provide basic support for data-driven transmission line inspection.

Abstract:The traditional warehouses use manual or man-machine methods to transport goods, use paper and pen to record related inventory and data of incoming/outgoing, which has many problems such as low efficiency, many security risks, and high employment cost. The development of modern industry and commerce has put forward a higher demand for warehouse storage technology, which is not only limited to the cost category, but also becomes a strategic tool to obtain profits. This study introduces the whole structure, the functions of each subsystem and the design of the main parts of the automated three-dimensional warehouse system. In order to further improve the production efficiency and improve the overall profitability of the company, this paper, based on the laboratory automated three-dimensional warehouse system, applies Apriori algorithm of association rule data mining, mines the outbound information, and provides more scientific suggestions on production and outbound. The system has the characteristics of high automation, intelligence and information management, which is of great significance to improve the efficiency of warehouse management and increase the profits of enterprises.

Abstract:The assembly of some parts during the assembly process of an automobile gearbox involves the process of selecting and adjusting gaskets. Whether the gasket size is appropriate will directly affect the overall assembly quality of the gearbox. In order to meet the production requirements of the new model gearbox and meanwhile improve the precise ability of the production system, it is necessary to transform the traditional manufacturing mode dominated by manual experience into a lightweight, intelligent new manufacturing model. In order to improve the digitization of the production site and the ability to manage the measurement tasks, the research has designed and implemented a new measurement system for gearbox assembly. The system is based on MQ250 gearbox assembly line. The research begins with the whole structure, and introduces the measurement module of the software system. First, based on the principle of relative measurement, a measurement module was designed, and the optimization problem of the measurement module was analyzed. Finally, the optimization of the measurement module was achieved by the least square method. The measurement system adapts to the assembly requirements of MQ250 gearbox components, and improves the assembly quality under the condition of ensuring assembly efficiency, and greatly improves the digital level of the assembly line while ensuring the assembly efficiency.

Abstract:Technology stack separation during development under the microservice architecture improves development efficiency and runtime service orchestration capabilities. However, the multiplication of micro-applications results in the increase of integration complexity of micro application management console and the poor experience of communication and operation interaction across micro-applications. This paper presents a micro frontends solution, combined with typical micro frontends framework, management console eventbus, micro-applications routing communication, separation of context resources in runtime and other technical mechanisms, the flexibility and operating efficiency of the management console is improved, the development and operation and maintenance costs are reduced, and the management optimization objective of the management console is better realized. Through the test evaluation, the flexibility is enhanced and the cost is smaller, the effectiveness of the solution is verified in the actual project.

Abstract:Power distribution network accident in actual application scenarios account for more than 80% of total grid accident, and the prediction of power distribution network accident has always been a difficult issue. This study, under the call of “Ubiquitous IoT” proposed by the State Grid, analyzes the research results of scholars on this issue, and proposes an accident prediction method for power distribution network based on graph neural network with the idea of graph neural network. Referring to the commonly used graph neural network design framework, the node information aggregation function, prediction function, and loss function are designed in detail, and reasonable depth parameters are selected according to the algorithm flow test. The algorithm fully considers the mutual influence between connected nodes, and uses the real grid operation data to compare the two other algorithms commonly used in this field. Experiments show that the proposed algorithm improves the accuracy by 3.0% and is more robust.

Abstract:Text style transfer is always a hot spot in Natural Language Processing (NLP). In recent years, as the development of sequence generation methods, many researchers focus on style transfer on non-parallel corpora. Specifically, this task wants to change the style of the sentence while keeping the original content. To achieve this target, many works have been proposed which based on the generative adversarial network. But due to the instability of adversarial training and the limitation of the independence assumption between the style and semantic information, these methods are hard to learn an effective and efficient transfer model. In this study, motivated by statistic learning methods, a definition of the text style is given. The style of the corpus can be captured by the covariance matrix of its sentences’ semantic vectors. From this perspective, the text style is dependent on all the semantic information. We then propose a learning free transfer method where the only thing we need is a pre-trained auto-encoder to produce the semantic vectors. With a pair of matrix transformations, including whitening transformation and stylizing transformation, performing on these vectors, we achieve text style transfer.

Abstract:With the rapid development of deep learning technology, more and more intelligent algorithms have been applied. The hardware equipment used for training and calculation is mainly GPU, which will incur high hardware procurement cost and power consumption cost in the actual deployment and use. Therefore, aiming at the high cost of the current deep learning system, this study proposes to use raspberry PI and Movidius neuron computing stick as the computing platform. SSD+MobileNet algorithm is adopted to realize the recognition and detection of vehicle targets, and the training model is tested and optimized in the actual environment to finally meet the effect of actual use, with a processing speed of 4 frames per second on average. The experimental results show that on the platform with weak computing power like raspberry PI, the algorithm can be accelerated by VPU modules like Movidius neuron computing stick, and the computing cost can be greatly reduced when it is in actual use.

Abstract:SSD (Single Shot multi-box Detector) algorithm is used to detect multi-scale objects on feature maps of different layers, which has the characteristics of fast speed and high accuracy. However, the feature pyramid detection method of traditional SSD algorithm is difficult to fuse the features of different scales, and because the convolutional neural network layer at the bottom has weak semantic information and is not conducive to the recognition of small objects, so this paper proposes a novel object detection algorithm RF_SSD based on the network structure of SSD algorithm. In this algorithm, feature maps of different layers and scales are fused in a lightweight way, and new feature maps are generated in the lower sampling layer. By introducing the receptive field module, the feature extraction ability of the network is improved, and the characterization ability and robustness of the feature are enhanced. Compared with the traditional SSD algorithm, the accuracy of the proposed algorithm is significantly improved, and the real-time performance of object detection is fully guaranteed. The experimental results show that the accuracy is 80.2% and the detection speed is 44.5 FPS on the PASCAL VOC test set.

Abstract:In view of the variety of buildings, and the confusion with the surrounding environment in the remote sensing image, the traditional methods are difficult to extract the buildings efficiently and accurately. This study proposes a method of building automatic extraction method based on the improved Mask-RCNN. In the proposed method, an improved Mask-RCNN network model framework is constructed by using PyTorch deep learning framework, path aggregation network and feature enhancement function are added in the network design, and multi-threaded iterative training and model optimization learning are conducted for the Inria aerial image tag data set by means of supervision and migration learning, thus automatic and accurate segmentation and extraction of buildings are achieved. The proposed method is compared with SVM, FCN, U-net, Mask-RCNN, and other building extraction algorithms with different open-source datasets. The experimental results show that the proposed method has competitive performance, which can extract buildings more efficiently, accurately, and quickly in different open-source datasets, and the four evaluation indexes of mAP, mRecall, mPrecision, and F1 scores extracted in the same dataset are better than the compared algorithms.

Abstract:Deep learning is a new filed in machine learning research, and to apply it to computer vision achieves effective result. To solve the problem that the traditional Scale-Invariant Feature Transform algorithm (SIFT) has low efficiency and extracts image features roughly, A SIFT image retrieval algorithm based on deep learning is proposed. The algorithm idea is that on the Spark platform, a deep Convolutional Neural Network (CNN) model is used for SIFT feature extraction, and Support Vector Machine (SVM) is utilized for unsupervised clustering of image library, then the adaptive image feature measures are used to re-sort the search results to improve the user experience. The experiment results on the Corel image set show that compared with the traditional SIFT algorithm, the precision and recall rate of the SIFT image retrieval algorithm based on deep learning is increased by about 30 percentage points and the retrieval efficiency is improved, the resulting image order is also optimized.

Abstract:Mobile phone text messaging has become an increasingly important means of daily communication, so the identification of spam messages has importantly practical significance. A self-attention-based Bi-LSTM neural network model combined with TFIDF is proposed for this purpose. The model first inputs the short message to the Bi-LSTM layer in a vector manner, after feature extraction and combining the information of TFIDF and self-attention layers, the final feature vector is obtained. Finally, the feature vector is classified by the Softmax classifier to obtain the classification result. The experimental results show, compared with the traditional classification model, the self-attention-based Bi-LSTM model combined with TFIDF improves the accuracy of text recognition by 2.1%–4.6%, and the running time is reduced by 0.6 s–10.2 s.

Abstract:Semantic segmentation of deep learning has a very broad development prospect in the field of computer vision, but many network models with better segmentation effects take up a lot of memory and take a long time to process a single picture. In response to this problem, we replace the bottleneck unit of the Deeplab V3+ model backbone network (ResNet101) with a 1D non-bottleneck unit, and decompose the convolutional layer of the Atrous Spatial Pyramid Pooling (ASPP) module. The algorithm can greatly reduce the parameter amount of Deeplab V3+ network and accelerate the speed of network inference. Based on the PASCAL VOC 2012 dataset, the experimental results show that the improved network model has faster speed and better segmentation, and takes up less memory space.

Abstract:Together with the particle swarm optimization algorithm, a physical programming-based multi-objective optimization model is developed to seek the optimal strategy for spare parts arrangement. The levels of satisfaction and the preference functions as well as the aggregate objective function for the supportability are designed which can reflect the preference of decision makers. With the proposed optimization method, the computational burden in large-scale multi-objective design problems can be greatly reduced. Meanwhile, the results from the proposed method are compared with that of single-objective optimization, demonstrating the effectiveness of the proposed model.

Abstract:Automatic image captioning is a hot topic which connects natural language processing and computer vision. It mainly completes the task of understanding image semantic information and expressing it in the form of human natural language. For the overall quality of Chinese image captioning is not very high, this study uses FastText to generate word vector, uses convolution neural network to extract the global features of the image, then encodes the pairs of sentences and images〈S, I〉, and finally merges them into a feature matrix containing both Chinese description and image information. Decoder uses LSTM model to decode the feature matrix, and obtains the decoding result by calculating cosine similarity. Through comparison, we find that the model proposed in this study is better than other models in BiLingual Evaluation Understudy (BLEU). The Chinese description generated by the model can accurately summarize the semantic information of the image.

Abstract:Equipment health is a quantitative assessment of the health status of a device and can more accurately reflect the real health status of the device. Existing health assessment methods do not reflect the health status of the device. Aiming at the health evaluation of a certain type of radar transmitter, an accurate health assessment method based on dynamic weight is proposed. In this study, the exponential function is used to model the dynamic variation of the weight of each parameter, and then the weighted standard Euclidean distance between the acquisition vector and the best vector is calculated, which can evaluate the health status of the equipment more accurately. The actual test of a certain type of radar transmitter shows that the calculation result of this method is accurate and reliable, and can better reflect the health status of the equipment.

Abstract:Transformation of silica complex melting process to its data information establishes a reliable timing rule for the high temperature melting process, and is of great significance for improving the direct fiber forming technology of blast furnace slag. Firstly, hierarchical clustering algorithm is used to detect the edge of gray image. The region growing and morphological processing are used to segment the image, so as to determine the position coordinates of the silica particle’s center of mass and draw the motion track. Secondly, the area and the generalized radius are selected as the indexes of edge contour feature. The experimental results show that the relationship between the area, generalized radius and time of silica and the fitting degree of quadratic curve is the highest. The above two indexes can be used to represent the melting process of silica and to estimate the actual melting rate of silica.

Abstract:With the vigorous development of social economy, the demand for large buildings such as subways, tunnels, and bridges is growing. Through analyzing the structural deformation data, it can judge the future development trend of the structure so that emergency measures can be taken in advance to prevent the occurrence of disasters. Due to the instability and nonlinearity of deformation monitoring data, the prediction of monitoring data has become a problem in structural monitoring researches. Aiming at the problems of structural deformation prediction models, a long and short-term memory network (LSTM) structural deformation prediction model is proposed based on orthogonal parameter optimization. The long-term memory of the time series can be obtained through the LSTM network structure, and the internal time characteristics of the structural deformation data can be fully mined, the parameters of the LSTM model can be optimized through the orthogonal experiment. Finally, the model was verified by measured data. Experimental results show that the predicted value of the model is closed to the actual monitoring value. Compared with the WNN, DBN-SVR, and GRU models, the average RMSE, MAE, and MAPE are reduced by 56.01%, 52.94%, and 52.78%, respectively. The LSTM structural deformation prediction model based on orthogonal parameter optimization proposed in this study is an effective structural settlement method, which provides reliable information for the safe construction and operation of the structure, and is of great significance to ensure the safety of the structure.

Abstract:Correction of image distortion is one of the problems in the field of image processing, especially in the field of automated inspection. A polynomial distortion imaging model is used to establish the conversion relationship between coordinate systems. Based on the LabView platform, the calibration dot matrix is used to calibrate the machine vision system, and the bilinear interpolation algorithm is used to correct the image in different working modes. The experimental results show that the acquired images have been well corrected and the expected results have been achieved.

Abstract:A dynamic gesture recognition algorithm based on convolutional neural network and support vector machine classification (CNN-Softmax-SVM) is proposed to solve the problems of low recognition rate and few gesture recognition types in monocular vision. Firstly, the fast fingertip detection and tracking algorithm based on YCbCr and HSV color space is employed, which can acquire fingertip trajectory in real time under complex background. Secondly, fingertip trajectory is used as input of joint CNN-Softmax-SVM network, and finally dynamic gesture trajectory is recognized by trained network. The test results show that the combined CNN-Softmax-SVM algorithm can identify the dynamic gesture trajectory well.

Abstract:When the CascadeClassifier cascade classifier provided in OpenCV uses Haar features for face detection, the detection speed is too slow to meet the real-time requirements of the video, and the impact of lighting is also great. Based on these two points, a new face detection algorithm is proposed, which uses Camshift target tracking and face detection to improve the detection speed and uses histogram equalization to reduce the impact of light. The algorithm first sets the face area detected by the CascadeClassifier cascade classifier method as the ROI area, operates on the ROI area and uses the Camshift algorithm for target tracking, and secondly performs face detection regularly to update the ROI area to ensure the tracking accuracy. The analysis of the experimental results shows that: with the improved algorithm, the speed of face detection has been significantly increased (about 40%), and the impact of light is reduced.

Abstract:Blockchain has the characteristics of decentralization, security, reliability, and immutability, and has received widespread attention recently. Merkle tree is the core component of the block, accounting for more than 96% of the block storage. It is mainly used to handle the problem of simplified payment verification in Blockchain transactions. Therefore, choosing the appropriate Merkle tree structure will greatly affect the performance of the Blockchain. However, there is no public platform to analyze and verify the performance of Merkle tree under different Blockchain systems at present. In this study, we propose a set of related performance evaluation and analysis indexes in terms of storage, verification, and build time. The performance of the Merkle tree of the three mainstream Blockchains of Bitcoin, Ethereum, and Hyperledger is evaluated. The index and evaluation method proposed in this study not only provides quantitative data support for further research on Merkle trees, but also provides guidance for Blockchain practitioners in choosing Merkle tree structures.

Abstract:Sales forecasting has always been a hot research topic and has great significance for all enterprises. In recent years, with the rise of deep learning, there are more and more models for sales forecasting, and the performance of single models is often not ideal. Therefore, there are more and more combinatorial models. In this study, we use Stacking strategy to support XGBoost, Support Vector Regression (SVR), GRU neural network as the basic model, then lightGBM as the final prediction model, with new features are merged. The advantages of several models are condensed, which greatly improves the prediction performance of the model, good enough to be more close to the real sales data, and provide a new prediction method for regression prediction.

Abstract:The combine model of the RFM model and K-means is used to classify customer value and AHP method is mostly used to determine the weight of indicators, without considering the relationship between the indicators of RFM model. In this study, firstly, we select the average time interval, the customer purchase frequency in a period of time, average transaction money of each order, and customer active time to structure RFMT model in order to measure the customer value. Then, determine the index weight by using grey correlation degree. Finally, aiming at the shortcomings of K-means, K-means ++ and elbow law are used to carry out cluster analysis of RFMT model. This model can make a more detailed division of customer base. It can help e-commerce enterprises to identify the customers that need to be focused on. Meanwhile, the enterprise customers can be divided into customer groups with high value to low value, and put forward specific marketing suggestions for different customer groups.

Abstract:Since the existing Practical Byzantine Fault Tolerance (PBFT) consensus algorithm applied to the consortium Blockchain has the problems of poor scalability, high communication overhead, and low efficiency, the Credit-Layered Byzantine Fault Tolerance (CLBFT) consensus algorithm was proposed. Based on the PBFT, a node credit score rule was formulated. A mechanism based on credit-layered was proposed, which divides the nodes into four categories, to enhance the initiative of trusted nodes and reduce the participation of abnormal nodes in order to achieve the purpose of good system operation. The experimental results show that under long-term operation, CLBFT can reduce communication overhead and improve system efficiency.

Abstract:In the environment of intelligent manufacturing and “Industry 4.0”, the intelligent manufacturing line as an important carrier of intelligent manufacturing has high research significance in its algorithmic IT operations, condition monitoring, and data acquisition. This article designs an identity authentication method called R-ECC (Random-Elliptic Curve Cryptography) for the mobile terminal of intelligent manufacturing line, which is based on the security model provided by OPC UA. The method addresses the problems of limited communication resources and high security requirement of data transmission in the process of communication between an Android-based mobile terminal and OPC UA server. This method introduces random numbers and elliptic curve cryptography in the authentication process, which improves the security of OPC UA identity authentication and reduces the consumption of communication resources in the same time. The experimental results show that the R-ECC identity authentication method for the mobile terminal of intelligent manufacturing line can effectively improve the security of the authentication process, reduce the hardware resource consumption of the mobile terminal, and accelerate the speed of identity authentication.

Abstract:Aiming at the problems of low recognition accuracy and low recognition efficiency in the current abnormal flow detection methods of industrial control network, combined with the periodic characteristics of industrial control networks, this study proposes an abnormal flow detection model based on Long-Short Term Memory network (LSTM) time series prediction. This model takes the LSTM network model as the core, and uses the normal historical traffic sequence of the first 15 minutes to predict the traffic data at the next moment. On the premise that the accuracy on the test set is 98.12%, the model’s predicted value can be considered to be normal. By comparing the actual value with the predicted value, it is determined whether there is an abnormality. On the premise of not reducing the recognition accuracy rate, because the predicted value is calculated in advance, this method greatly improves the detection efficiency.

Abstract:To improve SARS-CoV-2 information emergency response capability in hospitals, an emergency response method based on information systems was proposed. Relying on computer technology, Internet technology, and 5G technology, it has constructed an online emergency treatment area, established a current limiting management platform for outpatient and emergency, and conducted remote consultation for COVID-19 patients. It has realized the application of the whole process of online consultation, online prescription, and drug delivery for patients with chronic diseases and special diseases, established remote consultation services for COVID-19 patients between institutions, adjusted outpatient number allocation algorithms, and reduced cross infection rate. Adopting the information system of emergency measures, it has achieved the hospital from online to offline complete coverage, enhanced the control effect of SARS-CoV-2.