Advanced Hybrid Information Processing

In order to overcome the problems of accuracy and low efficiency of the communication channel selection algorithm, a communication channel selection algorithm considering the balance is proposed. The communication channel selection algorithm considering the balance first needs to collect the data of the communication network, and extract the channel impulse response and power delay spectrum from the original data, and then select the noise floor of the channel impulse response and power delay spectrum And multipath search and analysis of channel fading characteristics, and finally through the communication channel selection algorithm for optimal channel selection to realize the communication channel selection algorithm considering the balance. The comparison of experiments verifies that the channel selection accuracy and efficiency of the communication channel selection algorithm considering the balance is always higher than the ALOHA algorithm.

In view of the imperfection of intelligent construction cost specification, the complexity of cost influencing factors and the lack of historical cost data, the expert system and support vector machine theory are combined to achieve knowledge acquisition and data integration. By using the expert system module, the regression calculation, the establishment of project cost prediction model and the model test of parameter setting and optimization are realized. In addition, the investment prediction speed of the model is faster. Finally, through the empirical data analysis, the accuracy and effectiveness of the model are verified, which provides the economic indicators and reference materials for the design stage of intelligent building projects.

Because the traditional coal-fired heating mode consumes a lot of energy and is harmful to the environment, it produces a clean heating mode using electric energy, which is realized by energy storage heating equipment. The operation of energy storage heating equipment needs to be planned according to the electricity characteristics of clean heating. Therefore, a method based on large data model is proposed to integrate the electricity characteristics of clean heating using Hadoop platform. Then, according to the integrated data, the electricity load characteristics, electricity consumption characteristics, electricity consumption cycle characteristics and regional characteristics are identified to complete the electricity characteristics of clean heating. Farewell. Through experimental demonstration, it is proved that this method can effectively identify the electrical characteristics of clean heating and accurately predict the future heating data.

With the continuous development of science and technology, in the context of current big data, the research on the law of traditional virus propagation dynamics had been developed to the bottleneck. The traditional law of virus propagation dynamics was less sensitive and the mathematical model was not easy to operate. Therefore, it was proposed to study the dynamics of viral propagation based on the combination of big data and kinetic models. The model was established by using differential equations and so on, and the accurate prediction law of virus propagation dynamics was completed by experimental tracking control. A graph of the number of patients over time was obtained by bringing the problem into the model, and the changes in the model results were derived from this graph. In this way, corresponding countermeasures was drawn based on the changes in the results. Finally, through simulation experiments, it was proved that the combination of big data and kinetic model of viral propagation kinetics scientifically and accurately studied the laws of viral propagation dynamics. The established mathematical model was easy to operate and had a good guiding significance for practice.

With the increasingly complex network environment and the interference of various other radio waves, the quality of mobile communication network is seriously affected. Aiming at the above problems, this paper studies an auto-disturbance rejection method for mobile communication network nodes based on artificial intelligence. According to artificial intelligence, an interference identification analysis model is constructed, which is used to identify and analyze the interference factors of mobile communication network nodes. Based on the recognition results, the characteristics of different interference types are summarized, and the interference problem is accurately judged. Then, the anti-interference work of mobile communication network nodes is completed by checking and processing the results. The experimental results show that the user is more satisfied with the quality of the mobile communication processed by this method than the traditional method of UAI participating in the identification and analysis of interference factors, which proves that this method is effective in anti-jamming and can meet the needs of users.

The existing methods of dynamic reconfiguration of network information flow have some drawbacks, such as security, reliability and bad influence on the performance of the original network. Therefore, an anonymous reconfiguration method of multi-serial communication information flow under large data is proposed. Firstly, the original information flow is acquired in the communication network, and the cooperative filtering of multi-serial communication is carried out. After filtering, the notification information of relay nodes is obtained in the information flow, and the communication status of the information flow is extracted. The characteristic information of the information flow is reconstructed and anonymized. Finally, the anonymous reconstruction of multi-serial communication information flow is completed. By analyzing and comparing the experimental results, it can be seen that the method proposed in this paper is superior to the traditional method in terms of both the effect of anonymity and the efficiency of operation when reconstructing the anonymous information flow of multi-serial communication, it effectively solves the shortcomings of traditional methods, such as poor anonymous effect of information flow and slow speed of information flow reconstruction. It shows that the method has a high degree of anonymity and has a strong practicability.

In mobile communication systems, the purpose of channel allocation is to maximize the use of spectrum resources. The existing channel allocation is at the cost of frequent channel reallocation, so its practical application is not strong. Aiming at the problems of inaccurate allocation results and high bit error rate of traditional channel allocation methods, a channel allocation method based on big data technology is proposed and designed. This method makes use of the advantages of big data technology to discretize the channel data of mobile communication network. According to the requirements of the channel discretization standard and allocation algorithm of mobile communication network, it optimizes the channel allocation algorithm and realizes the effective channel allocation of mobile communication network. The validity of big data channel allocation method is confirmed by experimental demonstration and analysis. In the mobile communication network channel allocation, the allocation accuracy is high, and the allocation error is almost zero, which is better than the traditional method, and the allocation time is much lower than the traditional method. It shows that this method can realize the effective allocation of mobile network channel, and the allocation result is very reliable, which can guarantee certain network security.

Aiming at the problem that the reactive voltage sensitivity parameter of large-scale distributed wind farm is low overall, the parameter intelligent optimization design of the reactive voltage sensitivity of large-scale distributed wind farm is carried out. Firstly, design a wind farm equivalent circuit and optimize the parameters of the traditional reactive voltage sensitivity optimization model, and set the objective function to adjust the model weight coefficient. Then the bat algorithm is improved according to the parameter intelligent optimization model, and the reactive volt sensitivity parameter of the scaled distributed wind farm is intelligently optimized according to the improved bat algorithm. Finally, a simulation experiment is carried out to test the performance of intelligent optimization of reactive voltage sensitivity parameters of large-scale distributed wind farms. It is concluded that the reactive power sensitivity parameter of the large-scale distributed wind farm reactive voltage sensitivity parameter optimization is significantly higher than that of the reactive voltage sensitivity parameter optimized by the traditional reactive voltage sensitivity parameter optimization method.

In the traditional distributed hybrid energy storage structure, there are security, island operation and capacity blocking. For this reason, a distributed hybrid energy storage structure voltage reactive power control algorithm is proposed based on big data analysis. First, establish a voltage reactive power control model, using the control or manual operation of the communication system to achieve automatic operation of the transformer tap or capacitor input capacity. Secondly, the Nash theorem is used to calculate the physical quantity obtained by the secondary user under certain transmission power conditions. Based on the game theory, the operating data of the power grid is obtained, and then the voltage reactive power control algorithm is realized. Finally, experiments show that the distributed reactive energy storage structure voltage reactive power control algorithm based on big data analysis has certain advantages compared with the traditional voltage reactive power control algorithm.

Materials have always been a hot issue in people’s eyes. With the increasing demand for materials, the performance of various carbon nanotube composites is insufficient to meet people’s needs. Therefore, the performance of carbon nanotube composites based on fuzzy logic is proposed. Optimization Analysis. Firstly, the performance equivalent parameters are calculated. On this basis, the material ratio and the standard geometry are refined. Finally, the performance of the carbon nanotube composite is optimized by the fuzzy relation matrix. The experimental results show that the optimization method can effectively improve the stability, conductivity and bearing capacity of composite materials, and prove that the optimization method can improve the performance of composite materials.

In view of the low filtering accuracy of traditional bad information in the massive data environment, the security filtering algorithm of network dynamic bad information is innovated and improved in the big data environment. Combining the data set analysis algorithm with the grey statistics theory, this paper evaluates the dynamic information security status of the network structure, extracts the information security features in the evaluation results, compares the data features in the network structure, detects the dynamic time domain range of bad information, and filters and corrects the information in the time domain by nodes and channels, so as to realize the security of the dynamic bad information of the network The experimental results show that the dynamic bad network information security filtering algorithm based on big data analysis is more accurate and effective than the traditional algorithm, with high accuracy and the shortest time, and can be used in the network dynamic bad information security effectively, which meets the research requirements.

In order to introduce the grid power flow model to intelligently monitor the network security situation, a model based on grid power flow is established. In the construction of the network security situation intelligent monitoring system, the hierarchical database is protected and managed, the attacks brought by the network security situation are changed, and the network security situation level protection system is improved and improved. On this basis, the network trend correction factor is introduced, and the network security situation is normalized according to the network security situation value. The network information flow is processed uniformly, and the intelligent monitoring model of network security situation based on power flow is built. Compared with the traditional network security situation intelligent monitoring model, the application of network security situation intelligent monitoring model can effectively solve the uncertainty and fuzziness of information provided by various network security devices.

In the power system network, aiming at the low accuracy of traditional network hazard online monitoring method, an online monitoring method for hazard source of power system network based on mobile internet is proposed. Based on mobile internet, a power system network communication is constructed. The model uses this model to collect dangerous source data. After the hazard data is collected, the WAMS system is used to calculate the relative residuals of the hazard source data, and then the relative residuals are used to identify the hazard source parameters, and the branch with the hazard source parameters is present. The traveling wave positioning network is used to locate the dangerous source. After the hazard source is located, the hazard source is monitored online by the hazard source indicator. Under the condition that the experimental environment is the same, the method is compared with the online hazard source online monitoring method based on feature recognition technology and the online hazard source online monitoring method based on communication message parsing. The monitoring accuracy of these three methods is improved. The results are 41.1%, 68.8%, and 94.5%, respectively. The experimental results show that the monitoring accuracy of this method is higher than the traditional online hazard source online monitoring method, which proves the superiority of the method.

The classification algorithm of rotating machinery fault cannot effectively recognize the false components and true components in fault signal of rotating machinery. Therefore, an intelligent classification algorithm of rotating machinery fault based on optimized support vector machine was put forward. The K-L divergence was used to measure the nonlinear and symmetry of probability distribution of two processes in rotating machinery, and the error of information in the process of rotating machinery was measured to eliminate the false component of fault signal of rotating machinery. Meanwhile, the multi-value classification support vector machine algorithm based on decision directed acyclic graph was used to process the signal that only had a true component. Moreover, the value of each node in support vector machine decision function was calculated. Finally, based on calculation results, the fault categories were excluded. Thus, the intelligent classification of rotating machinery fault was completed. According to experimental results, the proposed algorithm can accurately eliminate false components in the rotating machinery fault signal. Meanwhile, the classification result is accurate.

Due to the coexistence of multiple electromagnetic interference, the operational performance of radar equipment will be seriously affected. Therefore, it is necessary to study the anti-jamming problem of airborne radar. In view of the problem that airborne radar is easily affected by point source signal interference under the traditional method, an airborne radar anti-jamming method based on artificial intelligence is proposed. The anti-jamming method is designed. Firstly, the airborne radar is detected by frequency shift, and the detected information is analyzed to judge the jamming environment and identify the point source target intelligently. Then the suppression jamming filter is generated based on the analysis of the point source jamming information, and then the suppression jamming signal is output. Finally, the anti-jamming method of airborne radar is obtained. The performance results of the airborne radar anti-point source jamming method are analyzed by simulation experiments. Compared with traditional method, the proposed anti-jamming method can effectively suppress the point source jamming information, the radar signal is clearer and the anti-jamming effect is better. The results verify the effectiveness of the proposed method.

Different temperature, power, flow rate and other factors have different effects on the removal of soot particles in the tail gas of simulated diesel vehicles, and the removal effect of each kind of soot particle catalytic removal method is also different. In order to further improve the effect of soot particle catalytic removal, a statistical analysis method of soot particle catalytic removal method based on big data is designed. Using large data technology to extract catalytic removal methods of soot particles, detailed analysis of each method was carried out, and the soot combustion performance of soot particles catalytic removal method was compared. The results showed that the removal of soot particles based on perovskite catalyst was more effective than that of soot particle removal method based on sol-gel preparation method, and that soot particles were catalyzed by low temperature plasma. The combustion performance of the removal method is better, and the catalytic removal performance is more superior.

In order to solve the problems of low control accuracy and poor stability of traditional electric drive control methods, a parallel computing based electric drive control method is proposed. According to the selected motor parameters, the transmission point with the maximum power or the strongest mechanical rigidity is selected as the main node, so that the parameters are equal to the load rate of the load distribution motor, and the load distribution is completed by parallel calculation; the air gap magnetic field is generated inside the motor, and the electromagnetic thrust is generated under the interaction with the excitation magnetic field generated by the permanent magnet by using the concepts of coordinate transformation and space vector Force is used to push the motor to move synchronously and linearly at the same speed to complete the motor vector control and the electrical drive control based on parallel computing. The experimental results show that the control accuracy of the proposed method is high and the operation is stable. It can effectively reduce the position tracking error and improve the control accuracy.

In the cloud computing environment, traditional social network community discovery algorithms have low accuracy in social network community discovery, leading to information waste, community overlap and low scalability, and unable to achieve ideal computing results. Therefore, a social network based on parallel recommendation is proposed. Network community discovery algorithm. By mining the candidate trusted user set, the number and composition of the community are obtained, and the communication units are divided into overlapping communities and non-overlapping communities according to the different numbers of communities belonging to the nodes in the network. Combining the mining of candidate trusted user sets and community division, social networking is realized Network community discovers and calculates. Experiments show that the algorithm improves the accuracy and stability of social network community discovery, and has good application value.

The traditional deployment optimization method of perception layer nodes in the Internet of Things has the drawbacks of poor optimization performance. Therefore, this paper proposes a research on deployment optimization of perception layer nodes in the Internet of Things based on NB-loT technology. The genetic algorithm is used to code the nodes in the perception layer of the Internet of Things, and the initial population is determined. Based on the coding of the nodes in the perception layer and the initial population, the fitness function is designed, and the NB-loT technology is used to optimize the deployment of the nodes in the perception layer of the Internet of Things. Experiments show that the average coverage of the proposed method is 24% higher than that of the traditional method, which shows that the proposed method has better optimization performance.

In order to solve the problem that the transmission link of wireless volume domain network is likely to be interrupted and consume unnecessary energy, this paper introduces probability statistics and proposes a research on energy saving of wireless volume domain network multi-relay nodes based on probability statistics. The energy consumption of network is analyzed and the formula of total energy consumption per bit network is derived. The simulation results show that compared with the traditional multi-path multi-relay node forwarding method, this method can greatly reduce the overall energy consumption of the network. Consumption also plays a role. This method can reduce the overall energy consumption of the network and prolong the life cycle of the network. When the optimal relay node is used for transmission, the transmission power is greatly increased.

Aiming at the problem of unstable equilibrium probability in modern load scheduling applications, a statistical method of unbalanced probability in cloud load scheduling is proposed. The weights and anti-saturation factors are calculated, the servers are grouped, the fuzzy cyclic iterative control of dynamic network resources is realized, and the network packet cloud load scheduling is designed. By comparing with the common methods, it is proved that the method designed in this paper can guarantee high equilibrium probability and good stability in a certain program.

In order to solve the problem that traditional cloud trust-driven mobile nodes in the Internet of Things lack credible authentication, a cloud trust-driven intelligent authentication method for trusted access of mobile nodes in the Internet of Things is proposed. The mobile nodes in the Internet of Things are determined based on cloud trust-driven, relying on the processing of mobile nodes in the Internet of Things and the intelligent authentication of trusted access of mobile nodes in the Internet of Things. The cloud trust-driven Internet of Things migration is realized. Mobile node trusted access intelligent authentication. The experimental data show that the proposed intelligent authentication method can not only improve the credibility of the traditional authentication method, but also simplify and standardize the authentication process. It enhances the adaptability and flexibility of trusted access authentication of Internet of things driven by cloud.

On the traditional method of dynamic integration of multi-objective data in UI color interface, because of the single integration algorithm, it is easy to lose the target data when there is too much target data. Therefore, based on the use characteristics of UI color interface, a new integration method of multi-objective data is proposed. This method obtains the sampling target through deep web data, detects and tracks the target image, optimizes according to the multi-objective integration, realizes the optimal path multi-objective equilibrium integration. Experimental results: the proposed detection method is fully in place in data integration, the occupancy rate of arm is 0%, the load line of DSP is 20%, the system maintains reliable real-time, and achieves the ideal state of UI color interface operation. However, the traditional data integration method of SLR is not in place; it can be seen that the traditional integration method is not suitable for the requirements of UI color interface with large target data.

The existing interactive interface of online teaching mobile terminal is not well used in the process of user experience. To optimize it, this paper puts forward the application of the visualization of the Internet of Things in the optimization of the interactive interface of online teaching mobile terminal. Utilizing the visualization technology advantages of the Internet of Things, we can improve the user’s sense of use by optimizing the mobile interactive interface vision and human interaction design with the use module. Design simulation experiment compares the number of user choices before and after optimization to verify the validity of the design.

In order to extract the key and useful features of the target in the UAV video image and strong marking ability, a feature extraction method for the UAV video image based on target tracking is proposed. The sparse beam method is used to adjust the splicing of UAV video images. Based on this, the pixel coordinates are obtained through the frame difference method to detect and locate the target. According to the target detection and positioning results, the video image of the target area is selected and preprocessed by the wavelet transform algorithm Target area video image, and extract the target area video image feature, through hierarchical particle filtering to achieve target tracking, to achieve the extraction of UAV video image feature. The experimental results show that: in the ORL database experiment, the average feature extraction percentage is 78.08%, and the average target tracking error is 1.16; in the COIL-20 database experiment, the average feature extraction percentage is 82.55%, and the average target tracking error is 1.20, which meets the needs of UAV video image feature extraction and target tracking.

The existing recognition method of tea shoots is only to judge the single color or shape features, resulting in low recognition accuracy. Therefore, an automatic recognition method of tea shoots image based on support vector machine is designed. In this method, two kinds of image features, color and shape texture, are extracted from the tea bud image for discrimination. The RGB model is used to extract color features, and LBP/C operator is used to extract the shape and texture features of the bud. The extracted features are used as the feature vectors of the training samples, and support vector machine model training is carried out to obtain the support vector machine classifier, and the tea bud image is recognized. The experimental results show that the recognition rate, recall rate and comprehensive evaluation index of the method are higher than those of the traditional method, which proves that the method has high recognition accuracy and improves the recognition efficiency.

Aiming at the problems existing in traditional color image segmentation methods, namely, image noise and image quality are poor, a color image automatic segmentation method based on visual characteristics is proposed. The method first analyzes the human visual characteristics, then uses the weighted average method to grayscale the color image, then uses the histogram equalization method to enhance the image, and then detects the edge of the image through the binary wavelet, and finally in the image. Image segmentation based on edge detection. The results show that compared with the traditional image segmentation method, the segmented color image of this method has a SNR of 5.3 dB, less noise and improved image quality.

In order to solve the problem that the average recognition degree of moving target line is low by the traditional method of moving target behavior recognition. Therefore, a motion recognition method based on deep convolutional neural network is proposed. Construct a deep convolutional neural network target model, and use the model to design the basic unit of the network. The returned unit is calculated to the standard density map by the set unit, and the moving target position is determined by the local maximum method to realize the moving target behavior recognition. The experimental results show that The experimental results of the multi-parameter SICNN256 model are slightly better than other model structures. And the average recognition rate and the recognition rate of the moving target behavior recognition method based on deep convolutional neural network are higher than the traditional method, which proves its effectiveness. Since a single target is more frequent than multiple recognitions and there is no target similar recognition, similar target error detection cannot be excluded.

Aiming at the problems of low efficiency and accuracy in the traditional 3D image feature point detection system, an efficient 3D image feature point detection system based on artificial intelligence is designed. Firstly, the whole frame of the system is designed. Then the hardware system is designed, including the development board, peripheral equipment and interface, basic engineering reconstruction and feature point detection unit. Then the software system is designed, including image collection module, image feature point display module. Image feature point processing module, image feature point extraction module, image feature point description module, and using the combination of hardware system and software system to achieve three based on artificial intelligence Dimension image feature point detection system. Finally, the effectiveness of the 3D image feature point detection system based on artificial intelligence is verified by experiments, and the detection efficiency and accuracy are much higher than the traditional methods. This study lays a foundation for the further study of images.

The manipulator has dynamic characteristics, and the trajectory tracking system of the manipulator has non-holonomic constraints and various uncertainties, which makes tracking control of the mobile manipulator more difficult. There is a big error in tracking a rigid flexible coupling manipulator with a single neural network. A new method for trajectory optimization tracking of a rigid-flexible coupled manipulator based on big data analysis is proposed. This method takes neural network as the research object, introduces fuzzy control into neural network, optimizes a single neural network, forms a composite method of fuzzy neural network, and uses a hybrid method to track the trajectory of the manipulator. Experimental results show that the tracking error of this method is less than 0.035 rad, which improves the tracking efficiency and improves the tracking accuracy. The method can complete the operation faster and more accurately according to the predetermined trajectory, and has higher practical applicability.

In order to overcome the low efficiency of traditional recognition technology, a fast recognition method of multi-combination features of moving images based on large data analysis is proposed. Based on feature extraction of multi-combination target, denoising of moving image and determination of Boolean correlation coefficient, fast recognition of multi-combination target feature of moving image under large data analysis is realized. The experimental data show that the proposed recognition method can not only effectively improve the efficiency of traditional recognition technology, but also make the recognition result more stable, and enhance the adaptability and flexibility of image recognition technology.

Aiming at the problem that the adaptive convergence of noise iteration performance is not fast enough in the traditional spatial scene visual information communication method, a method for accurately transmitting spatial scene visual information based on big data analysis is proposed. The digital image acquisition of the spatial scene is realized by signal filtering processing, then the spatial scene image is transcoded, intercepted and preprocessed, and the spatial scene image distortion correction, image smoothing processing and image segmentation are performed. Finally, through the serial communication method, the visual information of the space scene is accurately transmitted. The experimental results show that the noise iteration performance of the spatial scene visual information accurate communication method based on big data analysis can quickly and adaptively converge compared with the traditional spatial scene visual information transmission method.

The traditional detection method has the problems of complicated operation and slow search speed, which brings great impact to the efficient operation of the local search system of community structure. To this end, it studies the rapid detection method of local search target of community structure under big data. Analyze the key technologies for constructing detection methods, use quantitative algorithms to achieve rapid target location, perform resource entry on targets, and calculate data convolution kernel size. The convolution data is statistically analyzed, and the detection result is subjected to parsing and storage, thereby realizing the extraction of the target and completing the rapid detection of the local search target of the community structure. It is proved by experiments that the fast detection method of local search target of community structure has obvious advantages in search time consumption and has a good development prospect.

Based on the comprehensive research of image segmentation technology, an adaptive segmentation algorithm based on pattern recognition for image weak targets is proposed. By systematically designing the image segmentation algorithm by analyzing the algorithm requirements and principles, the modules such as image preprocessing, weak target detection, image feature extraction and adaptive threshold selection are designed and implemented according to the algorithm implementation flow. In order to verify the experimental performance of the algorithm, experimental analysis shows that the adaptive image segmentation algorithm can be used to preserve image details, improve the quality of the segmented image, and shorten the image segmentation time.

Because the traditional single-channel information transmission algorithm ignores the real-time control of the transmission data, the signal transmission tracking accuracy is low. For this reason, a target tracking algorithm for multi-channel information transmission in a big data environment is proposed. The algorithm solves the echo signal of each point, determines the transmission range of the multi-channel information, uses the interrupt mechanism to optimize the decoding algorithm, and obtains the position of the data of each point through the classification of the classifier, so as to realize the transmission target tracking of the multi-channel information. The traditional single-channel information transmission algorithm and the target tracking algorithm of multi-channel information transmission are compared and analyzed. The experimental results show that the information transmission and tracking accuracy of the multi-channel information transmission target tracking algorithm in the big data environment is better than that of the traditional single-channel information transmission algorithm The information transmission tracking accuracy is high, and it has a better information transmission tracking effect.

Under manipulator trajectory motion algorithm, the time consumption of manipulator motion is long, and the stability coefficient is low, and the accuracy of obstacle avoidance is poor. An algorithm that six degree of freedom manipulator moves with end trajectory based on particle swarm optimization was proposed. By analyzing the positive solution of kinematics and inverse solution of kinematics of manipulator, the expected pose of manipulator was obtained to realize the analysis of manipulator trajectory. According to the constraint condition of the short time consumption of motion, the high accuracy of obstacle avoidance and the strong motion stability, the model that six degree of freedom manipulator ARM moved with end trajectory is built. The model was subdivided into particle swarm optimization algorithm to realize the solution of model. Then, some parameters such as initial position and speed of particle swarm were set, and particle swarm fitness function was calculated to get the optimal solution. Finally, we determined whether the current optimal solution was the global optimal solution. Thus, we obtained the optimal planning results that the manipulator moves with the end trajectory. Experiment shows that the time consumption of manipulator motion is short. The average accuracy of obstacle avoidance is 95%. The stability coefficient is high. This algorithm can effectively solve the problem of current algorithm, which has practicality.

In the case of multiple UAVs, the navigation area of UAV is planned to effectively improve the accuracy of track control and ensure the navigation safety. Because there are some problems such as track deviation and delay of obstacle avoidance when using traditional methods to control the multi-UAV track, it is difficult to meet the requirements of track control accuracy and safety, an automatic control method of multi-UAV track based on embedded system is proposed. The mathematic model of UAV track control is designed based on the fuzzy algorithm, in order to obtain the route deviation parameters accurately, and the area range of UAV track is standardized according to the calculation results, and the control steps of UAV track are planned within the track range, so as to achieve the automatic control target of multi-UAV track. The experimental results show that the embedded multi-UAV track automatic control method can effectively solve the problem of large track deviation, and can avoid navigation obstacles and achieve the research goal of effective control of multi-UAV track. The experimental results show that the UAV under the control of this method can avoid obstacles accurately, solve practical problems effectively, it can effectively solve the problems of the traditional methods in track control and obstacle avoidance, it shows that the proposed method has practical application value.

In the visual non-destructive rendering of three-dimensional animation images, the traditional visual non-destructive rendering method is slow, so a visual non-destructive rendering method of three-dimensional animation images based on large data is proposed. The theoretical model of pixel-by-pixel time-domain denoising process is used to denoise, and GPU is used to achieve time-domain consistent processing according to the denoising results. The non-linear Kuwahara filter is used to smooth the three-dimensional animation image, and the first-order differential operator is used to highlight the dramatically changing pixels in the image, so as to detect the edge of the image. After obtaining the distinct contour of the three-dimensional animation image, the non-destructive rendering of the three-dimensional animation image vision is realized. In order to verify the effectiveness of this method, the average rendering speed of the proposed method is 83.2%, which is significantly higher than that of the traditional method. The experimental results show that the average rendering speed of this method is the highest, the image rendering effect of this method is better, and the effectiveness of this method is verified.

The human perception that more than 80% of the external information is visually acquired, therefore, in the interactive animation interface design, the visual effect is very important. In this background, an interactive animation interface visual reconstruction method based on Web technology is proposed. The method is mainly described by two aspects, firstly, the related description is carried out on the Web technology, and then the visual reconstruction of the interactive animation interface is realized by using the technology, and the method comprises the visual feature extraction, the visual feature matching and the visual feature 3D reconstruction. The results show that, after the visual reconstruction, the visual effect of the interactive animation interface is improved, and the visual existence in the design of the interactive animation interface is solved.

The traditional micro image surface defect detection system had slower running speed and less detection precision, which made the detection system operate inefficient and could not meet the requirements of small image surface defect detection. To this end, the optimization design of the micro image surface defect detection system based on machine vision-based big data analysis was carried out. The system design was optimized with MATLAB 7.0 programming environment; MATLAB technology was used to process small images to visualize calculation results and programming; The filtering of the micro image was detected by the method of spatial domain filtering to complete the detection task of the surface defect of the micro image. The design method was validated and the test data showed that the micro image surface defect detection system ran faster and the detection was more precise. The detection accuracy was 92% and the detection quality was high.

Aiming at the problem that the accuracy of the frame strength detection method is not high, the strength detection method of the subway vehicle bogie frame is studied in the big data environment. Firstly, the new structure of the subway vehicle steering frame is taken as the research object. The CATIA software is used to carry out the solid modeling of the subway steering frame to construct the frame 3D model to obtain the frame strength detection parameters. Then, the frame test rig is built, and the strength test of the frame test rig is carried out by the finite element model of the frame strength detection to realize the strength detection of the subway vehicle bogie frame. The simulation experiment is carried out to verify the detection accuracy of the strength detection method of the metro vehicle bogie frame. The experimental comparison shows that the strength detection method of the metro vehicle bogie frame is higher than the traditional frame strength detection method.

In the process of monitoring the abnormal load of big data in network behavior, more network traffic resources are consumed, which leads to the low efficiency of its operation. Therefore, an on-line monitoring method for the abnormal load of big data in network behavior based on deep learning is proposed. The online monitoring model of load anomaly is established, the network data distribution is analyzed, and the adaptive random link configuration is adopted to improve the channel balance and the positioning ability of the abnormal load. The load anomaly is identified through the load pattern and the online monitoring is completed. The experimental results show that the proposed method consumes about 50% of the traffic of the traditional method, which can effectively reduce the traffic consumption and improve the utilization rate of network resources. This method is more suitable for online monitoring of big data load anomalies in network behavior.

In order to solve the problem of discrepancy between simulation results and measured results of building energy consumption simulation and analysis, a method based on big data and BIM technology is designed. The 3D building information model is constructed by BIM technology, and the factors affecting building energy consumption are obtained. The building energy consumption is simulated and predicted by heat balance method and Design Builder. Finally, data mining technology is used to modify the prediction results, and static energy analysis method is used to analyze the revised results. So far, the design of building energy consumption simulation and analysis method based on big data and BIM technology is completed. Compared with the original method, the simulation results of this method are close to the measured ones. In summary, the energy consumption simulation ability of this method is better than the original method.