19results about How to "Remove redundant information" patented technology

The invention discloses a multi-modal medical data fusion and analysis platform, and particularly relates to the technical field of medical assistance, and the platform comprises a data collection and preprocessing module which is used for integrating, cleaning and standardizing multi-source data, and is used for efficiently fusing and safely storing the multi-modal data; the data fusion and storage module supports rapid retrieval and management; and the data analysis and modeling module is used for carrying out feature extraction and pattern recognition by utilizing a machine learning algorithm and providing deep prediction analysis. The system integrates, cleans and standardizes medical data from multiple sources through the data acquisition and preprocessing module, ensures data quality and consistency, then efficiently fuses data of different modalities into a single data set through the data fusion and storage module, supports rapid retrieval and management, and finally utilizes the data analysis and modeling module to analyze and model the medical data. And deep prediction and risk assessment of diseases are provided through feature extraction, model training and prediction analysis.

The invention particularly relates to a document abstracting method based on domain knowledge and a multi-granularity graph network, which comprises the following steps: S1, constructing a knowledge coding module, carrying out domain knowledge graph transfer learning, extracting triple structure features in combination with a capsule network, and generating a text entity coding module; s2, constructing a knowledge fusion module, converting a document text into a semantic vector, and adaptively fusing the semantic vector with external knowledge; s3, constructing a word-level semantic graph, and aggregating neighborhood information through a semantic graph attention mechanism to obtain word embedding representation containing context; s4, constructing a sentence-level semantic graph, performing self-attention calculation on the word-level semantic vector in the S3 to obtain semantic node representation of the sentence-level semantic vector, and performing feature aggregation on sentence-level nodes to obtain global semantic information of the document; and S5, constructing an output layer, and predicting according to the document feature vector in the S4 to obtain a final extraction type abstract result. The method has the advantages that redundant information in the document is removed, and ambiguity of text semantics is eliminated.

The invention relates to the technical field of data processing systems for supervision or prediction, in particular to a road and bridge deck sudden collapse early warning system and method based on the Internet of Things, and the system comprises a deformation monitoring system which is used for collecting the deformation data of a road and bridge deck; the data processing system is used for processing the deformation data based on the Internet of Things so as to obtain effective data meeting a preset storage condition; and the early warning management center is used for inputting the effective data into a pre-constructed road and bridge surface collapse risk prediction model to obtain a risk assessment result and a trend prediction result of the road and bridge surface so as to carry out early warning on sudden collapse of the road and bridge surface. Therefore, the problems of low data precision, poor credibility, early warning information delay, increased safety risk and the like caused by the fact that an early warning system only depends on a single technology, is easily influenced by environmental factors and cannot comprehensively capture various abnormal signals which may predict collapse in related technologies are solved.

The invention discloses an industrial process monitoring method based on a lightweight deep principal component analysis-auto-encoder model, and the method comprises the steps: collecting the related process variable data of an industrial process on line, carrying out the preprocessing of the collected data, inputting the preprocessed data into a trained lightweight deep principal component analysis-auto-encoder model, and carrying out the recognition of the related process variable data. The T2 statistical magnitude and the SPE statistical magnitude are obtained and compared with corresponding control limits, and a monitoring result is output; the lightweight deep principal component analysis-auto-encoder model uses a PCA dimension reduction module for dimension reduction to obtain the data score, inputs the score into the auto-encoder, and outputs the reconstruction data. According to the method, the calculation efficiency of the model is improved, compared with a traditional deep network model, the calculation speed is obviously improved, compared with a single multivariate statistical model, the strong feature extraction capacity of the deep hierarchical cascade model structure is incomparable to that of the single multivariate statistical model, and the problem of complex industrial process monitoring is effectively solved.

The application discloses a variable working condition cross-component causal coupling modeling and fault diagnosis method and device, and relates to the technical field of rotating machinery fault diagnosis. The method comprises the following steps: collecting and preprocessing multi-component signals under variable rotating speed working conditions of a rotating machinery; based on multivariate Granger causal modeling, prior information and dynamic disturbance information are fused to construct a dynamic causal diagram; a double-view structure of a main view of the dynamic causal diagram and an auxiliary view of a correlation diagram is constructed, multi-scale coupling features are extracted through a multi-order spectrum filter and an attention mechanism, a symmetric InfoNCE contrast learning loss and a supervised classification loss are combined to optimize and train a model, and fault classification is realized. The device is correspondingly provided with signal acquisition, dynamic causal diagram construction, feature fusion, model training and fault prediction modules. The application improves the accuracy of fault feature representation and the robustness of diagnosis under variable working conditions, is suitable for complex industrial scenes with multi-component coupling and frequent rotating speed switching, and has strong engineering practicability.

The present application relates to the technical field of hydroelectric generating set bolt monitoring, and particularly relates to a hydroelectric generating set bolt state dynamic monitoring method and system. The method comprises the following steps: performing resonance frequency prediction on the bolt, and performing adaptive resonance excitation to obtain a bolt resonance characteristic spectrum; performing bolt acoustic signal collection according to the bolt resonance characteristic spectrum to obtain a multi-channel original acoustic waveform set; performing working condition noise preprocessing on the multi-channel original acoustic waveform set to obtain an acoustic propagation time-domain-frequency domain matrix; performing multi-dimensional stress state mapping analysis according to the acoustic propagation time-domain-frequency domain matrix, and performing bolt loosening state recognition to obtain a bolt structure stress distribution atlas; and performing multi-source heterogeneous data fusion and edge computing according to the bolt structure stress distribution atlas to obtain a connection reliability index. The present application realizes high-precision, real-time and online monitoring of the bolt state through acoustic characteristic change analysis under a complex stress state.

The application provides a video-language understanding method and system, relates to the technical field of video understanding, and obtains a video and text to be understood; based on a trained understanding model, the input video and text are processed to generate a final understanding result; wherein the understanding model comprises a video encoder and a text query component, the video encoder comprises a clustering component and a timing component; the image blocks of all video frames in the video are filtered by using the clustering component to obtain video frame embeddings after eliminating redundant information; the timing dependency between the video frame embeddings is reconstructed by using the timing component to obtain a video feature matrix; based on the video feature matrix, the embedding of the text is performed by using the text query component to obtain the final understanding result; the application comprehensively considers three challenges of information redundancy, time dependency and scene complexity, proposes an understanding model composed of three key components, each component is aimed at a specific challenge, and the video-language understanding ability of an intelligent agent is significantly improved.

The invention provides an RAG system answer generation method, device and equipment based on a large language model, and relates to the field of artificial intelligence, and the method comprises the steps: receiving a query request input by a user; obtaining a target document block according to the query request, and further generating a target prompt word; according to the target prompt word, adopting a preset large language model to generate an answer text corresponding to the query request; wherein the target document block comprises a first document block; the obtaining method of the first document block comprises the steps that keyword retrieval is conducted on a preset knowledge base according to keywords in a query request, and a target knowledge document with the keywords is obtained; and according to the query request and the target knowledge document, generating an abstract text corresponding to the target knowledge document as a first document block. The answer generation accuracy of the RAG system can be improved.

The invention discloses a rapid remote sensing image change detection method based on superpixels, and the method comprises the steps: obtaining remote sensing images of the same region in different periods, dividing the remote sensing images into a T1 period and a T2 period, and carrying out the preprocessing of the remote sensing images in the two periods; overlapping and fusing spectral features of the two periods of images to form a new data block so as to obtain super-pixel blocks with the same scale in different periods; performing principal component analysis on the obtained superpixel blocks, and extracting representative features in each superpixel block; and carrying out binary clustering analysis on the extracted features by using a k multi-mean algorithm, and dividing the extracted features into changed image blocks and unchanged image blocks to realize change detection processing. Compared with the prior art, the rapid remote sensing image change detection method based on the superpixels has the advantages that the spatial information of the image is reserved, the calculation complexity is reduced, and the change detection accuracy is improved.

The present application provides a method, device, electronic device and storage medium for generating recommendation reasons, which relates to the field of artificial intelligence technology, wherein the method includes: in response to a query request from a user side, obtaining a target vector and a target label corresponding to the query information from the user side; recalling first data from a first database according to the target vector, and recalling second data from a second database according to the target label, wherein the first database includes vectors corresponding to the doctor's non-business type data, and the second database includes the doctor's business type label data; merging the first data and the second data to obtain the target text content; and generating a recommendation reason associated with the query information and conforming to a predetermined format from the target text content based on a preset Prompt instruction. The implementation of the technical solution provided by the present application effectively improves the matching degree of the recommendation reason and improves the user's consultation experience.

The present invention provides an indoor heading estimation method, device and medium, wherein the method includes grouping IMU data to obtain local features, integrating them with preset additional features, inputting the local features into a deep learning model to obtain correlation to update the additional features, and finally reducing the dimension of the integrated features to obtain the heading estimation result. The method can carefully mine the local features of the data and capture specific information in different regions; update the additional features through correlation to achieve global feature extraction and enrich feature representation; the dimensionality reduction process can reduce the computational complexity, reduce the training and inference time, remove redundant information and noise, and improve the generalization and robustness of the model. Compared with the existing technology, it can better adapt to the complexity of the indoor environment and improve the accuracy and reliability of heading estimation.

This invention discloses a data-driven modeling method and system for air compressors, comprising: acquiring operating data of an air compressor under various operating conditions; preprocessing the operating data to obtain cleaned operating data; dividing the cleaned operating data into time windows, performing multi-order polynomial regression fitting within each time window to generate an initial fitting parameter set corresponding to each time window; constructing a fitting parameter matrix based on all initial fitting parameter sets, performing matrix decomposition operations, and extracting principal component features of the fitting parameters; establishing an air compressor performance degradation trend model based on the principal component features to describe the performance decay law of the air compressor under different operating durations; and dynamically modeling based on the performance degradation trend model and combined with real-time acquired operating data to generate a real-time performance mapping model. This achieves automatic extraction of performance degradation features from historical data and dynamic compensation modeling.

The present invention discloses an image encoding and decoding method with multi-resolution output, the method comprising: S1: extracting multi-scale features of an image and constructing a multi-scale feature library; S2: based on the multi-scale feature library, training an extreme learning machine model and a support vector machine model to predict the target existence probability and the target size; S3: when the predicted target existence probability is greater than the target existence probability threshold, selecting a multi-scale feature matching the predicted target size in the multi-scale feature library, encoding, and obtaining the encoding feature; S4: performing multi-scale expansion convolution on the encoding feature to obtain a decoded image; S5: based on the decoded image, identifying the target, and correcting the encoding and decoding system based on the identification result. The present invention selects multi-scale features of appropriate resolution for encoding according to the target size category, and corrects the encoding and decoding system according to the decoded image identification result, which is conducive to adapting to scene requirements and improving encoding and decoding efficiency and quality.

This invention provides a method and system for predicting multimodal drug interaction events based on compact representation learning. The method includes: acquiring multimodal features of each drug in a drug pair to be predicted, wherein the multimodal features include at least two of biological features, molecular structure features, and knowledge graph features; performing intramodal compact representation learning on each modal feature of each drug to eliminate redundant information within a single modality and generate compact sub-representations for each modality; using the compact sub-representations corresponding to the molecular structure features as anchors, performing cross-modal alignment of the compact sub-representations corresponding to the biological features and knowledge graph features through mutual information minimization constraints to eliminate intermodal redundant information; constructing a multimodal fusion representation for each drug based on the compact sub-representations of each modality; concatenating the multimodal fusion representations of the two drugs in the drug pair and inputting them into a classifier to output the prediction result of the interaction event type of the drug pair.

The invention relates to a neural network specific misclassification repair method based on disturbance feature recovery, belongs to the technical field of misclassification repair, and solves the problems that in the prior art, the specific category distinguishing capability is insufficient, and the classification accuracy of an original model for misclassification repair is reduced. The method comprises the following specific steps: performing initial classification on a to-be-classified image data set by using a neural network model obtained by training an original image data set; screening out image sample data classified as a category P from the to-be-classified image data set as a doubt set; based on the doubt set and the trained feature mapping model, performing disturbance feature repair on the potential features extracted by the initial feature extractor; using the trained verifier to verify whether the repaired features correspond to a specific misclassification # imgabs0 #, if yes, repairing the image sample data misclassified as a category P in the suspected set, and obtaining a repaired classification result; according to the method, the accuracy of the original model is not greatly reduced, and meanwhile, the repair of specific misclassification is completed.

The present invention provides a cutting quality assessment method and system for an automobile caliper production line, which relates to the field of automobile manufacturing technology, including collecting image data of the cutting surface of the automobile caliper at different use stages; extracting the first texture feature and the second texture feature of the cutting surface of the automobile caliper; constructing and training an autoencoder model; obtaining a comprehensive feature representation through feature fusion processing; processing the comprehensive feature representation through feature selection based on mutual information to obtain the feature that contributes most to the judgment of the cutting surface state, and obtaining the classification result of the cutting surface state through long short-term memory convolutional neural network model training and optimization processing. The present invention realizes rapid and accurate assessment of the cutting surface of the automobile caliper, real-time monitoring and assessment of the cutting surface condition, provides a scientific basis for the production and quality control of the automobile caliper, and ensures the safety and reliability of the automobile.

The invention belongs to the technical field of computer vision, and provides a lightweight aerial photography target detection method based on mixed cavity regulation and control perception and partial decoupling aggregation. The invention provides a partial channel decoupling aggregation network structure aiming at the problems that tiny target features in aerial images are easy to lose and complex background interferences exist. The method comprises the following steps: firstly, designing a mixed cavity regulation and control sensing unit (MS-DGPU), introducing deep convolutional layers with different expansion rates into a regulation and control branch, constructing a multi-scale receptive field to capture context information of a fine-grained space, and generating a dynamic weighting factor through nonlinear activation; and secondly, constructing a partial decoupling aggregation module (PDAM) based on the unit, decoupling a feature channel into a spatial representation and semantic retention part by utilizing partial convolution (PConv), only performing regulation and control weighting on the spatial representation part, and then realizing feature multiplexing through residual connection. The module is embedded into a neck network of a target detection model, so that redundancy calculation is effectively reduced, and the detection precision and reasoning speed of the model on a multi-scale aerial photography target are remarkably improved.

This invention discloses a method, apparatus, and storage medium for event content verification based on a large model, belonging to the field of data processing technology. The method includes: identifying events in target text using a large model, rewriting the events in the target text according to target rewriting rules to obtain reference events; the target text is standard data that has been reviewed; storing the reference events in an event knowledge base; retrieving target reference events associated with the event to be verified from the event knowledge base based on the event to be verified; and comparing the event to be verified and the target reference events using the large model to obtain a comparison result characterizing whether the event to be verified is correct. This application utilizes the powerful semantic understanding capability of the large model to accurately identify events in standard data and rewrite the events to form structurally unified and semantically clear reference events, thereby constructing a high-quality event knowledge base and improving the accuracy of event content verification.

The present invention belongs to the field of image coding technology, and specifically discloses a method for image precoding based on an attention mechanism. The present invention proposes a scenario in which image transmission is required after image instance segmentation is performed, and the image is firstly subjected to an attention mechanism operation based on the result obtained by image segmentation, and then the image is encoded, which can not only ensure the image quality and details of the ROI area, but also remove the redundant information of the non-ROI area of ​​the image, and the present invention can also flexibly configure the weight coefficient, so as to flexibly adjust between detail retention and redundancy removal, so as to balance the compression ratio and quality.