In order to improve the monitoring and early warning efficiency of network public opinion, and to reveal the spread of network public opinion, an evolution process model of public opinion in the Internet is proposed, It tries to analyze the evolution mechanism of network public opinion in the Internet, and provides certain theoretical support and method guidance for network public opinion monitoring and prediction. At the semantic level, the implementation method of knowledge fusion of different levels of public opinion information resources is proposed, which is supported by semantic technologies such as the semantic web in the big data environment. Then the multi-agent modeling and simulation method is used to establish the network public opinion information communication simulation model. The attributes of each participating entity were constructed in the model, and the influence of various factors on the network crisis information transmission was analyzed. The experimental results show that the proposed verification simulation model has high credibility. By analyzing the number of "opinion leaders" in the model, the participation of netizens, the credibility of the government, the speed of information disclosure and the transparency of the public, it can improve the monitoring and early warning efficiency of network public opinion, and better reveal the propagation law of network public opinion.

Dual-driving powertrain system of pure electric vehicle has two power sources and the main function of power control strategy for finished vehicle is: power distribution in order to control operating condition of all components for the powertrain system according to operating characteristics of powertrain system and components of finished vehicle as well as driving status data of vehicles. In overall consideration of the impacts of efficiency of all components for powertrain system and system efficiency in different operating patterns on efficiency of finished vehicle, control strategy of pattern division and power distribution is based on optimum efficiency of driving system in order to guarantee basic performance of finished vehicle and increase driving range of finished vehicle. MATLAB/Simulink simulation analysis software is used to establish simulation platform for powertrain system of pure electric vehicle with two driving force and to carry out simulation verification for control strategy of finished vehicle. The results show that all power components can be put into synergetic operation and vehicles will have higher performance, energy saving effects as well as efficient and reasonable control with the control strategy which has been developed.

Short-term load forecasting plays a key role in the safe dispatching and economic operation of the power system. The lease square support vector machine (LSSVM) has the power system. The least square support vector machine (LSSVM) has great potential in forecasting problems, particularly by employing an appropriate algorithm to determine the values of its two parameters. In order to improve LSSVM load prediction accuracy, this paper proposes a LSSVM based on the Variational mode decomposition(VMD) electric load forecasting model that uses an artificial fish swarm-shuffled frog leaping algorithm to determine the appropriate values of the two parameters. The historical data such as load and weather in the first 15 days of the forecast day are the input into LSSVM. The AFSA-SFLA-LSSVM forecasting model, the LAVAFSA-SFLA-LSSVM forecasting model, the AFSA-LSSVM forecasting model, and the VMD-LAVAFSA-SFLA-LSSVM forecasting model were established for electrical load forecasting in a certain area within 24 hours of a specific day. The results of the example show that the accuracy of the VMD-LAVAFSA-SFLA-LSSVM forecasting model was higher than the other three forecasting models and the prediction error was smaller as well.

Existing image steganalysis methods based on deep learning assume that the embedding rates are known, whereas for most practical applications, these rates are unknown, leading to a sharp drop in model detection performance. This study combined transfer learning (TL) and multi-task learning (MTL) and proposed an image steganalysis method for a specific steganographic algorithm and unknown embedding rates. The proposed method used stego images with high embedding rates to pre-train the steganalysis model, constructed a steganalysis model based on MTL, and then transferred the parameter values of the pre-trained model as the initial values. The parameters were further fine-tuned on the training set, which consists of cover images and stego images with various embedding rates. A new objective function was designed by applying the weighting losses to the uncertainty method, dynamically adjusting the weight of each sub-task during the training process. The proposed method extracted the common features of images with various embedding rates more effectively, achieved better detection accuracy on images with unknown embedding rates, and demonstrated improved generalization ability.

For the issue of resource load prediction in cloud computing, a modified artificial bee colony algorithm and SVM are combined to construct a predictive model. First, by using reverse learning to initialize the population, differential evolution selects the individual population. The point strategy is used to construct the honey source selection route of the algorithm. The feedback mechanism reduces the shortcomings of the algorithm falling into the local optimum. Second, the parameters in the SVM prediction model are optimized and the best ones are found by using the improved bee colony algorithm. In the final simulation experiment, the proposed IABC algorithm has better prediction accuracy than the SVM, the LSSVM and other prediction algorithms, and so it has a certain promotional value.

This paper presents an algorithm for detecting overlapping communities in complex networks. The algorithm draws on the idea of the clique as the core of the community, and proposes to treat the overlapping community as a collection of all k-cliques. The algorithm uses random nodes as the initial community, and each iteration selects the node with the maximum fitness value of the community neighbor. All k-cliques of the node are added to the community. During the process, nodes with negative fitness are removed. It then realizes the partition of network community structures and detects overlapping nodes. In many experiments of computer-generated networks and real-world networks, algorithms based on this idea have achieved good experimental results, which also illustrates the feasibility of this idea. Furthermore, the time efficiency and complexity of the algorithm is also acceptable. This algorithm also has better community discovery results.

Fault diagnosis is important for the working conditions of 3D printers, because the failure of 3D printers will have a great impact on the quality of printed products and result in unqualified printing. In this paper, the reservoir computing (RC) method and the data collected by the attitude sensor are analyzed to obtain the health status of a 3D printer. Considering the economics and viability of fault diagnosis, a low-cost attitude sensor is installed on the moving platform of the 3D printer to collect tri-axial angular velocity, tri-axial acceleration, and tri-axial magnetic field strength signals. Then, the collected data is divided into training data and test data. The training data is used to establish the optimization parameter of the RC model to improve its performance, and the test data is used to identify the failure patterns using the model. Finally, compared with the SAE and SVM intelligent diagnosis techniques, the RC method achieves the best fault recognition accuracy, which further proves its superiority.

Manual program debugging is tedious as well as time-consuming. The high costs have motivated the development of automatic program debugging approaches, which mainly focus on helping programmers identify fault locations. Xie et al. revisited automatic debugging via human focus-tracking to validate its effectiveness. However, their observations implied that there exists interference between the mechanism of the automated debugging approach and the actual assistance needed by programmers during program debugging. To solve this problem, we propose a search-based software debugging approach based on weighted fault propagation graphs (WFPG). We firstly use spectrum-based fault localization techniques to generate a suspicious module-fault ranking list and then construct a WFPG for each suspicious program module to assist programmers in understanding fault propagation. Our approach integrates automatic fault localization and program understanding to help programmers debug. We conduct a case study to demonstrate the effectiveness of our approach, and the results are promising.

With the increasing application of computers and networks, several network security problems have emerged, and thus network intrusion detection systems have become the focus of network security research. In order to achieve the purpose of intrusion detection and protection, the traditional network intrusion detection system extracts features from the data of the network data stream according to the feature recognition algorithm and compares the extracted features with those in the training set to recognize the behavior. However, if a user wants to effectively detect malicious behaviors in the network, a large feature library is needed, and it cannot be shared with other users, which makes the quality of single user detection lower than the highest detection quality of the whole network. Blockchain, which is a new network system of decentralization, de-trust, tamper-proof, anti-counterfeiting, and traceability, plays an important role in the transmission and sharing of high-value data. In this paper, a new network intrusion detection system is designed based on blockchain, which can enable users to share feature libraries over the whole network by means of P2P network transmission. Meanwhile, its network structure and consensus algorithm are presented, and its security and performance are analyzed. Analysis results show that this system has lower false negative rates.

The influence of intelligent fault diagnosis on industrial development is becoming more and more important. In order to study the fault diagnosis technique of delta 3D printers using extreme learning machine (ELM), a low-cost attitude sensor was used in our designed machine. In the research process, the cross validation method was used to train ELM to obtain the optimal model. Through the analysis of different activation functions, we found that the correct recognition rates corresponding to the same activation function are different, and there are great differences among training samples and fault categories. The sin function, mexihat function, and tribas function recognition effects were better. The analysis of different activation functions revealed that the correct recognition rates corresponding to the same activation function are different, and there are great differences in different training samples and fault categories.

Knowledge graph inference has a wide range of applications in semantic search, question answering systems, entity disambiguation, link prediction, and recommendation systems. However, the accuracy and operational efficiency of existing methods do not meet the needs of large-scale knowledge graphs. Aiming at the problem of large-scale knowledge graph inference, this paper proposes a knowledge graph inference method based on probabilistic soft logic (KGIPSL). Firstly, KGIPSL uses the Markov logic network to construct the relationship between entities. Secondly, KGIPSL employs probabilistic soft logic to represent non-deterministic knowledge and infers the relationship between entities in the knowledge graph. Thirdly, KGIPSL conducts accurate knowledge inference. Experiments on real knowledge graph datasets show that the KGIPSL method is superior to the existing baseline method in accuracy, recall, and efficiency. Among them, the average accuracy of KGIPSL on the YAGO dataset is 14.9% higher than that of the baseline method.

3D-high efficiency video coding (3D-HEVC) structure is a development of HEVC, but some new coding techniques are added on the basis of it to make it more conducive to encoding depth maps and multi-view. In 3D-HEVC, the intra prediction mode decision for depth level contains closed connections with coding unit (CU) partition. This process, quad-tree block splitting, gives the gorgeous coding efficiency to 3D-HEVC, but it incurs unacceptable computational burdens because each possible coding mode is tested to rank the most suitable one. According to previous works, whether the current CU will be divided into smaller sizes is dependent on encoding contexts. In view of that, this paper proposed a novel method to speed up intra coding unit splitting, relying on data clustering and data mining. The experimental results showed that our new approach can reach a satisfied balance between computational burdens and RD cost.

Cloud computing is different from distributed computing and grid computing and has its own characteristics. The existing cloud system is not sufficient in the unified identification of cloud resources and the dynamic joining management of new resources. According to the characteristics of cloud computing, this paper introduces the idea of ontology on cloud monitor system (CMS) based on bionic autonomic nervous system (BANS) and uses ontology web language (OWL) language to describe the resources of the system. It also establishes a reusable extended resource expression model. At the same time, the use of the third-party tool Jena for OWL semantic query also gives the monitoring system the characteristics of a rapid semantic query, which further enhances the convenience of cloud resource management. In addition, based on the application of ontology, we also introduce multivariate decision diagram (MDD) multi-valued decision graph technology, which allows B-CMS to self-diagnose complex system faults. The combination of ontology and MDD greatly simplifies the monitoring and management of large-scale systems, providing a fast and standardized means for the intelligent diagnosis of systems.

In regard to e-commerce, this paper adopts the current mainstream framework technology and distribution architecture and proposes a high-performance online mall design. Firstly, the service architecture of the online mall system is introduced in detail. Then, the main functional modules of the system are described, and the business process of the system is analyzed. Finally, the key technologies used in the system are presented. The relevant design schemes proposed in this paper have important reference value for small and medium-sized enterprises to build low-rise stations, and the solution can also guide developers.

Image quality evaluation is significant in many fields of image applications and research, such as the performance evaluation of image processing algorithms, image compression, encoding, and transmission. This study aims to design an evaluation method by using the phase and structure information of images. Compared with the amplitude structure of the image, the visual system is more sensitive to phase information. Based on this premise, a new image quality assessment method, which is named the phase similarity index (PSI), is proposed. It considers three important features for visual perception: brightness, contour, and phase information. To test the performance of PSI, relevant experiments are conducted on TID2008, TID2013, and CSIQ databases. The experimental results show that PSI can achieve higher consistency with subjective evaluation compared with the most advanced image quality metrics.

Minimal failure-causing schema (MFS) may be affected by masking effects in software complex configuration space. A method for mining MFS based on combination testing and its testing results is proposed. The method firstly constructs a combination-fault tree (CFF-tree) based on the combined test suite and its results, extracts the frequent parameter-value-combinations from the tree as suspicious MFS and calculates their suspiciousness scores, and finally sorts them according to their suspicious scores. Though an iterative framework, MFS is repeatedly mined and checked by programmers until a certain stopping criterion is satisfied. Simulation experiments are used to validate the effectiveness of our method with and without masking effects. The experimental results show that the proposed method can mine MFS in the two scenarios and effectively reduce the number of additional test cases.

This paper aims to study the eye fixation and eye movements of air traffic control (ATC) controllers while working. Based on the establishment of the experimental platform of control simulation, every behavior feature of eye movement was recorded under different difficulty levels of experiments. With the geographic coordinates of the controlled area being converted to their corresponding screen coordinates, the controlled area could be divided into six zones by means of clustering analysis. The percentage of fixation points per unit in Zones 1 and 2 is dramatically greater than that of other zones, as determined by computing the fixation point per unit within the six zones. The Markov theory was applied to compute the probability of one-step fixation movement within the six zones. The results demonstrate that the highest probability of one-step fixation movement exists in Zones 1, 2, and 5, where the airspace of departure and arrival, airway crossover points, and navigation stations exist. Therefore, an experienced controller attaches a greater amount of importance and attention to the congested airspace and areas where conflict could easily occur. This paper can be used as a reference for the training of newly-employed ATC controllers to work in a safer and more efficient manner.

With strong nonlinear characterization ability, a BP neural network can effectively describe the characteristics of nonlinear time series. However, there are still some limitations, such as the ease of falling into a local optimum. Aiming at this problem, the Bayesian regularization optimization algorithm was used to improve the BP neural network. Under the premise of minimizing the objective function, the algorithm adjusts the weight update function through the conditional probability density and the prior probability of the historical data. Thus, the generalization capability of BP neural network will be enhanced. After an empirical study on stock time series prediction, we found that the improved network could prominently increase the prediction ability, while the ability of volatility prediction was better than that of other traditional algorithms.

In order to maintain the normal operation of production equipment, achieve the maximum utilization of personnel and the purpose of reducing personnel without reducing efficiency. It is necessary to research the control method of multi-split wireless remote monitoring system. At present, the method of controlling the multi-split wireless remote monitoring system based on fuzzy PID controller is commonly adopted. Firstly, the data of the multi-split wireless remote monitoring system was collected. Secondly, the wavelet transform was used to smooth the data and remove the interference data. Finally, fuzzy PID controller was used to control the multi-split wireless remote monitoring system online. Due to the fuzzy PID controller, although the current method can achieve comprehensive control, large control error and serious network congestion are the main problems. In order to accurately analyze the current network situation and reduce the network delay rate, a method to control the multi-split wireless remote monitoring system based on neuron PID controller was put forward. Firstly, the relevant data of multi-split wireless remote monitoring system were collected. Then, the network prediction function was used to analyze current network situation of the remote monitoring system, and the relevant network data parameters were calculated. According to the calculation results, the neural PID controller was used to achieve the real-time control for the multi-split wireless remote monitoring system. Simulation results show that the proposed method can predict and analyze the current network conditions accurately and avoid network congestion. Thus, real-time control for the remote monitoring system is achieved.

In order to improve the big data analysis and traffic data management ability of the Jiangxi high-speed traffic condition monitoring platform, a multi-dimensional multi-scale analysis model of Jiangxi high-speed traffic condition characteristics based on vehicle networking and traffic big data information fusion is proposed. On the basis of building a good knowledge base, model base, and method base, the model base feature analysis and dynamic detection method are used to effectively mine big data of the Jiangxi high-speed traffic status monitoring platform. Combined with information fusion theory, the multi-dimensional model of the Jiangxi high-speed traffic status monitoring platform based on text information, location information, picture, audio, video, and other Jiangxi high-speed traffic status data can be realized. The test results show that big data mining in the platform with this method has better clustering and achieves high multi-dimensional multi-scale fusion of high-speed traffic condition characteristics in Jiangxi. The efficiency of data scheduling and access are improved effectively, enhancing the multi-dimensional multi-scale analysis and resource scheduling ability of Jiangxi high-speed traffic condition characteristics.

Image acquisition in heterogeneous environments tends to lead to inadequate information and poor image quality. In order to improve the image quality of dynamic data in heterogeneous environments, an enhanced technology of dynamic data information based on Radon scale transformation is proposed. The gray histogram feature information parameters of images in heterogeneous environments are extracted. The feature quantities are fused and optimized in the central region of image clustering, and the multi-scale Retinex color feature components of dynamic data are extracted. Radon scale transformation is used to extract image centers in heterogeneous environments, enhance dynamic data of image centers, and improve image quality. The simulation results show that this method can enhance the dynamic data information of image centers in heterogeneous environments, and the output images have better imaging performance. The normalized correlation coefficient and peak signal-to-noise ratio (psnr) of the output images are higher than those of the traditional methods, which improves the peak signal-to-noise ratio (psnr) of the output image and improves the recognition performance of the image.

Current fault detection methods based on the immune mechanism for energy routing nodes of power grids have low peak strengths of estimated fault signals, resulting in a low probability of fault detection and fault location accuracy and a lack of fault isolation performance. A fault detection method for energy routing nodes of smart grids oriented to power information security is proposed. According to the fault characteristics of energy routing nodes of power grids, fault diagnosis criteria are given. The necessity and sufficiency of the fault diagnosis criteria are proven. The peak strengths of fault signals are estimated, and fault detection is realized. The fault line is judged by the natural frequency value of the fault traveling wave, and the traveling wave propagation that reflects the fault points is utilized. The intrinsic frequency value of the path can accurately calculate the fault distance and obtain the exact location of the fault. The fault isolation is accomplished by using the distributed power supply and combining it with the current power grid structure with the switch position. Experiments show that this method is superior to the current method in peak estimation strength, fault detection rate, and fault location, and the highest fault detection rate can reach 99%. The implementation of the proposed method can effectively improve the current situation of power grid fault detection and provide a more scientific basis for the development of this field.

Under asynchronous transmission, the hierarchical distribution of unbalanced big data is large and the recognition ability is not good. In order to improve the hierarchical mining ability of unbalanced big data under asynchronous transmission, it is necessary to carry out unbalanced big data elimination ability. An unbalanced big data hierarchical elimination algorithm based on fuzzy association rules feature extraction is proposed. The storage structure model of unbalanced big data's ternary table under asynchronous transmission is constructed. The cooperative filtering method is used to purify the unbalanced big data and filter the interference components in unbalanced big data under asynchronous transmission, the spatial grid clustering method is used to mine the unbalanced big data classification under asynchronous transmission, and the unbalanced autocorrelation features are extracted. The optimal mining and information fusion of unbalanced big data under asynchronous transmission are realized by the hierarchical distribution detection method, and the hierarchical distribution elimination and data storage optimization of unbalanced big data under asynchronous transmission are realized in the reconstructed phase space. The simulation results show that the correlation matching of unbalanced big data elimination under asynchronous transmission is good, the detection and recognition ability of data are improved, and the data storage structure is optimized. It has good application value in data storage, transmission, and output conversion control.

In order to improve the dynamic monitoring and feature recognition ability of coconut red ring disease, the image visual feature recognition method is used to detect the disease, and the method of coconut red ring disease feature recognition based on the Apriori algorithm is proposed. A two-dimensional dynamic hyperspectral image acquisition model of coconut red ring disease is constructed. The dynamic hyperspectral images of the disease are detected by block fusion, while the features are detected according to its texture distribution. The visual fractal features are extracted, the surface texture registration and block regional feature matching method are used to calibrate the feature points, and the feature decomposition of dynamic hyperspectral images are carried out by multi-scale wavelet decomposition. The simulation results show that the accuracy of the method for the identification and dynamic monitoring of coconut red ring disease is high, and the false detection rate is low, which improves the ability of prevention and recognition of coconut red ring disease.

In order to improve the prediction ability of network security level, a design method of network security level prediction system based on time window parameter identification and spatial interval sampling is proposed. According to the prediction feature of network security level, the browsing information and intrusion information feature of network users are detected, the distribution set of network intrusion and attack behavior characteristics reflecting network security level is constructed, the statistical feature quantity of network security level prediction is extracted, and the distributed feature extraction and adaptive detection of network intrusion are carried out according to the combined feature distribution of the network security level prediction network behavior feature set. The feature set and similarity attribute of network spatial distribution information of the load network security level are mined and combined with the estimation results of time parameters, the security level evaluation and optimization prediction are carried out, and the network intrusion optimization detection and security early warning are realized. The simulation results show that the method has strong response ability to predict the network security level, and the accurate detection performance of the network intrusion is better, which improves the prediction ability of the network security level and ensures the network security.

In order to improve the quality of network communication, it is necessary to transfer anti-blocking dynamic regulation of network communication data. A block chain wireless network data transmission anti-blocking dynamic regulation method based on block chain load equalization regulation technology is proposed. The block chain wireless communication network data transmission anti-blocking dynamic regulation transmission channel model is constructed, the anti-jamming design of block chain wireless communication is carried out by an adaptive multipath interference suppression algorithm, and the data transmission anti-blocking dynamic regulation state characteristic quantity of block chain wireless communication network data is extracted. The data allocation model of block chain wireless communication network is constructed, and the data allocation iteration model of block chain wireless communication network is obtained according to the different characteristics of data transmission impedance. The anti-blocking dynamic adjustment optimization of block chain wireless communication networks is realized, and the anti-jamming and security of data transmission anti-blocking dynamic regulation are improved. The simulation results show that the channel balance and output signal-to-noise ratio (SNR) of block-chain wireless communication networks are better, the anti-jamming ability is enhanced, the anti-blocking dynamic adjustment ability of communication data transmission is improved, the output bit error rate is reduced, and the security and load ability of data communication are increased.

In order to allow information acquisition robots to run steadily on land and in the air, it is necessary to study the angular acceleration control method for ground-air dual-purpose information acquisition robots. When the angular acceleration control method for the current robot is used to control the angular acceleration of ground-air dual-purpose information acquisition robots, there are problems of low control accuracy and low stability. Thus, a method of angular acceleration control for ground-air dual-purpose information acquisition robots is proposed. The dynamic equation of the information acquisition robot motion process is constructed according to the DC motor equation of the right and left track driving wheels and the driving motor, as well as the potential balance equation. According to the dynamic equation, the initial values of the information acquisition robot system and the inverse system are obtained. The state equation model of the information acquisition robot system is obtained by using the inverse method of the α-order integral. On the basis of the state equation model, the angular acceleration regulator control model is obtained to control the angular acceleration of ground-air dual-purpose information acquisition robots. The simulation results show that the control results of the proposed method have high accuracy and stability.