77 results about "Nonlinear structure" patented technology

The invention relates to the technical field of pulsed ion beam processing, and discloses a positioning system for a pulsed ion beam processing optical element, which can accurately predict nonlinear structure deformation in a complex thermal environment and eliminate compensation deviation caused by a traditional linear model. A multi-sensor data fusion mechanism effectively suppresses the interference of local measurement noise on a control decision, and improves the reliability of a compensation strategy. And a closed-loop control system realizes real-time verification and dynamic optimization of the compensation effect, and machining precision reduction caused by error accumulation is avoided. The submicron executing mechanism ensures that the thermal drift compensation amount is accurately converted into platform pose adjustment, the high-precision optical element machining requirement is met, and the problem that a traditional positioning system is incomplete in model input due to the single temperature field information collection dimension is solved. Meanwhile, the dynamic characteristics of the temperature-displacement coupling relation are verified through high-precision displacement data.

The invention discloses a wind turbine blade semi-coupling aeroelastic modeling method considering nonlinear deformation, and belongs to the technical field of fan simulation calculation. In combination with calculation parameters used by three-party software, establishing a blade structured load model and a blade correction aerodynamic model, calculating an aerodynamic load and a structural load, and superposing: superposing the aerodynamic load, the gravity load and the centrifugal force load of the blade as an external load of a nonlinear structure control equation of the blade, and taking the external load as a calculation parameter; according to the method, the blade nonlinear structure model and the semi-coupling simulation process of the three-party wind turbine simulation software are constructed, and under the same simulation environment and load conditions, the blade nonlinear structure model and the three-party wind turbine simulation software can be subjected to semi-coupling simulation, so that the blade nonlinear structure model and the three-party wind turbine simulation software can be subjected to semi-coupling simulation calculation. The higher-precision non-linear motion response characteristics of the super-long flexible wind turbine blade are calculated, and a good effect can be achieved in the aspects of initial structure design and response evaluation of the large wind turbine blade.

The invention discloses an intelligent rail transit signal system and an automatic scheduling method, and relates to the technical field of rail transit, and the method comprises the steps of data acquisition, machine learning modeling, scheduling optimization algorithm based on machine learning, scheduling strategy generation, real-time signal control, automatic scheduling decision, and system monitoring and optimization. By adding the physical constraint to guide the generation process, the problem that the rail transit data generation sample does not conform to the actual physical law is solved; the problems of insufficient data generation diversity, generation mode collapse and the like are solved through methods of dynamic memory mechanism optimization generator training and the like. The problems of high-dimensional redundancy, nonlinear structural features, poor model generalization ability and the like of rail transit data are solved by using a negentropy excitation function.

The invention discloses an electrical system relay protection fault diagnosis method and device and a storage medium, relates to the technical field of relay protection, and adopts the scheme that relay protection data of a power system are efficiently processed and analyzed by combining two technologies of U-MAP and kernel ridge regression. The method comprises the following steps: firstly, performing feature selection and data preprocessing by using U-MAP, and capturing a nonlinear structure and a complex mode in data; and then, classification prediction is carried out by using kernel ridge regression, so that the accuracy and response speed of relay protection are improved. The multi-model fusion method is greatly advanced in accuracy, recall rate and F1 value. A traditional method can judge based on a single electrical quantity threshold value, is easy to interfere and is low in performance. Therefore, various indexes of the multi-model fusion method are remarkably superior to those of a traditional method.

The invention relates to the technical field of heat supply network operation efficiency evaluation, and discloses a heat supply network operation efficiency evaluation method based on data prediction. The method comprises the steps of constructing a structure transformation result matrix, establishing a graph modeling structure, extracting disturbance characteristics and constructing a graph propagation modulation mechanism, generating structure disturbance response representation, identifying a structure behavior type, completing node flow prediction in combination with a basic predictor, and performing comparative analysis on a prediction result and actual operation data. And the matching degree between the node operation state and the target heat supply demand is judged, so that the operation efficiency evaluation result of the heat supply pipe network is output. According to the method, by introducing a disturbance-driven graph propagation mechanism and a multi-mode structure recognition strategy, the modeling capability and prediction accuracy of nonlinear structure response in a complex heat supply system are effectively improved, and support is provided for real-time evaluation and efficiency management of the system operation state.

The embodiment of the invention discloses a nonlinear system dynamic characteristic calculation method based on a structural dynamic modification method, and the method achieves the high-precision reduced-order modeling of a nonlinear structure through enabling a complex nonlinear effect to be equivalent to an additional stiffness field and combining with a structural dynamic modification algorithm. And based on the decoupled linear substructure modal space, a nonlinear correction operator is introduced, and the global dynamic response of the nonlinear structure is solved through a resolving method after modification. Structural vibration frequency, vibration shape and energy dissipation characteristics under a nonlinear effect can be efficiently obtained, and dynamic characteristics under different nonlinear amplitudes can be obtained. Compared with a traditional iteration method, the method has the advantage that the calculation complexity is remarkably reduced by equivalently linearizing a nonlinear part.

The invention discloses a constitutive model constrained physical information neural network seismic response prediction method, and relates to the field of artificial intelligence and big data analysis, a CM-PINNs framework shows reliable performance in the aspect of predicting nonlinear structure response under seismic excitation, and compared with an existing PhyLSTM model, the method has the advantages that the reliability is high, and the reliability is high. Excellent performance is shown in the aspects of prediction precision and robustness; the framework has good expandability, can be expanded from a single-degree-of-freedom system to a multi-layer shearing building structure, and provides an efficient and reliable method for earthquake response prediction of a complex structure system.

The invention discloses a data processing method and device, a data retrieval method and device and a storage medium, and relates to the technical field of document retrieval, and the method comprises the following steps: constructing a plurality of data nodes corresponding to non-linear structure data according to a hierarchical relationship of the obtained non-linear structure data; determining description information corresponding to each data node; generating semantic descriptions corresponding to the data nodes of which the hierarchical relationship is a father-son relationship; segmenting the nonlinear structure data according to each data node, the description information corresponding to each data node and each semantic description to obtain corresponding fragments; determining a fragment vector corresponding to each fragment; and storing all the fragment vectors and the hierarchical relationship into a vector database, and constructing an index corresponding to each fragment vector. Semantic fragmentation and vectorization optimization processing can be performed on nonlinear structure data of a tree structure and a graph structure, a data basis is provided for retrieval of the nonlinear structure data, and accurate retrieval of the nonlinear structure data is realized.

The invention provides a structure vulnerability analysis method and system based on earthquake magnitude, distance and site conditions, a storage medium and a computer program product. The method comprises the following steps: adopting a preset sampling algorithm to obtain preset earthquake magnitude, distance and site conditions, and acceleration time history samples corresponding to the preset earthquake magnitude, distance and site conditions; based on the acceleration time-history sample, performing nonlinear dynamic time-history analysis on a to-be-evaluated structure by using a finite element analysis method to obtain nonlinear structure seismic response corresponding to the to-be-evaluated structure and the acceleration time-history sample; establishing a change relationship between the statistical parameters of the seismic response of the nonlinear structure and the preset magnitude, distance and site conditions by using a regression fitting method; and constructing a vulnerability curved surface of a to-be-evaluated structure based on the preset earthquake magnitude, distance and site conditions based on the statistical parameters of the seismic response of the nonlinear structure and the change relationship between the preset earthquake magnitude, distance and site conditions. According to the embodiment of the invention, the four elements of a seismic source, a propagation path, a site and a structure can be completely coupled, and the method can be suitable for earthquake disaster risk analysis and rapid evaluation of post-earthquake disasters under various medium-short period structure systems.

The invention discloses an intelligent dynamics rapid simulation method. The invention relates to the technical field of finite element simulation, and solves the problem that real-time prediction of high-dimensional nonlinear structural dynamics and online inversion of impact acceleration cannot be met in the prior art. According to the invention, based on a deep learning algorithm, structure-level reconstruction of an existing operator learning normal form is realized through a mapping thought from a graph structure depth operator function to a function. According to the method, the double-tower type branch-trunk topology on which DeepONet leans is thoroughly got rid of; according to the method, a space correlation mode and a time evolution rule are synchronously coded in a hidden space and are directly projected to a function at a target moment to be expressed without following a frequency domain kernel integral path of the FNO instead of an end-to-end serial progressive single calculation path. According to the design, the function-to-function micro-mapping capability is reserved on the semantics of operator approximation, and induction deviation caused by structure splitting of a frame is overcome in a graph domain time sequence task.

The invention relates to the technical field of gene editing, and discloses an sgRNA activity prediction method and device based on deep features and automatic machine learning. The method comprises the following steps: acquiring sgRNA training data with editing efficiency labels and discretizing the sgRNA training data into classification labels; after the sgRNA and the PAM sequence are spliced, inputting the spliced sgRNA and PAM sequence into a pre-training depth sequence model to extract high-dimensional features; obtaining a low-dimensional feature vector through nonlinear dimensionality reduction; an automatic machine learning framework is combined with cross validation, and a prediction model is automatically trained and optimized based on low-dimensional features; and finally, carrying out activity prediction and sorting screening on candidate sgRNA by utilizing the model. According to the method, through fusion of deep semantic extraction, nonlinear structure maintenance and a full-automatic modeling process, the accuracy, robustness and development efficiency of sgRNA activity prediction are effectively improved, and an efficient pilot screening tool is provided for a gene editing experiment.

The invention relates to a multi-target equivalent static wind load calculation method suitable for a nonlinear structure. The method comprises the following steps: firstly, carrying out a nonlinear structure wind tunnel test and establishing a corresponding finite element model; secondly, obtaining an average component of the equivalent static wind load through a wind tunnel test result, obtaining an equivalent static wind load not including the average component through time-history analysis of a finite element model, namely a pulsation component, and obtaining a multi-target linear equivalent static wind load through combined analysis of the pulsation component and the average component; and finally, on the basis of the multi-target linear equivalent quiet wind load, correcting the multi-target linear equivalent quiet wind load in combination with a nonlinear coefficient, calculating to obtain a multi-target nonlinear equivalent quiet wind load, and expressing the multi-target nonlinear equivalent quiet wind load as a wind pressure coefficient form. The simultaneous equivalence of multiple extreme responses of a nonlinear structure is realized, and a convenient and accurate equivalent static wind load is provided for engineering design.

The invention relates to a structure period segment division method based on structure dynamic response similarity, which is characterized in that on the basis of structure dynamic response under seismic action, objective division of structure period segments is realized by constructing response characteristic matrixes corresponding to different structure periods and introducing a fuzzy clustering mechanism to perform unsupervised classification on the structure periods. According to the method, linear and non-linear structure response characteristics are considered at the same time, the optimal period segmentation number and the boundary threshold are determined in combination with the clustering effectiveness evaluation index, and a more accurate thought and method are provided for more rapidly evaluating seismic damage of the structure in the specific period range and matching input vibration of the structure in the specific period range.

The application discloses a kind of sensor fault detection method and device of structural health monitoring system, it is related to structural health monitoring technical field, comprising: obtaining the nonlinear structure monitoring data of sensor system collected by structural health monitoring system, nonlinear structure monitoring data is handled using improved geometric PNL hybrid model, obtain linear to-be-separated mixed signal, using FastICA model to process to-be-separated mixed signal, obtain multiple independent elements, determine the separation matrix of improved geometric post-nonlinear independent component analysis model according to multiple independent elements, using the processing of improved geometric post-nonlinear independent component analysis to real-time collection nonlinear structure monitoring data, determine whether sensor system exists fault and the sensor of fault occurrence.The method can still complete linearization processing to sensor mixed signal under the condition that prior knowledge is unknown, compared with simple linear ICA analysis algorithm, it is more suitable for complex nonlinear structure.

In a dynamic response test method for a nonlinear structure using a virtual signal provided in the present application, a load is applied to the free end of the nonlinear structure by an actuator according to the target dynamic parameter A(t) of the nonlinear structure, and then an excitation signal is applied to the nonlinear structure to capture the dynamic response of the nonlinear structure. The present application uses a virtual signal to control the actuator to generate a target load to load onto the nonlinear physical structure to simulate target parameters, such as stiffness, damping, and the force under action. Load simulation can be accurately performed through virtual signals, and accurate load simulation can finely control and adjust dynamic parameters such as damping and stiffness to solve the problem of sensitivity to dynamic parameters of nonlinear structures.

The present invention discloses a method, a system and a storage medium for early fault detection of a hydroelectric generating unit, belonging to the technical field of mechanical fault diagnosis methods. The method includes: collecting vibration data of a water turbine and using the vibration data as source data; performing blind source decomposition on the source data based on a binary decomposition method to decompose it into multiple source signals; respectively performing surrogate data processing based on the fast Fourier transform on the decomposed multiple source signals to obtain surrogate data; and performing non-linear structural damage determination and fault propagation path drawing based on the surrogate data. The present invention is beneficial to the discovery and maintenance of early faults, thereby reducing the maintenance cost of the hydroelectric generating unit.

The invention relates to a nonlinear structure reduced-order model construction method based on sparse recognition and a mixed mode, belongs to the technical field of structural dynamics analysis and aeroelastic mechanics analysis, and solves the problem that complex motion caused by geometric nonlinearity under large deformation cannot be accurately described in the prior art. Comprising the following steps: S1, establishing a nonlinear finite element model of a target large flexible wing to obtain a training data set; s2, solving a mixed modal basis based on the displacement residual error and SVD (Singular Value Decomposition); s3, establishing a nonlinear stiffness coefficient solving problem model, introducing LASSO regression to establish an LASSO regression optimization objective function, and solving to obtain a sparse nonlinear stiffness coefficient; s4, based on the sparse nonlinear stiffness coefficient and the structural kinetic equation, establishing a nonlinear structure reduced-order model; and S5, applying the nonlinear structure reduced-order model to statics response solution and dynamics response solution of the large flexible wing to obtain statics response and dynamics response results.

The invention discloses a safety analysis method and system based on rotation symmetry and degree of freedom in S box linearization. The method comprises the following steps: step 1, obtaining an S box with rotation symmetry and a differential path probability corresponding to the S box according to a target algorithm; 2, according to the S box with rotation symmetry and the differential path probability corresponding to the S box, constructing a linearized round number inequality, and solving the linearized round number inequality; and 3, according to the solution of the linearization round number inequality, judging whether the S-box linearization technology can be utilized to attack the target algorithm or not. According to the method, whether a general target algorithm can be attacked by using an S box linearization technology or not can be quickly judged, the affine subspace of the S box can be quickly obtained and linearized, the number of rounds capable of being linearized is calculated, the method is suitable for all cryptographic algorithms with non-linear structures, the efficiency of S box linearization attack is improved, and the method has good application prospects. And resource waste is reduced.

The invention discloses a method for predicting the association between lncRNA and disease based on a collaborative matrix decomposition of high-order perturbations and flexible ternary regularization, which belongs to the technical field of matrix decomposition and biological gene combination. The method comprises the following specific steps: obtaining the association matrix Z of lncRNA and disease; calculating the functional similarity matrix R of lncRNA based on the known association matrix Z; s and the semantic similarity matrix D of the disease s ; The multi-order correlation matrix calculated based on the known correlation matrix is ​​added to the known correlation matrix as a high-order perturbation; the global nonlinear structural similarity information of the similarity matrix is ​​obtained through similarity learning of flexible ternary regular terms; the objective function is optimized and iterated based on the collaborative matrix decomposition method to obtain the final prediction score matrix, so that the final prediction of lncRNA-disease association is more accurate.

The present invention discloses a method for identifying impact excitation loads of nonlinear structures based on a hybrid regularization strategy, which is specifically implemented according to the following steps: first, obtaining a transfer matrix between the impact excitation position and the vibration response measurement point position; measuring the vibration response of the nonlinear mechanical structure under the impact excitation load to be identified; constructing a sparse characterization model; solving the constructed sparse characterization model using a convex optimization algorithm to obtain the expansion force acting on the nonlinear mechanical structure; extracting the impact excitation load and the nonlinear force peak information from the expansion force; constructing a nonlinear response transfer equation based on the obtained impact excitation load, solving the nonlinear response transfer equation using an l2 norm regularization method, and obtaining the slow-changing trend information of the nonlinear force. The present invention solves the problem of low recognition accuracy and unstable results caused by directly using the l1 norm sparse regularization method in the prior art.

The invention provides a seismic data filtering method and device based on kernel principal component analysis and a medium, and belongs to the field of seismic data processing. The method comprises the following steps: extracting a trend time difference attribute of an input three-dimensional seismic data volume; setting a surface element size and extracting a corresponding line data volume according to the surface element size; selecting a target point and calculating surface element coordinates according to the trend time difference data; setting the size of a time window and extracting a small three-dimensional data volume in combination with surface element coordinates; and performing kernel principal component analysis on the small three-dimensional data volume to obtain a first principal component, and taking central point data of the first principal component as filtered data of the target point. According to the method, original data are mapped into a high-dimensional space through nonlinear mapping by adopting kernel principal component analysis, so that nonlinear structures and characteristics in the original data can be better reserved; in addition, the trend time difference attribute of seismic data is also considered, geologic structure data can be more accurately obtained by opening up a time window along the event trend, the problem of data discontinuity caused by the influence of stratum inclination is improved, and the filtering effect is improved.

The invention discloses a non-linear structure frequency and vibration mode coefficient identification method based on a sliding window power spectrum and Polymax. The method comprises the following steps: 1, testing to obtain an acceleration response of an engineering structure; 2, calculating an auto-power spectrum and a cross-power spectrum of each moment signal in the acceleration response signal of the engineering structure; 3, for each instant, selecting an instant power spectrum function of a specific time period by using a sliding window with a certain length; 4, calculating to obtain a mean power spectrum function of the instantaneous power spectrum functions in the selected time period range; and 5, taking the mean power spectrum function as input of polymax, and combining the stability diagram and clustering analysis to obtain an instantaneous modal frequency and a vibration mode coefficient. According to the non-linear structure frequency and vibration mode coefficient identification method based on the sliding window power spectrum and the Polymax, the outstanding advantages of the Polymax in conventional conditions, large damping and dense modal analysis can be effectively played, and the defects of an existing method in the aspects of analysis parameter selection and model overfitting can be overcome.

The invention discloses a nonlinear industrial process quality monitoring method and system, and the method comprises the steps: collecting an original data set X, and carrying out the mean removal of the original data set X, and obtaining a standard data set Xc; performing nonlinear mapping on the standard data set Xc by using kernel principal component analysis; taking the first k feature variables in the kernel feature space as kernel key component variables, and calculating a nonlinear structure between feature vectors of a kernel matrix K; calculating a projection value Fp of a kernel key component variable in a space where a kernel matrix eigenvector is located, calculating a projection value # imgabs0 # of a kernel matrix K after mean removal by combining the projection value Fp, and constructing a multivariate T2 control chart based on multivariate T2 statistics; and a preset control limit is inserted into the multivariate T2 control chart, and whether the to-be-monitored process is abnormal or not is judged based on the control limit. According to the method, the kernel principal component analysis and the multivariate T2 control chart are combined, complex nonlinear data can be effectively processed and monitored, and the sensitivity and accuracy of anomaly detection are high.

The application provides a high-order interaction prediction method and device with mixed graph deep learning, which first constructs a drug molecule graph, a microorganism weighted graph, a disease weighted graph and a supergraph connecting the three based on multi-source heterogeneous data such as drug molecular structure, microorganism classification information and disease semantic network, forming a mixed graph structure. Subsequently, through a mixed graph deep learning module fusing a graph convolution network and a supergraph neural network, nonlinear structure features and high-order interaction features of each entity are extracted, and the adaptive fusion of the features is realized by using an attention mechanism. Then, the fused deep features are mapped to the prior expectation of the latent factor matrix in the Bayesian logic tensor decomposition model, a probabilistic graph model is constructed, and the joint adaptive inference of the model parameters, latent variables and deep learning mapping is carried out through a variational expectation maximization algorithm, so that the high-order correlation probability prediction of the whole tensor space is realized without negative sampling.

The invention relates to a depth robust reconstruction method for a smooth L1 auto-encoder and a hierarchical sparse group lasso, which comprises the following steps of: firstly, constructing an SL1AE network comprising the smooth L1 auto-encoder, and introducing a smooth ReLU activation function into the network to solve a black spot problem generated when L1 loss is combined with the ReLU activation function; meanwhile, layered sparse group lasso LSGLasso regularization is added into the network, flexible control over element-level and group-level sparseness of each layer is achieved so as to capture a complex nonlinear structure and improve model robustness, an optimization algorithm based on near-end gradient descent PGD is used, a composite objective function composed of reconstruction loss and regularization items is decomposed into processable sub-problems, and the reconstruction loss and the regularization items are subjected to reconstruction. And end-to-end training is realized. A large number of experiments prove the effectiveness and universality of the method in data reconstruction in a complex noise environment.

The invention relates to the technical field of data mining, and discloses a method for improving clustering performance, which comprehensively applies technical means such as subspace learning, nonlinear processing, tensor nuclear norm constraint and consensus similarity learning. Through subspace learning, the method can reveal hidden structures and features in high-dimensional data, and effectively deals with dimension disasters. The nonlinear processing part optimizes nonlinear information in the data and reveals a deep nonlinear structure of the data. By using the tensor nuclear norm constraint, the method not only improves the performance of a clustering algorithm, but also shows a remarkable effect in practical application, for example, the recognition accuracy is improved in face recognition, analysts are helped to more accurately distinguish different consumer groups in the market analysis field and insight into consumer demands, and the market analysis efficiency is improved. Therefore, the accuracy and effectiveness of market analysis are remarkably improved.

The present invention provides a data encryption method and system based on chaotic block cipher, which relates to the technical field of data encryption. The method includes: receiving plaintext data input by a user; dividing the plaintext data into multiple groups and inputting them into a round function for encryption operation to obtain corresponding ciphertext data; wherein, a chaotic sequence is used to dynamically construct an S-box, system parameters and fixed parameters; in each round of encryption process, the constructed S-box is called to participate in data permutation, and the round key determined by the system parameters and fixed parameters is called to encrypt multiple groups of the plaintext data to generate the next-round grouped data. In this way, the chaotic sequence is used to dynamically construct the S-box in the non-linear structure part and the system parameters and fixed parameters in the linear structure part to securely encrypt the plaintext data, reduce the risk of being broken during data transmission, and improve the security of information transmission.

This invention provides a dual-motor steer-by-wire system and its active fault-tolerant control method. The method includes: a front wheel steering angle tracking controller, a vehicle stability controller, and a nonlinear model control method based on a variable parameter (LPV) model. The front wheel steering angle tracking controller and the stability controller are designed based on a linear quadratic regulator (LQR) and an optimal control algorithm, including: solving for the ideal front wheel steering angle; establishing a dynamic model of the dual-motor steering actuation system; and designing the steering angle tracking controller and the stability controller based on the ideal front wheel steering angle and the steering system dynamic model. The dual-motor active fault-tolerant control includes: comparing the control torque output by the dual-motor torque sensors with that output by the front wheel steering angle tracking controller, calculating the fault coefficient, and optimizing the steering actuator model into a variable parameter (LPV) nonlinear structure containing motor fault parameters. This invention can ensure vehicle safety after a dual-motor failure.

The invention provides a variance-based aircraft structure design parameter global sensitivity analysis method, and aims to solve the problems of high calculation cost, difficulty in quantizing parameter interaction effect and coupling propagation influence and the like in the application of existing Sobol and other methods, the method comprises the following steps: firstly, identifying key structure parameters and establishing a probability model of the key structure parameters; then, a Sobol sequence is adopted to generate a sample matrix, and output response is calculated through an aircraft structure system model; constructing a mixed sample matrix to perform variance decomposition, and respectively calculating a first-order sensitivity index and a total sensitivity index of each parameter to quantify main effect and interaction effect contribution; and finally identifying performance key parameters and interaction sensitive parameters according to the index sequence. According to the method, global analysis of a high-dimensional nonlinear structure system is realized under acceptable calculation cost, the result can provide a quantitative decision basis for lightweight design, tolerance allocation and multidisciplinary collaborative optimization of an aircraft structure, and the design efficiency and reliability are remarkably improved.

The invention provides a data feature extraction method and system based on a combined kernel sparse auto-encoder, and relates to the technical field of network communication security detection. After a first data set and a second data set are obtained, a combined kernel function is constructed to obtain a third data set, then the third data set is utilized to train a sparse auto-encoder, parameters are updated to obtain a weight matrix and a sparse cost function, then a target function is constructed and solved, and the parameters are optimized based on an adaptive genetic algorithm. The objective function of the combined kernel sparse auto-encoder is optimized through an iteration method of the adaptive genetic algorithm, continuous crossing and variation are carried out, an optimal solution is solved, and a dimension reduction result is obtained. According to the method, the kernel technology and the sparse auto-encoder are combined, and the nonlinear structure and high-dimensional characteristics of the data are fully utilized, so that efficient extraction and dimension reduction processing of network security data features are realized, the effectiveness of feature extraction is improved, the robustness and accuracy of the model are enhanced, and the network security data analysis and processing capability is improved.