68results about How to "Fully capture" patented technology

The invention provides an unstructured data automatic processing method and system based on deep learning, and relates to the technical field of data processing, and the method comprises the steps: obtaining unstructured data through an intelligent data collector, building a data feature fingerprint database, constructing a data quality evaluation model through an incremental learning algorithm, and carrying out the data standardization processing; extracting dominant features and recessive features by using a multi-layer cascade network structure and a self-attention mechanism, and constructing a feature association graph and a bidirectional reasoning channel; and performing rule matching and causal reasoning based on the domain knowledge base, and finally realizing automatic structuring of the unstructured data. According to the method, the automation degree and the accuracy of data processing are improved, the manual intervention cost is reduced, and the method has relatively high universality and expandability.

The invention discloses a remote sensing data processing method and system, a product and a readable medium, and relates to the field of remote sensing data processing, and the method comprises the steps: carrying out the registration and correction of remote sensing images in different time periods, and obtaining a remote sensing image spatial-temporal index; calculating shadow ranges generated by each building in the target area at different times based on the building height data and the remote sensing image spatio-temporal index, and superposing the shadow ranges corresponding to the buildings to obtain a shadow coverage frequency map; calculating a shielding degree score of an area of each time phase in the shadow coverage frequency diagram based on a remote sensing image spatio-temporal index, and selecting an optimal time phase with the lowest shielding degree as an optimal observation time phase according to the shielding degree scores; based on the optimal observation time phase, dividing a region in the shadow coverage frequency diagram into a plurality of target sub-regions according to a shielding degree score; and repairing each target sub-region by adopting a preset repairing strategy according to the shielding degree score. By implementing the method, the accuracy of restoring the shadow-shielded region in the remote sensing image can be improved.

The invention discloses a remote sensing image ground object segmentation method based on hierarchical feature extraction and decoding, which relates to the technical field of remote sensing image processing, and comprises the following steps: step 1, collecting and preprocessing a remote sensing image, and constructing a training sample set; 2, constructing an image segmentation network, wherein the image segmentation network comprises a hierarchical encoder, a receptive field module unit and a network decoder which are connected in sequence; 3, training and optimizing the image segmentation network by adopting the training sample set to obtain a remote sensing image ground object segmentation model; and 4, collecting a remote sensing image to be segmented, and inputting the remote sensing image to the remote sensing image ground object segmentation model to obtain a ground object segmentation result. According to the method, the advantages of the advanced visual basic model and the U-shaped network architecture are combined, so that rapid and accurate segmentation of the ground target is realized.

The invention provides an influence maximization method and system based on deep reinforcement learning and a graph neural network, and belongs to the field of influence maximization in social network analysis. The method aims to optimize selection of seed nodes through a reinforcement learning technology so as to realize maximum propagation of information in a network, and the graph neural network can accurately capture an interaction relationship between nodes in a graph structure by virtue of a message transmission mechanism of the graph neural network, which is highly matched with an information propagation process in a cascade propagation model. Therefore, in the influence maximization model based on the deep reinforcement learning and the graph neural network, the graph neural network is combined with the deep reinforcement learning, the embedded representations of the nodes are learned by using the graph neural network, and the embedded representations are used as the input of the deep Q network so as to fully capture the complex structure and node relationship in the network. In addition, a multi-head attention mechanism is introduced to aggregate information of each round, expression of key features is enhanced, and information redundancy is effectively reduced.

The invention provides an aero-engine residual life prediction method and system based on deep learning, and relates to the technical field of engine life prediction. According to the method, a multi-scale convolution feature extraction module (SDMCN module) based on a graph attention mechanism and a GRU module (F-GRU module) fused with the feature attention mechanism are established. Through an SDMCN module, the spatial relevance of the features is fully extracted from the two aspects of the relevance between the feature maps formed under the convolution kernels of the same size and the relevance between the feature maps formed under the convolution kernels of different sizes. And meanwhile, the F-GRU module is used for dynamically adjusting the weights of the input characteristics under different time steps, so that differential updating of the sequence data is realized. By adopting the two parallel branches, extraction of spatial relevance and time sequence dependency relationship between features is realized, so that feature degradation information is captured more comprehensively, a more accurate life degradation model is established, and the accuracy of model prediction is improved.

The invention provides a multi-source remote sensing data classification method, which belongs to the field of hyperspectral image processing in the field of remote sensing, and comprises the following steps: acquiring remote sensing data of various target objects; extracting shallow layer features of each kind of remote sensing data; obtaining a plurality of frequency characteristics of each kind of remote sensing data; fusing the same frequency features of all remote sensing data to obtain a plurality of same frequency fusion features; splicing the plurality of same-frequency fusion features to obtain a multi-source fusion feature; enabling the multi-source fusion features to sequentially pass through a frequency modulation layer and an attention layer which are superposed to obtain fused global features and local features; and weighting the global features and the local features in the spectral dimension to obtain a prediction classification result of the target object. According to the method, directional frequency characteristic decomposition and fusion of the multi-source data are realized, the multi-source fusion characteristics are extracted based on the multi-source remote sensing data to serve as the basis of classification, more comprehensive information is provided for classification of target objects, complex ground object characteristics can be fully captured, and the classification result is more accurate.

The invention relates to the technical field of lithium battery water content detection, in particular to a lithium battery water content detection method and system. The method comprises the following steps: acquiring a lithium battery surface thermal imaging image; carrying out temperature gradient distribution identification on the lithium battery surface thermal imaging image, and carrying out local distribution change sudden change detection to extract a temperature imbalance region and a temperature conventional region; carrying out optimal penetrating electromagnetic wave frequency calculation according to the temperature imbalance region and the temperature conventional region, carrying out a multi-band electromagnetic wave penetration test, and collecting electromagnetic reflection signals of the temperature imbalance region and the temperature conventional region; and performing multi-frequency-band reflection signal intensity calculation and region-by-region comprehensive reflection characteristic mining on the electromagnetic reflection signal to obtain a temperature imbalance region signal reflection characteristic and a temperature conventional region signal reflection characteristic. According to the invention, efficient and accurate lithium battery water content detection is realized.

The invention discloses a vehicle-mounted multi-target classification and detection method based on multi-teacher distillation and dynamic reasoning, and the method comprises the steps: collecting an image through a vehicle-mounted image collection device, taking the image as the input through a dynamic neural network, and outputting the classification and bounding box regression of multiple targets. The dynamic neural network is composed of a multi-teacher network and a student network. The multi-teacher network is composed of a plurality of teacher models; a dynamic weight distribution system based on an attention mechanism is introduced into the multi-teacher network, and different weights are distributed to each teacher model according to the characteristics of the current input scene; a confidence coefficient threshold value is set in a multi-teacher network, and regional knowledge possibly containing a target is only taught to a student network. The student network is composed of a strategy network and a detection network, and the detection network is a two-stage target detection network obtained by integrating a feature extraction network and a target positioning network; the strategy network adaptively extracts task features according to knowledge provided by the multi-teacher network; and the detection network dynamically selects a route according to the policy network decision vector.

The invention relates to a hyperspectral image classification method, in particular to a hyperspectral image classification method based on a greedy strategy guided graph self-attention network. The invention provides a hyperspectral image classification method of a graph self-attention network based on greedy strategy guidance in order to solve the problems that in existing HSI classification, samples are limited, GAT tends to only consider directly-connected nodes, indirectly-related relations are insufficiently considered, and the classification effect is affected. The method comprises the following steps: step 1, constructing a graph self-attention network based on greedy strategy guidance; the greedy strategy guidance-based graph self-attention network comprises a data preprocessing stage, a super-pixel-based graph self-attention branch stage, a pixel-based multi-attention fusion branch stage, a classification stage and a greedy training strategy stage. And step 2, classifying the hyperspectral image to be measured based on the trained greedy strategy guided graph self-attention network to obtain a classification result. The method is applied to the field of hyperspectral image classification.

The invention relates to the technical field of aquaculture, in particular to aquaculture filtering equipment which mainly comprises a filtering tank and a stirring mechanism, a water drainage opening is formed in the bottom face of the filtering tank in a penetrating mode, a water inlet and a medicine adding opening are formed in the top face of the filtering tank in a penetrating mode, and the stirring mechanism is arranged in the filtering tank and comprises a baffle located in the filtering tank. And the vibrating mechanism is arranged below the barrier plate. A motor drives a rotating shaft, a limiting strip and a rotating plate to rotate, when the rotating plate drives a connecting shaft and a stirring blade to rotate, a roller on the connecting shaft rolls on the inner wall of a fixed frame, and under the action of the roller, the fixed frame, a groove body and a fixed rod, the connecting shaft and the stirring blade revolve, rotate and move up and down in a reciprocating manner at the same time; therefore, contact and reaction between the flocculating agent and the water body are more uniform, the flocculating agent can fully play a role, impurity particles are fully captured or coagulated, and the impurity separation and precipitation effect and the water body filtering quality in the water treatment process are improved.

The invention discloses an SOFC-oriented multistage solid hydrogen storage system and method. The SOFC-oriented multistage solid hydrogen storage system comprises a gaseous hydrogen storage device and an air conveying device, the gaseous hydrogen storage device is connected with the SOFC and the combustion device through the hydrogen heat supplementing device; the air conveying device is connected with an air supply device, the end part of the air supply device is connected with an air heat compensation device through an air heat recovery device, and the air heat compensation device is connected with the SOFC; the hydrogen storage reactors are installed between the gaseous hydrogen storage device and the hydrogen heat supplementing device, the hydrogen storage reactors comprise a TiFe reactor, an Mg2Ni reactor and a TiH2 reactor, and the TiFe reactor, the Mg2Ni reactor and the TiH2 reactor are sequentially communicated. The system provides a safe, high-density and low-cost hydrogen storage function for the SOFC through integration of the SOFC and multi-stage solid hydrogen storage, high-temperature waste heat generated in the SOFC power generation process just provides a driving heat source for solid hydrogen storage and hydrogen desorption, and the defects in the prior art are overcome.

The invention relates to the technical field of aircraft maintenance project management, and discloses an aircraft defect accurate recording system and method, and the system comprises a data collection module, an image processing module, a feature processing module, a defect recognition module and a result output module. The data acquisition module comprises an image acquisition unit and an auxiliary information acquisition unit; the image processing module is used for preprocessing the surface image; the feature processing module is used for extracting target features; the target features are converted into feature vectors; an identification model is arranged in the defect identification module; the recognition model takes the feature vector as input, and takes the defect rate and the defect category as output; and the result output module is used for positioning a defect position and visually displaying and storing a defect analysis result according to the output result of the defect identification module and the corresponding feature vector. According to the method, defect feature analysis dimensions are diversified, defect information can be fully captured, the identification precision is high, and systematic intelligent management of aircraft defects can be realized.

The invention discloses a method for enhancing the super-resolution of a single image through a double-domain learning network, electronic equipment and a storage medium, solves the problem of insufficient reconstruction details and textures in the prior art, and can integrate frequency domain and spatial domain information to improve the reconstruction quality from a low-resolution image to a high-resolution image. The method comprises the following steps: 1, inputting a low-resolution image into a double-domain learning network comprising a plurality of double-domain learning blocks, and generating two super-resolution image outputs through processing of the plurality of double-domain learning blocks: one is frequency domain output, and the other is spatial domain output; step 2, introducing a complex convolution discriminator and cross-modal distillation loss to enhance the performance of the double-domain learning network, specifically, the complex convolution discriminator is used for processing data for converting an image from an HSV color space into a complex form; the cross-modal distillation loss serves as a loss function, and effective fusion of frequency domain and space domain output is guided in the model training process.

The invention relates to an image recognition method for an electric power inspection scene, and the method comprises the steps: obtaining electric power inspection image data, carrying out the data enhancement processing, achieving the data amplification through a generative adversarial network, constructing an electric power inspection image data set through marking, and carrying out the sample balance processing; for a convolutional neural network model trained on a large-scale data set in advance, performing fine tuning by using an electric power inspection image data set; the method comprises the following steps: acquiring a to-be-identified electric power inspection scene image, inputting the image into a convolutional neural network model, extracting multi-scale features of the image in a residual connection mode, obtaining high-dimensional fusion features through cascade fusion or weighted fusion, improving the attention on a region of interest through an attention mechanism, and performing identification by using multiple models. And fusing the recognition results of the plurality of models to obtain a final image recognition result. Compared with the prior art, the method has the advantages of being good in performance, low in false detection rate and omission ratio and the like on power inspection image data with small volume.

The invention provides a method and a device for predicting the remaining service life of a cutter. The method comprises the following steps of: making a plurality of labels for wear values in a cutter wear experiment data set; vibration signal data and cutting force signal data of the tool are collected and preprocessed; randomly dividing the plurality of labels and the preprocessed vibration signal data and cutting force signal data into a training set, a verification set and a test set; extracting multi-domain features of the training set, the verification set and the test set; inputting the multi-domain features into a first training model for training, and outputting nonlinear features; inputting the multi-domain features into a second training model for training, and outputting linear features; fusing the nonlinear feature and the linear feature to obtain a fused feature; and inputting the fusion feature into a third training model, and outputting a predicted value of the remaining service life of the cutter. The method improves the prediction precision of the residual service life of the cutter.

The invention discloses a multi-source generalization method based on a Markov classifier. The method comprises the following steps: acquiring image data of a plurality of source domains; according to the image data of the plurality of source domains, a training data set is obtained, one image in the image data of each source domain serves as a sample image, and the training data comprises the sample image and a category label corresponding to the sample image; an image detection model is constructed, the model comprises a feature extractor and a generation classifier which are connected in sequence, the feature extractor adopts a convolutional neural network, and the generation classifier adopts a Mahalanobis classifier based on Mahalanobis distance; constructing a target detection loss function; training an image detection model by using the training data set based on the target detection loss function; and obtaining a to-be-classified image, inputting the to-be-classified image into the trained image detection model, and obtaining and outputting a category corresponding to the to-be-classified image. According to the method, cross-domain class change can be fully captured, image data of a transition fitting source domain is avoided, and the classification precision of the model on different source domain images is remarkably improved.

The invention belongs to the technical field of brain grain recognition in a brain-computer interface, and discloses an online brain grain recognition method for a cross-session scene, which comprises the following steps: acquiring electroencephalogram data of K users under a session, performing non-overlapping segmentation on each test time according to the length T into brain electron segments, and storing the brain electron segments in a database; performing sliding window processing on the brain electron fragment under S different dynamic window scales to correspondingly generate S groups of data sets with different dynamic window scales; performing offline training on M different random seed models by adopting each group of dynamic window scale data sets to obtain M brain grain recognition models; and in the second session, performing gliding window processing on data which is received online and has the length of T in a corresponding scale, inputting the processed data into the corresponding brain grain recognition model for user identity prediction, averagely integrating prediction results, and performing weighted integration to obtain a user identity prediction result, thereby realizing cross-session online brain grain recognition. According to the method, the accuracy, robustness and recognition rate of cross-session electroencephalogram signal identity recognition can be improved.

The invention provides a tool wear prediction method and device, and the method comprises the steps: obtaining vibration signal data, cutting force signal data, sound emission signal data and tool wear value label data of a tool, constructing a data set, and dividing the data set into a training set, a verification set and a test set; respectively extracting multi-domain features of the training set, the verification set and the test set to obtain a first feature, a second feature and a third feature; inputting the first feature, the second feature and the third feature into a first training model for training, and outputting an intermediate feature; inputting the training set, the verification set and the test set into a second training model for training, and outputting multi-scale features; fusing the intermediate features with the multi-scale features; and inputting the training set of the fused features into a third training model for training, inputting the test set of the fused features into the trained third training model, and outputting a target predicted wear value. According to the method, the prediction precision of tool wear is improved.

This utility model discloses an intelligent integrated drainage-demisting terminal for high-humidity exhaust gas, comprising a drainage-demisting box, and an air inlet and an air outlet respectively disposed on opposite sides of the box; two sets of demisting box bodies are symmetrically arranged inside the drainage-demisting box, one set of demisting box bodies is connected to the air inlet, and the other set of demisting box bodies is connected to the air outlet; several sets of baffles are rotatably installed inside the demisting box bodies; an adjustment mechanism is provided on the outside of the demisting box bodies; the bottom ends of the two sets of demisting box bodies are connected by a semi-circular connecting pipe, and the bottom end of the connecting pipe is connected to a drainage mechanism. This utility model belongs to the field of exhaust gas treatment technology, specifically referring to an intelligent integrated drainage-demisting terminal for high-humidity exhaust gas.

The invention relates to the technical field of nickel-containing dangerous and solid waste recycling, and discloses a complex nickel-containing waste smelting flue gas cleaning and comprehensive utilization method which comprises the steps that smelting flue gas is fed into a dry-method deacidification tower after being subjected to waste heat recovery and shock cooling, coarse-particle calcium hydroxide is sprayed into the dry-method deacidification tower, defluorination is preferentially carried out in a first temperature interval of the dry-method deacidification tower, and defluorination is carried out in a second temperature interval of the dry-method deacidification tower; a defluorination product is collected, and fluorine-containing dust is separated and collected through electrostatic dust collection; the dedusted flue gas enters a semi-dry deacidification tower, and fine-particle calcium hydroxide is sprayed into the semi-dry deacidification tower for deep dechlorination in a lower second temperature interval of the semi-dry deacidification tower; and discharging the purified gas after denitration, washing and ionic liquid desulfurization. The defluorination product is washed and purified to prepare hydrofluoric acid, and the dechlorination slag is dissolved and subjected to a double decomposition reaction to prepare hydrochloric acid. According to the method, a temperature gradient and absorbent particle size gradient coupling process is utilized, so that directional separation and enrichment of gas-phase fluorine and chlorine pollutants are effectively realized, the problem that carnallite is difficult to utilize is solved, and full-process recycling and ultralow emission are realized.

The invention discloses an insurance fraud detection method, a related device, equipment and a storage medium, and the method comprises the steps: carrying out the feature extraction based on to-be-detected insurance data, and obtaining a first scale feature, a second scale feature, and a first fusion feature of the first scale feature and the second scale feature; based on the first fusion feature, predicting to obtain a first weight representation and a second weight representation; obtaining a second fusion feature based on a weighted feature of the first scale feature by the first weight representation and a weighted feature of the second scale feature by the second weight representation; performing classification prediction based on the second fusion feature to obtain a detection result of the to-be-detected insurance data; wherein the detection result at least comprises whether the to-be-detected insurance data is suspected of fraud. According to the scheme, the insurance fraud detection efficiency and precision can be improved, and the insurance fraud detection cost is reduced.

The invention discloses a risk identification method and device, equipment and a storage medium. In the scheme, transaction log information to be identified is acquired; and generating risk prompt information based on the transaction log information and the corresponding business scene. And splicing the transaction log information with the risk prompt information to obtain to-be-identified risk information. And inputting the to-be-identified risk information into a target risk identification model constructed based on a BERT model for risk identification to obtain a risk identification result. According to the scheme, the risk information is enhanced by generating the specific risk prompt information for different business scenes, risk identification is performed on the enhanced information through the risk identification model constructed based on the BERT model, the risk information is fully captured, and the risk identification accuracy is integrally improved.

A method for constructing a large-scale knowledge graph pre-training model and a reasoning platform belongs to the technical field of knowledge graphs of computers, and comprises the following steps: adding prompt information into an entity and relationship sequence, so that a model captures richer information and more accurate entity and relationship representation; a structure adjustment function and a rule adjustment function are provided, the scores of the candidate entities are adjusted, and the knowledge graph completion precision is improved; experiments prove that the method is remarkable in effectiveness and superiority on a biomedical knowledge graph completion task, and the completion of the biomedical knowledge graph can be realized with higher precision.

This invention relates to the field of time series forecasting technology and discloses a stock index forecasting method based on 1DCNN-ResBiLSTM, comprising the following steps: S1, data acquisition and sample construction; S2, data preprocessing; S3, local feature extraction; S4, long-term dependency learning; S5, residual feature fusion: extracting the feature vector of the last time step output by the 1DCNN module in S3, linearly mapping it to the same dimension as the output feature of S4 through a fully connected layer, and then adding it element-wise with the output feature of S4 to achieve residual feature fusion; S6, prediction result output; S7, model training and optimization. By adopting a technical solution that combines the 1DCNN module with time-dimensional sliding convolution and ReLU nonlinear activation, the method accurately captures the local fluctuation features and short-term trend information of stock time series, solving the problems of insufficient characterization of short-term stock price changes and inability to effectively mine local time series patterns.

The embodiment of the invention discloses a point cloud data processing method and device, electronic equipment and a storage medium. The initial features of the point cloud data can be obtained; performing feature extraction processing on the output features of the (i-1) th layer of coding block by adopting the ith layer of coding block to obtain the output features of the ith layer of coding block; performing feature fusion processing on the output feature of the (K-j + 1) th layer of coding block and the output feature of the (j-1) th layer of decoding block by adopting the jth layer of decoding block to obtain the output feature of the jth layer of decoding block; and performing point cloud identification processing based on the output feature of the last layer of decoding block to obtain an identification result of the point cloud data. According to the method, geometric dependency between local and overall shapes of the point cloud data is considered, so that the dependency relationship is effectively captured through layer-by-layer refinement and fusion, features of different scales are fused in a decoding process, and effective transmission of local and overall features is ensured. According to the scheme, the accuracy of feature representation of the point cloud data can be improved.

The invention relates to a field rice leaf blast detection method, device and equipment and a medium. The method comprises the following steps: acquiring a to-be-detected rice leaf image; a convolution attention module is introduced after a backbone feature extraction layer of a Pyrami d-YOLOv8 model, a bottleneck module in cross-stage partial network bottleneck modules is replaced by a pyramid-bottleneck module, a feature map with higher resolution is introduced into a feature fusion layer, and an additional detection head is constructed, so that the improved Pyrami d-YOLOv8 model is constructed. The construction of a rice leaf blast detection model is completed; carrying out disease detection on the to-be-detected rice leaf image by adopting a rice leaf blast detection model trained to a convergence state so as to determine a disease area and a disease grade of leaf blast in the to-be-detected rice leaf image; and spraying the disease area of the leaf blast disease by adopting the bactericide type and dosage corresponding to the disease grade of the leaf blast disease in the to-be-detected rice. The detection precision of the rice leaf blast can be improved, and the rice leaf blast can be prevented and controlled in time.

The invention discloses a multifunctional composite material for synergistically purifying multiple pollutants and a preparation method of the multifunctional composite material, belongs to the field of waste gas purification, and is used for realizing industrial desulfurization, decarburization and denitrification, and the multifunctional composite material is prepared from the following raw material components: 20-40% of aliphatic polyamine or polyamine species; 20-30% of piperazine and derivative species thereof; 10-20% of manganese and vanadium species; and the balance of a natural clay carrier. The preparation method comprises the following steps: firstly, preparing an aqueous solution of TEPA and manganous nitrate, loading the aqueous solution of TEPA and manganous nitrate on a porous carrier, then drying to obtain a solid amino decarburization and denitrification composite material, finally loading synthesized HPP on the solid amino decarburization and denitrification composite material by dropwise adding or spraying, and then drying to obtain the amino multi-pollutant synergistic trapping and regenerating composite material. The step-by-step preparation mode can ensure the quality and stability of each link.

The invention discloses a knowledge graph embedding method and system for intelligent knowledge recommendation. A complex intelligent knowledge recommendation framework is built based on knowledge graph embedding and deep reinforcement learning technologies; on the basis of the built complex knowledge intelligent recommendation framework, a complex knowledge graph model is built in combination with complex knowledge characteristics; a TransEx knowledge graph embedding method is constructed by fusing an attention mechanism and a complex value embedding technology, entity nodes and relationships in a knowledge graph are converted into a vector sequence, and the structure and semantic relationships of the graph are reserved.

The present invention relates to the technical field of gait analysis, and in particular to a multi-head contextual attention gait phase classification method based on a TCN network, which is as follows: construct a model based on a TCN network and a multi-head contextual attention module, input the data collected by the sensor into the TCN network of the model for feature extraction, and obtain features; construct a multi-head contextual attention module, input the features into the module, and obtain fused features; input the fused features into a channel attention module to obtain comprehensive features; calculate the total loss of the model; input the comprehensive features into a linear layer classifier to obtain the final phase classification prediction score, and the category corresponding to the phase classification prediction score with the highest score is the predicted final category. The present invention can more accurately extract the temporal context features of sensor data and improve the accuracy of gait phase classification.