320results about How to "Fully excavated" patented technology

The invention relates to the technical field of electric power short-term load prediction, in particular to an industrial park short-term load prediction method and system based on a Diffusion-Transform-AR model. Acquiring non-stationary historical load data and meteorological data of the industrial park, and performing missing value processing on the data; performing correlation analysis on the data by using the maximum information coefficient, selecting meteorological data with relatively high correlation, and reprocessing time data; performing time sequence decomposition on the load data by using a seasonal trend decomposition method to obtain trend term, seasonal term and residual term features; constructing a diffusion Transform prediction model, and taking the residual term, the meteorological data and the time data as input prediction residual; constructing an autoregression prediction model and taking the trend term as an input prediction trend; according to the method, the advantages of various models are combined, time sequence features are deeply mined, the prediction precision and the model generalization ability are effectively improved, and the method is suitable for complex industrial park load prediction scenes.

The invention relates to the technical field of big data analysis and artificial intelligence, in particular to a multi-source heterogeneous big data fusion and reasoning method and system based on a knowledge graph, and the method comprises the following steps: constructing mapping interfaces of various data sources, collecting and storing entities in the data sources, and constructing nodes in the knowledge graph; performing representation learning on a graph structure of the knowledge graph based on a graph neural network, extracting an embedded vector of a node, and extracting a fusion node through the embedded vector and a mapping interface; performing knowledge completion on the fusion nodes by using a neural network based on symbolic logic; generating candidate reasoning paths by adopting a depth-first search algorithm based on the knowledge graph; reasoning is carried out based on an attention mechanism and path similarity, candidate reasoning paths are screened and sorted to obtain an optimal reasoning path, a graph neural network is adopted to carry out representation learning of a knowledge graph, symbolic logic and the neural network are combined, the interpretability of the reasoning process is guaranteed, and the learning ability is high.

The invention discloses a surveying and mapping geographic information data management method and system based on cloud computing, and relates to the field of geographic information, and the method comprises the steps: constructing a semantic segmentation model based on a convolutional neural network Deep Lab, and carrying out the feature extraction and semantic annotation of a surveying and mapping geographic data set through the semantic segmentation model, and obtaining a geographic element set; performing spatial division and indexing on the geographic element set by adopting an R tree spatial indexing method to obtain a hierarchically organized geographic element tree; constructing a graph neural network model based on a graph convolutional network GCN, and performing topological relation modeling on the geographic element tree by using the graph neural network model to obtain a time-space attribute graph containing geographic semantics and topological relations; and constructing a space-time prediction sequence model based on the space-time diagram neural network STGCN, and performing evolution trend prediction on the space-time attribute diagram by adopting the space-time prediction sequence model. Aiming at the problem that dynamic prediction of geographic elements is difficult to realize in the prior art, the spatial-temporal evolution trend of the geographic elements is predicted, and dynamic prediction is realized.

The invention discloses a high-resolution rain condition analysis and generation method based on multi-source modal fusion deep learning, and the method comprises the steps: obtaining a multi-physical-quantity data set based on a multi-source data set related to rain conditions, and building a dimensionless multi-modal feature factor library; performing multi-dimensional non-equal weight processing on the loss function, and constructing a multi-scale space conversion model MF-ST-Unet; after model training, an hour-by-hour regional high-resolution rain condition analysis data set is finally generated; and based on the trained MF-ST-Unet model, a high-resolution rain condition analysis data set applied to covering the whole area is obtained through transfer learning knowledge distillation. According to the method, uniform grid rain condition analysis with high spatial resolution can be quickly and effectively generated, and the method is more objective and accurate compared with a traditional method; the method fully considers the multi-scale features of the rain condition and the precipitation level difference, achieves the high-resolution objective rain condition data acquisition, improves the rain condition detection capability and disaster prevention accuracy, and has a very high application value.

The invention discloses a polyp image segmentation method based on boundary clue depth fusion, and the method comprises the steps: inputting an intestinal polyp image into a backbone network through constructing a boundary clue depth fusion network, and extracting different levels of feature information, so as to guide the extraction of a polyp region; a boundary clue mining module is constructed to obtain boundary detail information by means of partial hierarchical features, and more accurate positioning of a polyp area is realized; constructing a foreground target enhancement module to enhance the learning of the network on the overall structure characteristics of the foreground target in combination with the reverse attention branch, so that the network focuses on the foreground target; and constructing a depth feature fusion module to integrate boundary information and foreground information, further improving the segmentation performance through cascade feature fusion and boundary information guidance, and finally obtaining a polyp segmentation result by using the trained model. According to the method, the limitation of the polyp in boundary uncertainty and morphological diversity is overcome, the weighted joint scale difference loss optimization model is adopted in training, and high-precision recognition and segmentation of the polyp region are realized.

The invention discloses a marine diesel engine vibration fault early warning method based on an improved VMD, and the method comprises the steps: improving the VMD through combining a frost ice algorithm RIME, carrying out the hyper-parameter optimization, carrying out the VMD signal decomposition of an optimized data set based on an optimal combination, and obtaining a plurality of IMF values, so as to reconstruct vibration signal sequence data. Performing time domain and frequency domain feature extraction on the reconstructed vibration signal sequence data to obtain a time domain feature parameter and a frequency domain feature parameter; calculating and acquiring a Pearson coefficient between each feature parameter in the mixed domain feature set and the early warning value, and selecting the feature parameter of which the Pearson coefficient meets a preset Pearson coefficient threshold as an optimized feature parameter; and obtaining a residual error between a predicted root-mean-square value and an actual root-mean-square value based on the optimal BiGRU algorithm model so as to obtain a residual error threshold range under the condition of fault-free operation of the marine diesel engine, building an RIME-VMD-BIGRU model for vibration fault early warning of the marine diesel engine, and realizing automatic prediction of the vibration fault of the marine diesel engine. According to the method, sailors can be reminded of possible faults of the ship, the dependence on expert experience is reduced, and meanwhile, compared with other early warning algorithms, the method has higher diagnosis precision.

The invention discloses an electrocardiosignal open-set identity recognition method based on a hybrid network and a cross attention mechanism, and belongs to the technical field of identity recognition. The electrocardiosignal open-set identity recognition method comprises the steps that electrocardiosignals are preprocessed to obtain standardized input; constructing a hybrid network model to extract a long-time dependency relationship between multi-scale local space information and a time sequence; fusing different types of information by using a cross attention mechanism; and establishing an Openmax classification layer to realize identity tag identification in an open set environment. By means of the method, the known individual identity can be effectively recognized, unfamiliar persons are rejected, and the accuracy of electrocardiosignal identity recognition and the applicability in the practical application scene are improved.

The invention provides a full-automatic online PAT continuous crystallization control method, a medium and a system, and belongs to the technical field of PAT continuous crystallization, and the method comprises the following steps: firstly, carrying out time sequence analysis and Bayesian network correlation analysis on collected crystallization process parameters and effect data to obtain a change optimization curve and a frequency spectrum characteristic of each parameter; and then, establishing a multi-branch neural network model by using the frequency spectrum characteristics as input, and realizing online prediction of the current crystallization process effect. And then, through genetic algorithm optimization, finding out an optimal crystallization process parameter combination, converting the optimal crystallization process parameter combination into actual production equipment control parameters, and automatically adjusting production equipment. According to the method, the online effect evaluation model based on the neural network is established by collecting and analyzing the historical data of the PAT continuous crystallization process, and the technical problems that in the prior art, self-adaptive online optimization is difficult to achieve, and dynamic changes in production are difficult to respond are solved.

The invention belongs to the technical field of a computer system based on a specific calculation model, relates to a virtual and real dynamic interaction-based digital twin bearing fault diagnosis method, and aims to solve the problem that a bearing system in large rotary mechanical equipment is large in fault hidden danger and avoid serious consequences caused by faults. A generally used bearing is in a high-precision standard, so that system signal data under bearing fault operation is not easy to collect in a factory, and the unbalanced data sample distribution directly leads to finiteness of fault diagnosis in a real scene only based on a big data driving mode classification method, so that the system signal data cannot be easily collected in the factory. The method has good scene working condition self-adaption, can be applied to fault diagnosis of a bearing system in different industrial scenes, fully excavates limited industrial data information and rich physical priori knowledge, and ensures safe and reliable operation of an important rotating mechanical system through a digital twinning technology.

The invention discloses a low-dew-point double-rotating-wheel dehumidifier energy-saving prediction method based on time-space diagram attention. The invention provides a time-space diagram attention neural network based on an attention mechanism. According to the time-space diagram attention neural network based on the attention mechanism, a diagram attention network, a bidirectional gating circulation unit and a time sequence Transform model are combined. The method comprises the following steps: firstly, constructing a graph structure according to equipment topological structure information, vectorizing a time slice input matrix, introducing an attention mechanism, and adaptively distributing node weights to update node features; adopting two forward and reverse unidirectional GRUs structures, and fusing past information and future information for feature prediction; and finally, through position embedding and a time sequence Transform structure, hidden information between sliding window inputs is mined, loss of effective features is reduced, and the long-term prediction capability of the model is improved. According to the method, the spatial-temporal correlation of original data can be better captured, the equipment operation state parameters are effectively predicted, and the energy-saving optimization control process is guided.

The invention provides a multi-view clustering method based on uncertainty and a graph neural network, and relates to the technical field of multi-view clustering. The method comprises the following steps: obtaining a multi-view data set, and preprocessing to obtain an original feature matrix and an adjacent matrix; performing feature extraction and fusion on the input original feature matrix and the adjacent matrix to obtain a consistent embedded matrix and a transfer matrix; based on a Dempster-Shafer evidence theory and Dirichlet distribution, obtaining reliable consistent embedding by reducing the uncertainty of an embedding space in the consistent embedding matrix; mapping the reliable consistent embedding and the transfer matrix into final embedding; and performing decoding reconstruction and clustering analysis on the final embedding to obtain a clustering result. According to the method, potential information in the views can be fully mined, the uncertainty in the model is reduced, the noise problem existing in a multi-view clustering task is solved, and the clustering result of the multi-view data can be quickly, efficiently and accurately obtained.

The invention belongs to the technical field of computer application, and relates to an operation inspection equipment fault detection method and system based on an improved deep learning model, and the method comprises the steps: obtaining original data, employing EEMD to decompose the original data for feature enhancement, obtaining a decomposition signal, carrying out the noise reduction of the decomposition signal, and obtaining a feature signal after noise reduction; according to the feature signals after noise reduction, a deep learning model is constructed based on Resnet and a dynamic Transform encoder, and the prediction probability of each fault category is output; a ZOA optimization algorithm is used to optimize hyper-parameters of a deep learning model, an optimization model is obtained, the performance of the optimization model is evaluated, an optimal model is obtained, and the prediction probability of each fault category is output, so that false report and missing report of faults can be effectively reduced in a fault detection task, potential safety hazards caused by the fact that the faults are not found in time are reduced, and the fault detection efficiency is improved. Therefore, the stability and reliability of the power system are improved.

The invention belongs to the technical field of artificial intelligence, and relates to an occupational interest evaluation method and device, computer equipment and a storage medium, and the method comprises the steps: obtaining personal basic information, existing occupational experience and future occupational tendency of a user; a question recommendation model is adopted, and according to the personal basic information, the existing occupational experience and the future occupational tendency, occupational interest evaluation questions for the user are obtained; rewriting the occupational interest evaluation question according to the question rewriting cue word by adopting a language model to generate a rewritten evaluation question; obtaining reply content of the user for the rewritten evaluation question; and predicting the occupational interest tag of the user according to the reply content by adopting a tag classification model. In addition, the invention also relates to a block chain technology, and data such as personal basic information, existing occupational experiences and future occupational tendencies can be stored in a block chain. According to the invention, the participation degree of the user and the accuracy of the evaluation result can be improved.

The invention provides a child abdominal common solid tumor activity image evaluation method and device and a medium, and the method comprises the steps: capturing the change information of a focus on a vein phase image after plain scanning and enhancement through a 3D convolutional neural network model, carrying out the intelligent silhouette, and generating an enhancement region and the enhancement degree of the focus after chemotherapy, thereby achieving the evaluation of the activity of a child abdominal common solid tumor. And finally, integrating all the information to perform intelligent diagnosis, judging the tumor activity, and forming an artificial intelligence auxiliary activity evaluation scheme after chemotherapy of the abdominal solid tumor. According to the method, the automatic enhancement degree quantitative evaluation of the hepatoblastoma, the nephroblastoma and the neuroblastoma of the children is realized through automatic tumor segmentation and registration subtraction image calculation by utilizing the plain scanning and enhanced CT image, so that the focus characteristics are extracted from multiple angles, the image information is fully mined, and the accuracy of the diagnosis is improved. Accuracy and sensitivity of activity discrimination after chemotherapy of the solid abdominal tumor of children can be obviously improved.

The invention discloses a pixel equivalent measuring method based on mobile terminal pupil distance intelligent measurement, which comprises the following steps: calibrating a camera, and correcting distortion through internal and external parameters obtained through calibration; image preprocessing, including image pre-identification and image enhancement, and highlighting the edge and angular point features of the image; coarse positioning of the calibration object: acquiring rough positions of corner points at four corners of a square frame of the calibration object; carrying out angular point coordinate coarse positioning, and carrying out mask processing on the image by taking an obtained angular point and a two-dimensional neighborhood thereof as a range; refining the coordinates of the angular points, correcting and compensating the angular points, and obtaining the coordinates of the four corrected angular points; performing perspective transformation; determining a pixel equivalent and a transformation matrix; carrying out verification measurement; and verifying and optimizing the determined pixel equivalent. According to the invention, an image processing technology is utilized to solve the calibration problem during the pupil distance measurement of the mobile end, and a small-range straight angle point precision compensation method is utilized to automatically compensate error defects, so that the pupil distance measurement precision is improved, and a new idea is provided for the existing method.

The invention discloses an aerial image camouflage target detection method and device based on a multi-scale attention and boundary sensing network, equipment and a medium. A pyramid vision converter (PVT) is adopted as a trunk network, multi-scale features are extracted as attention-based feature extraction, and a high-resolution feature map is generated; a boundary sensing module (BAM) is developed, and the module aggregates local boundary information and global position information so as to fully mine boundary semantic information of the camouflage target; designing a boundary guidance module (BGM) which integrates the boundary features of the boundary sensing module with the multi-scale features of the backbone network to guide feature learning; a context aggregation module (CAM) is constructed, the module aggregates multi-layer fusion features, context semantic information of a camouflage target is fully mined by adopting extended convolution, and feature representation is further enhanced. According to the method, the problem of detecting the aerial image camouflage target is solved.

In order to improve the accuracy and reliability of impact damage identification based on machine learning, the invention provides impact damage identification based on multi-modal data fusion and ensemble learning. According to the method, multi-modal data including discrete impact damage characteristics, a time-domain impact signal sequence and a time-frequency image of an impact signal are adopted as input of a machine model, and impact damage information contained in an impact stress wave signal is fully mined; and the artificial neural network, the long-short-term memory network and the convolutional neural network are integrated by using a weight average technology, and impact damage is identified by fusing three distances including a cosine distance, a Manhattan distance and an Euclidean distance. The method is verified on a 400mm * 400mm composite material platy structure, and the result shows that the method can perform high-precision identification on the impact damage in the structure under the condition of using a small number of sensors.

The invention relates to the field of cold end optimization control systems, in particular to a method for reconstructing a wind field of an axial flow fan group of a direct air cooling unit based on a physical information neural network. Comprising the following steps: S1, arranging a plurality of wind speed and pressure sensors near an axial flow fan group of the direct air cooling unit, and collecting sample data when axial flow fans rotate at a high speed by taking wind speeds and pressures in all directions of a wind field of the axial flow fan group of the unit as target variables; s2, performing data preprocessing on the sample data; s3, determining a differential equation of fluid motion satisfied by the axial flow fan group area, and constructing a physical information neural network of the axial flow fan group; s4, training a physical information neural network model of the axial flow fan group by using the preprocessed data obtained in the S2; and S5, reconstructing a wind field of the axial flow fan group of the direct air cooling unit according to the trained physical information neural network model of the axial flow fan group in the step S4. According to the method, the wind field data of the axial flow fan group of the direct air cooling unit is quickly and accurately predicted by using a physical information neural network method.

The invention discloses a blind hyperspectral unmixing method and system based on end member query design and a Transform attention mechanism, and relates to the field of hyperspectral image processing. The problems that a current simplified mixing mechanism is mainly a linear mixing model, ground feature abundance and corresponding end member accuracy are low, a deep learning blind hyperspectral unmixing method depends on good end member initialization, and unmixing performance is seriously reduced during end member initialization are solved. The method comprises the following steps: constructing an abundance mapping module, extracting features in a hyperspectral image, and performing abundance decoding on the features to form an estimated abundance image; mapping the abundance graph into abundance embedding; constructing a Transform module fitted by a hybrid mechanism, and combining end member query, reconstruction query and abundance embedding through the Transform module fitted by the hybrid mechanism to obtain reconstruction query of end member information and abundance information; the original hyperspectral image is remodeled according to the reconstructed query. The method is applied to the field of blind hyperspectral image unmixing.

The invention relates to the technical field of natural language processing, in particular to a method and system for achieving intelligent number asking based on a large model, and the method comprises the following steps: integrating data from various data sources including a database, a data warehouse, WEB service and Excel files; the data is cleaned, converted, standardized and normalized through the data management center, so that the data quality is ensured, and basic data support is provided for subsequent intelligent question analysis; the beneficial effects are that the method for realizing intelligent number asking based on the large model provided by the invention responds to complex and changeable data requirements in real time, helps an enterprise to quickly make a data-based decision, and reduces the time cost of waiting for the assistance of an IT department or a data analysis team; through effective integration and intelligent search of mass data resources, various data in an enterprise can be fully mined and utilized, and idleness and waste of the data resources are avoided.

The invention belongs to the technical field of traffic, and particularly relates to a traffic flow prediction model for realizing co-mining of traffic point semantics and network topology. The name of the model is STIGNN, the STIGNN focuses on coupling of topological relations between traffic point features and traffic points in the aspect of spatial feature aggregation, and potential and existing spatial dependencies are comprehensively considered. The STIGNN can process a long-time sequence through a time convolution layer based on gating, and the model can be directly applied to an inductive learning task by introducing a dual attention mechanism of node features and a topological relation into a space-time framework, and can be popularized to any network with complex entities and relations. Experimental results on two public real traffic network data sets METR-LA and PEMS-BAY show that the STIGNN is superior to an advanced baseline model.

The invention belongs to the technical field of underwater sound, and particularly relates to an anti-bottom echo interference target detection method and system suitable for a multi-beam sounding sonar, and the method comprises the steps: constructing a binary sparse hypothesis test problem based on to-be-detected data and a constructed beam domain sparse signal model, the to-be-detected data is beam data corresponding to a certain sampling moment in the current frame; solving the background noise power, and solving the sparse vector based on the block sparse characteristic of the target and Bayesian learning; and judging whether a target exists in the to-be-detected data or not by using a generalized likelihood ratio detector. In the design process, priori information such as a beam data model and block sparse characteristics of a target is considered and utilized, so that the priori knowledge of the target is fully mined, and the sensing capability of the sonar to the environment is improved; interference suppression and target detection can be realized only based on the beam data of the current frame without using auxiliary data of front and back frames; parameters needing manual adjustment do not exist, and the method can be directly applied to an unmanned platform.

The invention discloses an urban traffic flow prediction method based on deep learning. The technical scheme of multi-source heterogeneous traffic data fusion and multi-scale spatial-temporal feature extraction is adopted. Comprising the following steps: acquiring multi-source traffic data through a vehicle-mounted GPS, road test equipment and a traffic monitoring video, performing data cleaning, space-time alignment and feature extraction, and performing fusion by adopting methods such as Kalman filtering and the like to obtain unified traffic state characterization; constructing a network model comprising three LSTM units with different scales, and modeling short-term, medium-term and long-term change characteristics of traffic flow by taking 5-10 minutes, 1 day and 1 week as time steps respectively; an Attention mechanism is introduced to carry out adaptive fusion on the multi-scale features; mapping various external environment data into low-dimensional feature vectors through representation learning, and introducing the low-dimensional feature vectors into the model; a knowledge distillation technology compression model is adopted, online service is deployed, and traffic flow prediction of the next one hour is generated every 5-10 minutes. The method improves the accuracy and real-time performance of traffic flow prediction, and has a good engineering application value.

The invention discloses a high-frequency ground wave radar formation target tracking method and system, and particularly relates to the technical field of target tracking. Comprising the following steps: constructing a position feature vector and an appearance feature vector of a formation target based on high-frequency ground wave radar echoes; constructing nodes by using the position feature vectors and the appearance feature vectors of the formation targets; screening adjacent nodes of the known node Node based on the position feature vector, and calculating a likelihood function; constructing a prototype network model based on the formation target appearance feature vector, and calculating the Euclidean distance between the appearance feature vector of a known node Node and the appearance feature vector of an adjacent node Adjacent; on the basis of a likelihood function Lsiran and an Euclidean distance Euddis, the correlation similarity between a known node Node and an adjacent node AdJacent is calculated, so that a connection edge is constructed; and training based on a graph neural network loss function to realize formation target matching tracking. By adopting the technical scheme of the invention, the problem that the formation target cannot be tracked in the prior art is solved, and accurate tracking of special targets such as formation is realized.

The invention provides a knowledge distillation passive domain unsupervised domain self-adaptive method, which realizes classification of a target domain without using source domain data by designing a teacher-student model and a competition mechanism. Firstly, a teacher network composed of a feature extractor, a classifier and a rotation classifier is constructed, source domain data is used for training, classifier parameters are fixed, a source model is finely adjusted through a self-supervised rotation sampling method, and a pseudo label of target domain data is generated through a clustering method. And then designing a student network composed of a feature extractor and a classifier, generating a pseudo tag only by using target domain data, and optimizing the discrimination and diversity of the tag by maximizing mutual information loss. And finally, screening the pseudo labels of the teacher-student network by using a competition module, and extracting the pseudo labels with higher confidence to update the student network, thereby completing knowledge migration. According to the method, source domain data is not needed, data storage and transmission cost is reduced, privacy and data security are improved, and unsupervised classification learning on unlabeled target domain data is supported.

The invention relates to the field of multi-modal feature processing, in particular to a multi-modal feature generation method and device, electronic equipment and a storage medium. According to the method, the audio and image data are extracted from the original video and are respectively coded and processed, so that the data characteristics of different modalities can be fully mined, and the limitation that a single-modal model can only process one type of data is avoided. And meanwhile, a prompt text containing large language model task content and different modal feature association relationships is also obtained, so that a clear task direction can be provided for the model, and the model is guided to better fuse multi-modal information. Then different modal features and prompt texts are spliced to obtain multi-modal features, the multi-modal features serve as input of a large language model, the large language model can receive and process feature information of multiple types of data such as texts, images and audios at the same time, the defects of a single modal are overcome by fusing the different modal features, and the multi-modal features are obtained. Therefore, the accuracy of processing complex tasks is effectively improved.

The invention relates to a knowledge graph construction method and system for multi-modal cultivated land data. The method comprises the following steps: constructing a cultivated land protection field knowledge system; entities, relations and attributes are extracted and stored in a knowledge graph, corresponding description information is obtained and input into an incomplete multi-modal data fusion algorithm model based on deep semantic matching for data feature fusion, cultivated land data in different modals are mined, the knowledge graph is supplemented and adjusted, and the knowledge graph of the multi-modal cultivated land data is constructed. Entities, relationships and attributes are extracted and stored in the knowledge graph, so that the constructed knowledge graph can be ensured to be more accurate, the coverage field is wider, and the application of the knowledge graph is more diversified and accurate; data feature fusion is carried out in an incomplete multi-modal data fusion algorithm model, so that the heterogeneous problem in the knowledge graph is effectively solved; the cultivated land data in different modes are fully mined, and the knowledge graph can become more powerful in tasks such as self-complementation and intelligent question and answer.

The invention discloses a load prediction method based on combination of CNN-BiLSTM and XGBoost, and belongs to the field of power load prediction. The method comprises the following steps: acquiring historical load data and historical meteorological data, and preprocessing the historical load data and the historical meteorological data to obtain preprocessed historical load data; decomposing the preprocessed historical load data by adopting a multiplication seasonal adjustment method to obtain a trend sequence, a seasonal sequence and an irregular sequence; a CNN-BiLSTM hybrid model is constructed, and a trend sequence is predicted; constructing an XGBoost model, and predicting a seasonal sequence and an irregular sequence; and multiplying the predicted trend sequence by the predicted seasonal sequence and the irregular sequence, and evaluating a result to obtain predicted load data, thereby realizing load prediction based on combination of CNN-BiLSTM and XGBoost. According to the method, the CNN-BiLSTM hybrid model is used, on the basis of unidirectional flow of the LSTM network from the past to the future, the data flow direction from the past to the past is added, the time sequence feature structure of the data is fully mined, and the precision and generalization ability of load prediction are improved.

The invention belongs to the field of universal software fuzz testing, and discloses a dynamic clustering fuzz testing method and system based on efficient byte exploration. According to the method, fuzzy test seed bytes are subjected to clustering analysis by using a density-based spatial clustering algorithm, the positions of key bytes are identified, and the effective variation times of the key bytes are recorded. Then, corresponding weights are distributed to the bytes according to the clustering result and the number of effective variation times of the clustering result, an energy distribution strategy is dynamically adjusted in combination with the path coverage rate and path novelty information, and the exploration efficiency of a new path is improved. And performing mutation on the high-weight bytes to generate effective seeds of a new path, locking key byte positions of the effective seeds, performing normalization processing on other positions, and continuing mutation operation. The test strategy is continuously optimized through repeated execution of the steps, and the fuzzy test efficiency and the vulnerability discovery capability are effectively improved.

The invention relates to an electroencephalogram signal automatic sleep staging method based on a residual network and multi-level attention, and belongs to the technical field of sleep monitoring, and the method comprises the following steps: S1, constructing a sleep data set, and carrying out data preprocessing; s2, building a single-channel electroencephalogram automatic sleep staging network model based on deep learning; s3, training the model by using a gradient descent method; s4, sleep staging of the single-channel electroencephalogram signals is achieved through the sleep staging model with the optimal parameters. According to the method, multi-scale feature extraction is realized by using convolution kernels of different sizes, signal features are extracted more comprehensively by combining a style pooling-based recalibration module with a space attention module, and time sequence information is captured and important sequences are highlighted by using a gated loop network fused with a self-attention mechanism. And finally, the sleep staging accuracy is improved.