30results about How to "Quick classification" patented technology

The application provides a kind of underwater structure apparent disease identification method and system based on deep learning, wherein, method includes: according to the combination of fusion model and target recognition model, generate the pre-set underwater structure apparent disease identification model;Underwater apparent disease is identified by the pre-set underwater structure apparent disease identification model;The fusion model is built by improved CycleGAN model and multi-scale Retinex algorithm (MSR) network, for converting underwater image into clear and clear image with obvious features;The target recognition model is obtained by YOLOv5 model, for realizing the positioning and classification of underwater structure apparent disease.The present application solves the problems of inaccurate classification and low recognition accuracy of underwater structure apparent disease by the underwater structure apparent disease identification model, caused by factors such as camera imaging blur, insufficient contrast, dispersion and noise in complex water area.

The application provides a die cutting and hot stamping equipment intelligent state monitoring and fault diagnosis system and method, and relates to the field of intelligent monitoring.The method comprises the following steps: collecting working parameters of multiple key parts of the die cutting and hot stamping equipment, constructing a time sequence collection window, slidingly collecting the working parameters, and obtaining characteristic information reflecting the equipment state; based on the characteristic information, constructing an anomaly detection model, performing anomaly detection on the equipment state, obtaining an anomaly score, and judging whether the equipment state is abnormal according to the anomaly score; for the characteristic information judged as abnormal, constructing a fault diagnosis model based on the fault type to which the characteristic information belongs, performing fault diagnosis on the equipment state, and generating a diagnosis result; and generating a comprehensive diagnosis and operation and maintenance decision report according to the diagnosis result.The application realizes comprehensive perception of the internal state of a closed host through multi-sensor collaborative monitoring and dynamic time sequence collection, breaks through the limitations of traditional monitoring, and provides accurate data basis for early fault warning and predictive maintenance.

The invention discloses an uplink service message identification processing method and device, a terminal and a medium. The method comprises the following steps: acquiring core feature information corresponding to an uplink service message; obtaining a target message core data structural body according to the core feature information; obtaining a PDU session identifier corresponding to an uplink service message based on the target message core data structure body; obtaining a QoS flow identifier based on the PDU session identifier and the target message core data structure body; based on the QoS flow identifier, obtaining an uplink service message after scheduling processing; performing query processing on the QoS flow identifier to obtain a target packet data convergence protocol instance; and completing a QoS identification processing flow of the uplink service message based on the target packet data convergence protocol instance. The invention aims to realize rapid QoS identification, classification and processing of the uplink service message, improve the message processing efficiency and accuracy, and reduce the system resource overhead.

The application discloses a multi-modal radio frequency authentication method based on a multi-scale signal representation, which comprises the following steps: firstly, pre-processing the original IQ signal of a target device received to obtain an instantaneous envelope signal; carrying out multi-scale decomposition and denoising processing on the instantaneous envelope signal to obtain a denoised envelope signal; constructing a multi-modal data set according to the original IQ signal and the denoised envelope signal; carrying out feature extraction and fusion on the multi-modal data set through a pre-trained target feature extraction network to obtain a fused feature representation; and carrying out classification on the fused feature representation through a pre-trained target classification network to output a classification result corresponding to the target device. The amplitude, frequency and time-frequency energy distribution information can be complementarily fused by constructing the multi-modal data set; the multi-modal features are extracted and fused through the target feature extraction network, so that the information loss is effectively avoided; and the target classification network is used for rapid classification, thereby reducing the calculation cost and ensuring the accuracy of identification.

The utility model relates to related technical field of screening technology especially, a kind of high throughput screening device of high-yield strain of wolfberry enzyme, including support plate, the upper surface of support plate is provided with screening mechanism;The high throughput screening device of high-yield strain of wolfberry enzyme, by the setting of screening mechanism, start motor when using, and the liquid needing screening is added to inlet pipe, due to the effect of rotation centrifugation, material is evenly distributed between each disc, and the strain with smaller specific gravity moves upwards along disc wall, enters filtration liquid storage interval, the contact area of material and it is increased by the design of multiple groups of disc, so that more material is rapidly classified under the action of reason, shorten separation time, improve product purity.

The application discloses an active intrusion prevention method, relates to the technical field of system network security, and aims to solve the problems of low detection precision and detection lag of existing network abnormal intrusion, and comprises the following steps: obtaining preprocessed network data; performing multi-dimensional traffic data analysis on the network data to obtain the network data labeled with network state credibility measurement results and various sub-dimension measurement labels; calculating the labeled network data through a machine learning classification algorithm to obtain characteristic behavior classification of the network data; and performing suspicious node detection through fine-grained real-time measurement according to the classification results and updating a security strategy. The application further discloses an active intrusion prevention device. The application can quickly analyze network intrusion through multi-dimensional analysis of network data and machine learning.

The application provides a cabin adjustment method, device, equipment, medium and program. It relates to the technical field of cabin adjustment. The method comprises the following steps: acquiring video data and pressure data of passengers in a vehicle within a preset time length; performing feature extraction on each frame of image in the video data to obtain a target seat of a child among the passengers, a head posture, a facial expression and a continuous closed-eye time length; obtaining a pressure sum, a pressure center and a pressure change of the child based on the target seat and the pressure data; performing weighted processing on the head posture, the facial expression and the continuous closed-eye time length, the pressure sum, the pressure center and the pressure change according to a fusion weight to obtain fusion data of the child; taking the fusion data as an input of a state estimation model to obtain a child state corresponding to the child; and adjusting the cabin of the vehicle to a state conforming to the child state according to the child state and a preset cabin adjustment rule. The application achieves the effect of improving the safety and comfort of the child during the ride.

This invention relates to an engine acoustic-vibration multi-information fusion diagnostic device and fault diagnosis method. It removes interference components from the acoustic time-domain data of a faulty engine; through three levels—acoustic-vibration time-domain feature correlation, acoustic order data, and vibration order data—it uses a lookup table method to quickly classify and locate fault features, thereby rapidly determining the possible fault types of the engine; it employs time-domain / frequency-domain multi-layer signal fault information extraction methods to better handle strongly periodic abnormal noise signals and non-steady-state signals, thus facilitating the identification of fault types with non-steady-state noise characteristics. Compared with existing engine on-site abnormal noise diagnostic schemes, this method has good adaptability, and by comparing the features of multi-dimensional information results and performing correlation analysis, it can effectively filter external interference signals during the testing process and locate the source of abnormal noise from a higher-dimensional mechanistic level.

The utility model relates to the technical field of corrugated board processing, in particular to an automatic corrugated board distributing mechanism which comprises a base, a conveying belt is installed and connected in the base, a second material guide plate is fixedly connected to the middle of the right end of the base, and a first material guide plate is fixedly connected to the middle of the front end of the base. The first guide plate and the second guide plate are both of an inclined structure, a guide device is fixedly connected to the upper portion of the front end of the base, a frame is fixedly connected to the left portion of the upper end of the base, an identification camera is installed and connected to the frame, and a control panel is fixedly connected to the upper portion of the front end of the frame. According to the automatic distributing mechanism for the corrugated boards, the pushing device is arranged, the electric push rod is started to work, the electric push rod pushes the push plate to move forwards, the first motor installed on the push plate drives the baffle to rotate, and after the baffle rotates, the baffle and the push plate are distributed in an L shape; therefore, the paperboard can be prevented from continuously moving on the conveying belt during pushing.

The application provides a non-target detection device and system based on a nanopore and a detection method thereof, the non-target detection device and system based on a nanopore are used for non-target detection of a complex matrix through a nanopore, feature extraction is performed on a nanopore detection graph, and sample classification is realized through model construction, the detection method does not make a prior definition on a to-be-detected target, and various molecular information existing in serum is acquired in a net catching manner, non-target capture and overall state analysis of various molecules in serum are realized, and a new way of rapid and portable disease screening is provided.The application breaks through the limitation of traditional single marker detection by combining the nanopore technology and machine learning, and is expected to promote the development of precision medicine in a more efficient and comprehensive direction.

The application discloses a breast CEM image lesion recognition system and equipment based on deep learning, and relates to the field of medical image processing. By acquiring a low-energy image and a subtraction image of an original CEM image and performing pretreatment, lesions in a breast region of the image are labeled, and two ResNet50 networks are used to extract preliminary extraction features of the low-energy image and the subtraction image respectively. Double-mode auxiliary feature fusion is performed to obtain a multi-scale pyramid feature layer. A detection head with an attention mechanism is connected after each pyramid feature layer to finally position and classify breast lesions. The application realizes rapid detection, positioning and classification of breast lesions based on a multi-modal CEM image based on a deep learning method, and improves breast lesion recognition efficiency and precision.

The application provides a vehicle detection data classification method and system, a computer and a readable storage medium. The method comprises the following steps: preprocessing vehicle detection data to generate input text; inputting an input vector into an ERNIE model to convert the input vector into a first word vector, and performing sequence feature processing on the input text to generate a second word vector; performing splicing processing on the first word vector and the second word vector to generate a word vector matrix, and inputting the word vector matrix into a DPCNN model; optimizing the DPCNN model through an equal-length convolution function, performing maximum pooling processing on the equal-length convolution function through the optimized DPCNN model to generate a maximum feature value; and outputting a predicted classification label corresponding to the input text according to the maximum feature value. Through the above method, the classification of vehicle detection data can be quickly completed, thereby greatly shortening the time consumed for classifying vehicle detection data.

The utility model discloses a kind of neodymium iron boron alloy finished product sorting devices, it is related to neodymium iron boron alloy processing equipment technical field, including base, classification frame is hinged on the base, two parallel distribution baffle plates are fixedly connected on the classification frame, two baffle plates divide the inside of classification frame into first sorting channel, second sorting channel and third sorting channel, first sorting channel is located between second sorting channel and third sorting channel;The inside of the first sorting channel is provided with first sorting assembly, and the first sorting assembly comprises first guide plate and first elastic strip, and the first guide plate and the first elastic strip are both inclinedly arranged.The neodymium iron boron alloy finished product sorting device is provided with classification frame, first sorting assembly, second sorting assembly and other structures, realizes quick classification, reduces working intensity, and the sliding of alloy finished product can be speed-reducing buffered, reduce the damage caused by collision, improve the use efficiency of neodymium iron boron alloy finished product sorting device.

The invention discloses a defect detection method and system based on multi-mode light-induced thermal wave imaging, and the method comprises the steps: carrying out the multi-mode light excitation of a honeycomb sandwich structure through employing a pulse light source and a modulation light source as excitation modes, and generating a multi-mode infrared thermal wave signal; infrared thermal wave signals are continuously collected based on an infrared thermal imager, and timestamp alignment is carried out; based on the time domain features and the frequency domain features of the multi-modal infrared thermal wave signals, signal-level fusion and image-level fusion are carried out respectively, and a signal-level fusion image and an image-level fusion image are obtained; and inputting the original image, the signal-level fusion image and the image-level fusion image into an improved YOLO-World model, and realizing automatic identification and classification of defects through multi-scale feature extraction and a RepVL-PAN module. According to the method, the multi-mode optical excitation technology and the YOLO-World deep learning model are combined, efficient recognition of complex defect types in the honeycomb sandwich structure is achieved, detailed information about defects is provided from different angles, and the detection precision is remarkably improved.

The present application relates to a kind of biological species homology analysis systems based on protein / gene sequence data, belong to biological species homology analysis technical field, solve the problem of high cost and low efficiency of detection method in prior art.The system of the present application includes: data preprocessing module, for the obtained protein / gene original data is carried out data cleaning, obtain protein / gene sequence data;Data analysis module includes: protein sequence analysis unit, for according to the frequency of each amino acid in protein sequence, position and its physical and chemical properties, generate protein sequence numerical representation vector;Gene sequence analysis unit, for obtaining the optimal matching sequence in two groups of gene sequences;Species homology judging module, for according to protein sequence numerical representation vector / optimal matching sequence, the distance of each two groups of protein / gene sequence data is calculated, and according to the size of distance, the analysis of protein / gene sequence homology nearness.

This application discloses an information processing method, apparatus, electronic device, and storage medium. The method includes: acquiring first interaction data of target information, the first interaction data being used to characterize the interaction frequency of the target information; determining second interaction data of the target information based on an information set; the second interaction data being used to characterize the interaction features of the target information relative to the information set; determining target interaction data of the target information based on the first interaction data and the second interaction data; and determining whether the target information can be classified into the information set based on the target interaction data.

This invention relates to a hyperspectral band selection method based on spectral-spatial hypergraph regularized sparse self-representation. To address the insufficient utilization of band adjacency relationships in traditional sparse self-representation models, this method first integrates spectral similarity and spatial proximity to design a comprehensive index for evaluating band correlation. It explores and evaluates adjacency relationships between bands from multiple perspectives, constructing a hypergraph to generate a more complete inter-spectral local structure. Second, based on the aforementioned local band constraints, it supplements this with global band characteristics to jointly guide the optimization process of the sparse self-representation band selection model, obtaining the optimal low-dimensional representation of high-dimensional data and achieving band selection. This method overcomes the limitations of using a single similarity metric for decision-making and fully explores the multi-faceted adjacency relationships between bands, effectively reducing band redundancy, improving the quality of band subsets, and contributing to improved analytical accuracy in subsequent tasks.

The invention discloses a seismic exploration data unsupervised similarity clustering method, electronic equipment, a storage medium and a device. The method comprises the following steps: constructing a seismic exploration data unsupervised similarity clustering model based on a self-organizing mapping network; constructing a training sample and a test sample based on actual seismic exploration data; training the clustering model based on the training sample; testing the trained clustering model based on a test sample; and carrying out seismic exploration data unsupervised similarity clustering based on the tested clustering model. According to the method, the seismic exploration data unsupervised similarity clustering model is constructed through the self-organizing mapping network to carry out seismic exploration data unsupervised similarity clustering, data are classified and identified according to geophysical rules and standards, effective and rapid classification of seismic data can be realized, the seismic data characterization capability is improved, and the seismic exploration data classification efficiency is improved. And a basis is provided for accurate processing and explanation of seismic exploration data.

This application discloses a noise identification method, medium, equipment, and product for thermal power plants. The thermal power plant includes multiple pieces of equipment. The method involves acquiring raw sound signals and historical acoustic feature data; digitizing each raw sound signal to obtain multiple acoustic features; determining a dynamic threshold range based on the historical acoustic feature data; identifying anomalous features among the multiple acoustic features based on the dynamic threshold range; determining a target label for each anomalous feature based on a predefined set of noise labels and the associated attributes corresponding to each anomalous feature; constructing an association graph based on the index information and target labels corresponding to the multiple anomalous features; and performing posterior probability analysis on the nodes in the association graph to obtain the noise source region of the thermal power plant. This application embodiment achieves anomaly detection through multi-dimensional features, dynamic thresholds, and label matching, and constructs an association graph for probability analysis, enabling precise location of noise sources.

This invention relates to a target depth classification method based on the difference in arrival angle of direct waves. In a known deep-sea area, there is a horizontal linear array, and a moving sound source moves in a straight line through the array towards its forward trajectory. The minimum value of the curve showing the change in arrival angle of the direct waves over time is compared with the angle boundary value for the same motion trajectory characteristic in the known sea area to obtain the depth information of the moving sound source; that is, the depth of the moving sound source is either less than or greater than the depth boundary value. This invention can quickly calculate the depth boundary value and angle boundary value based on simulation. Furthermore, it can pre-calculate the parameter information corresponding to different target motion trajectories in the known sea area, facilitating rapid target depth classification, greatly improving computational efficiency, and is easy to implement in engineering practice.

The application discloses a kind of based on self-adapting graph extension line and intelligent shunt's malicious domain name detection method and system, belong to network security field.Method includes: to seed domain name carries out DNS resolution and IP reverse search;To excessive IP constructs multidimensional feature vector and is filtered by three-stage cascade classifier noise to generate candidate domain name;For each candidate domain name, maintain discovery path label PathTag and adaptively select fingerprint primary key calculation fingerprint according to it, soft exclude same family split;Through confidence function, automatically determine and skip LLM to threat intelligence confirmed malicious domain name, and the rest are sent to LLM according to fingerprint group batch to judge;To clustering, execute time anomaly self-correction and intra-cluster consistency check.The application reduces extension line noise and determination cost by three-stage gating, PathTag driven fingerprint clustering and confidence shunt mechanism, improves clustering accuracy.

The application discloses an intelligent film image evaluation system and method adopting cloud detection technology, relates to the technical field of medical image processing, and comprises a cloud server, which is communicatively connected with the following modules, wherein: an image acquisition and uploading module is used for acquiring original film image data from various film image devices and uploading the acquired original film image data to the cloud server for preprocessing. The application realizes intelligent analysis of the film image through the cloud detection technology, gives a quality score in combination with a preset quality standard, and judges whether the image meets diagnostic requirements, so that evaluation differences caused by human factors are reduced. Meanwhile, according to extracted interference features, a pattern classification algorithm is used to perform quality analysis on the film image meeting the diagnostic requirements after quality evaluation, a quality classification coefficient is calculated, and the image is divided into different quality grades, so that accurate classification of the image quality is realized, and a reference basis is provided for medical diagnosis.

The present application belongs to the technical field of artificial intelligence text processing, and relates to a text recognition method, a classification method and a device based on a large model. After recognizing the text through OCR, the present application uses the semantic understanding and reasoning ability of the large model Transformer architecture to accurately extract context key information, obtains the data required by the user through Prompt Engineering, forms a vector cluster through a text summary, compares it with a vector library through inner product comparison, outputs a classification name, classifies the input text, and thus realizes conversion, recognition, classification and storage. The present application combines various technologies such as OCR recognition, text information extraction and file attribute classification, and realizes efficient and accurate processing of documents. The PPOCRv4 workflow and Macro-o1 large model are used to structurally extract key information of files, improving the accuracy and depth of key information extraction of files. The Jina-V3-Embedding large model and the vector library are used for inner product comparison, realizing fast and accurate classification of text attributes.

This invention discloses a real-time visual detection method, system, and storage medium for substation faults, innovatively constructing a hierarchical early exit mechanism. Through ultra-lightweight early exit branches and dynamic lightweight detection branches, efficient dynamic allocation of computing power is achieved. Deformable convolution and multiple CBAM attention mechanisms are introduced into the complete backbone network, enhancing the network's adaptive feature focusing ability on irregular fault edges. Motion-guided spatiotemporal regularization and temporal deformation alignment modules are introduced. This design overcomes the limitation of traditional algorithms relying solely on single-frame static images by caching the core features of the previous frame and utilizing optical flow motion fields for spatial deformation and alignment, injecting spatiotemporal constraints consistent with the physical evolution of faults into the feature fusion stage. Regarding the improved loss function, a Gaussian penalty term and a smoothing focusing factor are superimposed. Furthermore, a spatiotemporal consistency regularization loss is introduced to construct a novel joint loss function.

The utility model discloses a multi-station type key strike life testing machine, which belongs to the technical field of electronic component detection equipment and comprises a base, supporting legs are arranged at corners of the lower end of the base, supporting rods are arranged at corners of the upper end of the base, a top plate is arranged at the upper ends of the supporting rods, and three groups of electric push rods are arranged at the upper end of the top plate. The output end of the electric push rod is provided with a test box, the interior of the test box is provided with a center rod in a penetrating manner, and the lower end of the center rod is provided with a pressing plate. Through the arrangement of the keyboard positioning mechanism, keyboards of different sizes can be accurately fixed, and the stability of the keyboards in the test process is ensured; the transmission assembly drives the nut block to move along the lead screw, so that the position of the U-shaped positioning plate is adjusted, the keyboard is firmly clamped in cooperation with the rubber pad, the testing accuracy and reliability are improved, and meanwhile testing errors caused by unstable fixing are reduced.

The application discloses a three-dimensional CUS classification prediction method and system based on RFCN, and the method comprises the following steps: S1, carbon utilization and storage CUS suitability classification model data analysis; S2, constructing a cyclic full convolutional neural network RFCN and optimizing parameters; S3, using the optimized cyclic full convolutional neural network RFCN to perform visual training and prediction on the carbon utilization and storage CUS suitability classification process, and completing prediction result analysis and reliability evaluation. The application can realize fast, quantitative and three-dimensional visual prediction of carbon utilization and storage CUS suitability classification.

This application provides a method and system for intelligent identification of power equipment based on deep learning, relating to the technical field of equipment identification. The method includes: firstly, acquiring real-time and historical images of the power equipment; stitching the historical images together to form a stitched image; randomly cropping a sub-image of the same size as the historical image from the stitched image; adding training labels to the historical image and the sub-image to obtain training data; then constructing a classification model using a deep learning algorithm; training the classification model with the training data; and finally inputting the real-time image into the trained classification model to output the classification result. This application can improve the accuracy of power equipment identification results even with limited training samples.

This invention discloses a detection circuit for hidden defects in aluminum electrolytic capacitors, comprising an AC-AC high-power power frequency converter, a triangular wave signal generation module, an output module, a time detection and judgment system, and an LCD display system. The AC-AC high-power power frequency converter provides a stable and compatible DC power supply, ensuring the stability of subsequent modules. The triangular wave signal generation module can generate triangular waves with adjustable amplitude and frequency to meet the detection requirements of aluminum electrolytic capacitors of different specifications. The output module can accurately capture the charging and discharging characteristics of the capacitor and output a reliable detection signal through voltage division, filtering, and comparison circuits. By combining the time detection and judgment system with the LCD display system, the signal can be quickly analyzed, the results displayed, and automatically classified, improving detection efficiency and accuracy and solving the problem that traditional methods are difficult to identify internal hidden defects.

The invention is suitable for the technical field of power systems, and provides a power system fault classification method, device and equipment and a program product, and the method comprises the steps: obtaining electrical quantity time sequence data collected by power system monitoring equipment; performing multi-level signal decomposition on the electrical quantity time sequence data to obtain a plurality of signal components; extracting statistical features of the plurality of signal components to construct feature vectors to be classified; and inputting the feature vectors to be classified into a breadth feature mapping classification model trained in advance, and outputting a fault classification result, so that complex fault types can be effectively distinguished, and the classification accuracy and fineness are improved.

The invention discloses an overall flood classification method and system based on flood composition and encountering conditions, and the method comprises the steps: S1, collecting and sorting the overall flood data of a research section, including the flood process of a research section session and the inflow process of an upstream main inflow section at a corresponding time; s2, calculating the flood volume ratio of inflow water of upstream main inflow water sections of different overall floods to downstream floods in different time periods; s3, calculating the average proportion of incoming water of the upstream main incoming water section of each overall flood to the flood volume of the downstream flood in different time periods; s4, primarily classifying the floods according to the average proportion of inflow of upstream main inflow sections of different overall floods to the flood volume of downstream floods in different time periods; s5, flood encounter evaluation indexes are defined, secondary classification is conducted on the overall flood according to the evaluation indexes, and refined classification of the overall flood is achieved; according to the method, the overall flood is classified, the upstream and downstream overall flood can be quantitatively and finely classified, and a basic support is provided for researches such as overall design flood and similar flood recognition.