In order to overcome the shortcomings of the traditional experiment box of the communication principle in signal processing and detection, it is necessary to use simulation to show the experimental phenomena in a more comprehensive way. However, Simulink, LabVIEW, and other software can only simulate the digital communication system. To solve this problem, Multisim is used to perform component-level circuit simulation of analog communication systems, including PAM ( Pulse Amplitude Modulation) encoding and decoding, FSK (Frequency Shift Keying ) modulation and demodulation, and PSK ( Phase Shift Keying ) modulation and demodulation. Through simulation, the inherent working mechanisms of sampling theorem, phase-locked loop, and Costas loop are intuitively demonstrated, and the principles of two possible distortions that can occur during FSK demodulation are thoroughly analyzed. 

In order to reduce the interference of common mode noise and improve the SNR (Signal-to-Noise Ratio) of laser self-mixing grating interference signal under the condition of signal loss, a method of measuring the interference displacement of laser self-mixing grating based on differential technology is proposed. The +1 diffraction light is reflected by the vibrating object and returns to the laser cavity, where self-mixing interference occurs and is received by the photodetector PD1 encapsulated in the laser, while the 0 diffraction light is received by the external photodetector PD2 . According to the principle that there is phase difference in the forward and backward output of semiconductor laser, the phase of the self-mixing interference signal carried by the two beams is opposite, and the superimposed noise is partially correlated, so the difference between the two interference signals can eliminate part of the noise. By processing the differential self-mixing signal, the displacement can be reconstructed. The experimental results show that the mean error of the reconstructed vibration waveform is smaller, the signal is more stable, and the signal-to-noise ratio of the interference signal is improved by no less than 3 dB.

In order to improve the QoS(Quality of Service) of users in mobile communications and to solve the problem that the total energy efficiency of the system and the quality of service of users can not be reconciled, a resource allocation scheme based on the quality of service and energy efficiency of users is proposed for the application scenario of SCMA(Sparse Code Multiple Access) downlink. The capacity and energy consumption of the system are taken as the main directions for optimisation. A greedy algorithm is used to allocate subscribers to the codebook, which is transformed into a convex optimization problem using the GDA(Generalized Dinkelbach’s Algorithm) for iterative solution. The user code book and energy consumption are calculated in rotation until the minimum user communication speed in the system changes below a preset value or reaches the maximum number of operations, so that the capacity and energy consumption of the system reach a better situation. The theoretical analysis and simulation results show that the proposed SCMA allocation scheme can effectively improve the communication rate and system energy efficiency of the users in the downlink.

Aiming at the defect that the original LPP ( Local Preserving Projection) algorithm is difficult to accurately obtain the local manifold structure of non-uniform high-dimensional data and can not use the sample category information, a MIF-LPP (Multi-Information Fusion Local Preserving Projection) algorithm is proposed. MIF-LPP algorithm uses the improved standard Euclidean distance to obtain the nearest neighbor information and mutual neighbor information of samples, reducing the impact of uneven distribution of sample points and the difference of data dimensions of a single sample. The weight matrix is constructed by fusing the class information of the samples, and then the low dimensional essential manifold of the data is obtained. The validity of the algorithm is verified on CWRU(Case Western Reserve University) data set and our laboratory bearing data set respectively. The experimental results show that the feature extraction performance of MIF-LPP algorithm is obviously superior to other algorithms, and it is robust to neighborhood values. 

Chinese chemical named entities do not have strict word formation rules to follow, and the recognition entities contain letters, numbers, special symbols and other forms, and the traditional word vector model can not effectively distinguish between nested entities and ambiguous entities in chemical terms. The named entities of high school chemistry test resources are devided into four categories: substances, properties, quantities, and experiments, constructing a vocabulary of chemistry subjects to assist manual labeling. Then, the ALBERT pre- training model is used to extract text features and generate dynamic word vectors, and the named entity recognition is performed on the text of high school chemistry questions combined with the BILSTM-CRF (Bidirectional Long Short-Term Memory with Conditional Random Field) model. The accuracy, recall and F1 values of the proposed model reached 95. 24% ,95. 26% and 95. 25% , respectively. 

The cloud data of metal part surface damage point requires high local feature analysis ability of the segmentation network, and the traditional algorithm with weak local feature analysis ability can not achieve the ideal segmentation effect for the data set. The relative damage volume and other features are selected to classify the metal surface damage, and the damage is divided into six categories. This paper proposes a method to extract the attention feature of 3D map containing spatial scale area information. The obtained spatial scale area feature is used in the design of feature update network module. Based on the feature update module, a feature updated dynamic graph convolution network is constructed for point cloud semantic segmentation. The experimental results show that the proposed method is helpful for more effective point cloud segmentation to extract the local features of point cloud. In metal surface damage segmentation, the accuracy of this method is better than pointnet++, DGCNN(Dynamic Graph Convolutional Neural Networks) and other methods, which improves the accuracy and effectiveness of segmentation results. 

Measure planning is the main method to increase oil and control water in oilfield. In order to accurately predict the effect of various measures to increase oil production, a measure effect prediction model based on HDCNN(Hybrid Dilated Convolutional Neural Network)-BIGRU-Attention is proposed with monthly oil production and water content as the prediction targets. The model extracts multi-scale global features of production data through HDCNN. Aiming at the characteristics of strong timing and large volatility of measure production data, the BIGRU(Bidirectional Gated Recurrent Unit) is used to fully mine the long-term dependence between data to improve the utilization rate of time series information and the learning effect. The scaled dot- product attention mechanism (Attention) is introduced, and the weight adjustment strategy is used to make the network focus on the feature dimension with large correlation with the prediction target. In order to verify the effectiveness of the proposed model, LSTM(Long Short-Term Memory), CNN(Convolutional Neural Network)- LSTM and LSTM-attention are taken as experimental comparisons. The results show that the proposed model has lower prediction error and better generalization ability. 

 In order to solve the large difference between the results of multi-attribute group decision making, the convergence of measurable functions is studied based on intuitionistic fuzzy measures. Based on the possibility theory and intuitionistic fuzzy set theory, the intuitionistic fuzzy possibility measure is studied, and the boundedness, monotonicity, additivity and intuitionistic fuzzy additive properties of the intuitionistic fuzzy measure are determined. The intuitionistic fuzzy measure is based on the three decision-making rules of receiving decision, delaying decision and rejecting decision. Based on the above research results, the convergence of measurable functions in terms of measure is clarified. It is known that there exists a convergence state of set- valued bizero asymptotically additive set function in measurable space. The example shows that this method can effectively calculate the intuitionistic fuzzy possibility measure of different decisions under different attribute characteristics, and can effectively obtain the best decision scheme, can effectively solve the multi-attribute group decision-making problem with good practicality. 

 For the fault diagnosis problem of the indicator diagram of pumping wells, the feature information in the image is extracted by convolution neural network. In order to ensure the diagnostic performance of the network model, the structure complexity of the network model is reduced. Based on the lightweight convolution neural network, the attention mechanism is introduced to improve the diagnostic performance of the lightweight network model. First, in the network infrastructure the MobileNet-V2 network is used, and the attention ECA (Efficient Channel Attention Module) module is embedded in the inverse residual module of MobileNet-V2. Compared with the ordinary residual network, the features retained after convolution are more complete, so the fault diagnosis capability of the model is improved. Then, the ECA uses 1D convolution to achieve local cross- channel information interaction between adjacent channels and obtain the dependencies between local channels. The resulting channel attention re-calibration weights are multiplied by the corresponding channels of the input feature map of the module, and the attention-weighted feature map is obtained. The MobileNet-V2 accuracy rate is 90. 6% , and the improved MobileNet-V2 diagnostic accuracy rate is 97. 60% . 

 Microseismic inversion is an important way to complete main task of microseismic monitoring by inverting and inferring information such as the location of the epicenter, the time of occurrence of the earthquake, the true magnitude, the initial amplitude of the hypocenter, the focal mechanism and the medium parameters based on the microseismic monitoring data. Through the study of microseismic inversion technology, more accurate microseismic information can be obtained, thereby improving the reliability of reservoir fracture evaluation, reducing development costs and improving oil and gas recovery. We review microseismic inversion from three aspects: microseismic focal location inversion, microseismic focal mechanism inversion, and microseismic multi-parameter joint inversion. The research progress of microseismic inversion technology in recent years is reviewed, and the principle, advantages and disadvantages of various microseismic inversion methods are summarized, the improvement and application of various microseismic inversion techniques are summarized, and the future research ideas and development directions are prospected. It provide reference for further development of microseismic inversion in the future.

With the increasing complexity of computer network space, in order to effectively reduce the losses caused by network security events, a multi classification prediction model based on the quantum particle swarm lightweight gradient descent algorithm (QPSO LightGBM: Quantum Particle Swarm Optimization-Light Gradient Boosting Machine) is proposed to evaluate vulnerabilities of high-risk network assets. Synthetic MOTE(Minority Oversampling) technique is used to balance the data, QPSO(Quantum Particle Swarm Optimization) is used for automatic parameter optimization is realized, and LightGBM is used for modeling. Multi-classification prediction of network asset vulnerability is realized. In order to verify the rationality of the model, the proposed model is compared with the model constructed by other algorithms. The results show that the proposed model is better in various performance indexes.

In order to solve the problem of computer aided diagnosis of novel coronavirus pneumonia (Covid-19: Corona virus disease 2019), a bilinear convolutional neural network model is created and a feature extraction subnetwork with VGG(Visual Geometry Group network) 16 and VGG19 is employed. The algorithm is applied to COVID-19 image classification and compared with the basic image classification algorithm. The results and lesion visualization analyses demonstrate that the bilinear convolutional neural network model outperforms other deep learning network models in terms of accuracy, with an accuracy of 95. 19% . By replacing softmaxlayer and using SVM(Support Vector Machines) classifier, the model classification accuracy is improved to 96. 78% . The study provides a trustworthy tool for the quick and accurate diagnosis and treatment of neonatal pneumonia and a confirmation of the viability of fine-grained imaging algorithms for the categorization of COVID-19 CT images. 

In order to solve the impact of excessive charging load on the power grid, a two-layer optimization model of distribution network considering energy storage operators and electric vehicles is proposed. The upper layer of the model considers the economic benefits of energy storage operators participating in distribution network peak shaving and valley filling and electric vehicle charging, and the lower layer considers electric vehicle owners actively responds to the economic benefits of energy storage operators dispatching charging and discharging based on the time-of-use electricity price. The improved simulated annealing algorithm is used to solve the two-layer model, and the IEEE 33 node is used to verify that the model has considerable benefits in ensuring the safe operation of the distribution network. 

 With the advent of big data era, the development of recommendation systems has become more and more vigorous. It has become a research hotspot to push information that may be of interest to users in a timely manner among massive amounts of information. Traditional recommendation algorithm lack implicit information and contextual information about graph structures. In response to this, a recommendation model is proposed based on graph neural network. The main innovations are: 1) Based on the higher-order connectivity theory of graphs, the graph neural network is used to mine the hidden information in the user-item bipartite graph, and a the order is extended to multiple orders, so as to obtain more accurate embedded representation and recommendation effect; 2 ) Consider context information in the update process, which is conducive to understanding the interaction between contexts. The model is tested on the Yelp-OS, Yelp-NC and Amazon-book datasets, and the results show that it is better than the related comparison algorithms in both HR(Hit Ratio)and NDCG(Normalized Discounted Cumulative Gain) indicators, which proves that the algorithm can optimize the recommendation effect and improve the recommendation quality. 

The propagation of digital media images is widely used in daily life. At present, the fuzzy image de duplication method still has the problems of unclear image and low quality after processing. In order to solve the problems, an adaptive fuzzy de duplication algorithm for digital media images based on wavelet domain is proposed. Firstly, the digital media image is denoised by wavelet domain method. Secondly, the digital media image is divided into protected area and unprotected area by using the method of gradually labeling significant area, in which the protected area is the significant area. Finally, the image is processed by significance regularization, and the image adaptive fuzzy de duplication algorithm is completed. The experimental results show that the image noise is low, the image quality is high, the image information is rich, and the definition is good. 

 In order to cope with large-scale new energy access, a two-stage optimization scheduling model of an active distribution network is constructed. Based on the IEEE 33-node distribution network model, a day-to-day scheduling model of an active distribution network is constructed by adding dispatchable loads such as electric vehicles and energy storage devices, and reactive power regulating devices such as capacitor banks, static reactive power compensation, and on-load voltage regulating transformers. In the pre-day stage, the output of each equipment in the distribution network in the next 24 hours is scheduled with the optimization goal of comprehensive operation cost. In the intra-day rolling optimization scheduling stage, day-ahead scheduling results of energy storage equipment, electric vehicles, and capacitor banks are taken as constraint conditions, and voltage stability, network loss, and new energy utilization are taken as optimization objectives. The rolling optimization cycle is 4 hours. After dispatching load access, the new energy consumption rate of day-ahead dispatching increased by 4. 41% , and the power purchase cost decreased from 472. 03 yuan to 446. 90 yuan. Compared with the day-ahead scheduling, the absorption rate of new energy in day-ahead scheduling is reduced by 0. 53% , and the maximum voltage offset of each node is reduced from 0. 060 0 to 0. 025 3. Experimental results show that the proposed two-stage scheduling model has a lower voltage offset and higher new energy utilization. 

 Aiming at the difficulty of detecting subtle defects on O-ring surface, we present a method of detecting defects on O-ring surface based on six photometric stereoscopic method and image comprehensive feature analysis. First, the images of six different light source angles are collected, and the surface gradient map and reflectance map are reconstructed by photometric stereoscopic method. The surface gradient image is first converted into the average curvature and Gaussian curvature image, and then converted into the gray-scale image. The defect region is segmented using a fixed threshold. After the reflectivity map is filtered by Gauss, the local mean and variance thresholds are used to segment the defect area. Finally, the defects are accurately selected by analyzing the connected domain characteristics of the obtained defect regions. The experimental test results show that it has a good effect on the subtle defects such as weld marks, concave-convex and flow marks on the surface of the seal ring. In the application of the designed seal ring quality detection system, the detection accuracy is more than 98. 4% , which can solve the problem of low recognition rate of the current industrial sealing ring defect detection.

 Due to differences in camera parameters, shooting environment, and angles for pedestrian photography, the accuracy of pedestrian recognition algorithms still needs to be improved. To this end, a pedestrian re recognition algorithm based on pedestrian semantic perception information and deep learning is proposed. Firstly, super-resolution reconstruction of pedestrian views enhances the detailed features of pedestrian views, extracts the overall feature values of pedestrians, and uses them to identify pedestrians with significant body differences. Secondly, the Semantic information of pedestrian images is perceived, and the feature values of pedestrian Semantic information are extracted according to the above results to identify pedestrians with the same or similar body shape. Then, the macroscopic feature values of the human body and the semantic perception information feature values in the pedestrian video are fused into a comprehensive feature value. Use the generated feature values to calculate the distance between them and the video feature values of different individuals, and identify massive character images. Finally, this article validated the performance of the algorithm in different datasets. The experimental results show that the language perception based pedestrian recognition algorithm has the highest mAP and rand-1 values.

In order to solve the problem of unclear face information in face recognition under face occlusion, and optimize the face recognition system, a local occlusion face image recognition method based on face edge details is proposed. The face image denoised according to the sparse expression, the edge of the face image is detected according to the principle of image gray transformation, the edge region is segment, and its threshold is calculated to obtain the edge information of the face image. The face feature is marked points to enhance the accuracy of information recognition, the feature descriptor of face image is extracted, it is input into support vector machine model, and local occlusion face image recognition is realized through training. The experimental results show that the average recognition rate of face image under the application of the proposed method is higher than 73% and the recognition time is less than 20 s. 

In the case of zero sample remote sensing image scene segmentation without any observation data, there is no response reference, which results in long segmentation time and low accuracy. Therefore, a zero sample urban remote sensing image scene segmentation algorithm based on convolutional neural network is proposed. PCA ( Principal Component Analysis) and K-SVD ( K-Singular Value Decomposition) are used to denoise remote sensing images to suppress the patch effect. The denoised image is input into the Retinex enhancement algorithm to further improve the enhancement effect of zero sample urban remote sensing image. The mean shift algorithm is used to segment the remote sensing image scene to obtain the relationship between its pixels, and the convolution neural network is used to complete the accurate segmentation image scene. The experimental results show that the algorithm has high accuracy, high recall, high F-score rate and short consumption time.

When collecting spatial multidimensional data, collection devices are often discrete, and due to equipment failures, environmental factors, and other factors, there may be omissions or anomalies in spatial multidimensional data, a spatial multidimensional data calibration model based on the Lagrange interpolation algorithm is proposed. Firstly, Star shaped and snowflake shaped spatial multidimensional database structure is established to clarify the data distribution characteristics. Then, the initial data is preprocessed, and the consistent division of data dimension is realized through parameter initialization operation, so as to improve the data quality. Then the data classification is completed through the processes of information entropy ant colony clustering, optimization and merging. And the data with the same characteristics are gathered into the same cluster to reduce outliers. Finally, the Lagrange interpolation polynomial is established by using the basis function. And the normalization idea is introduced to ensure that the value floats in a certain range, avoid Runge phenomenon, and generate a new interpolation polynomial. The polynomial calculation result is the calibrated data value. The experimental results show that this method has good data preprocessing ability and can effectively reduce the calibration error.

Network spoofing attack detection is an indispensable link in maintaining the normal operation of campus monitoring network, but the detection process is easily disturbed by problems such as signal strength, monitoring configuration and router performance. Therefore, a spoofing attack detection method of campus monitoring network based on deep learning algorithm is proposed. The self encoder in the deep learning network is used to reduce the dimension of the campus monitoring network traffic data, and the stack encoder composed of the self encoder is used to extract the features of the reduced dimension traffic data, the extracted features into is input the confidence neural network, the type of network spoofing attack is judged according to the comparison between the output confidence value and the fixed threshold, and the detection of campus monitoring network spoofing attack is completed. The experimental results show that the proposed method has the advantages of short detection time, high detection rate and low false alarm rate. 

Because of the high cost of artificial fruit and vegetable picking and slow development of domestic fruit and vegetable picking robots, an intelligent robot is designed to reduce labor costs and time costs. According to the College Students’ Innovative Entrepreneurial Training Program, the design process of the fruit and vegetable piking robot is studied, using NVIDIA Jetson Nano development board to improve VI( Visual Identity) computing efficiency. The VI algorithm adopts the YOLO V5s algorithm and the motion control part to analyze the forward kinematics solution of the 5-DOF( Degree-Of-Freedom) mechanical arm, the forward kinematics formula of the 5-DOF mechanical arm is derived, and thus finding out about its activity space. In the End-Effector section, the gripper is used to facilitate the picking of various fruits and vegetables. A clearer intelligent structure for fruit and vegetable picking is established, the speed of VI is increased by more than 10 times by using NVIDIA Jetson Nano. Through the analysis of the forward kinematics formula, the End-Effector Action is more accurate, and also the fruit and vegetable picking aiming is completely realized within an acceptable deviation.