41 results about "Fuzzy recognition" patented technology

The invention discloses a light guide plate defect detection method and system based on a neural network, and particularly relates to the technical field of machine vision detection, and the method comprises the following steps: aiming at the problem of image instability of a light guide plate in a dynamic transmission or rotation process, continuously collecting an image sequence and extracting time domain features; and performing interference judgment in combination with the inter-frame consistency prediction coefficient and a first threshold to realize accurate identification of the abnormal image frame. For an abnormal image frame, further correcting the recognition credibility of the abnormal image frame by adopting a confidence adjustment and fusion mode, and meanwhile, introducing a frequency domain transformation and image enhancement strategy to compensate detail loss caused by motion blur; according to the method, inter-frame consistency analysis, confidence fusion regulation and control and frequency domain fuzzy recognition and compensation mechanisms are introduced, abnormal judgment and image quality restoration of the light guide plate image in the dynamic scene are realized, the recognition accuracy and stability of the neural network model on the defect type, position and confidence are improved, and the false detection and omission ratio is effectively reduced.

The invention discloses a chip design and manufacturing optimization system and method based on edge reinforcement learning, and relates to the field of data analysis, and the system comprises a scene recognition module, a strategy selection module, a decision correction module and a cloud updating module. A fuzzy recognition and alarm mechanism is introduced to enhance the adaptability; inter-process linkage optimization is realized by means of a double-index strategy library and an influence gradient transmission mechanism, and the process collaboration is improved; through dynamic uploading of the state tetrad and cloud oscillation identification, the system can automatically enter cold-quiet period locking under strategy abnormity and energy consumption fluctuation, and abnormal diffusion is effectively inhibited. Meanwhile, a lightweight edge deployment and reinforcement learning mechanism enables the system to have high response, low communication and self-learning capabilities, and the system is suitable for a large-scale complex chip manufacturing environment.

The invention discloses a knowledge graph data intelligent question and answer method and system based on voice activation, and the method comprises the steps: obtaining a voice instruction inputted by a user side, and converting the voice instruction into a natural language text; semantic fuzzy recognition is conducted on the natural language text, and then whether semantic fuzzy exists in the natural language text or not is determined; if yes, generating a voice prompt problem by using a standard entity in the knowledge graph and a corresponding known attribute, obtaining a correction text fed back by a user based on the voice prompt problem, performing named entity recognition on the correction text, and extracting a main entity and a target keyword; if not, named entity recognition is carried out on the natural language text, and a main entity and a target keyword are extracted; calculating the semantic similarity between the known attribute of the standard entity corresponding to the main entity in the knowledge graph and the target keyword; and determining a user intention based on the semantic similarity, generating a query statement for executing a corresponding query operation in the knowledge graph database, and obtaining a corresponding query result.

The invention discloses a camouflage target segmentation method based on multi-scale cross-spectral feature interaction, and belongs to the technical field of computer vision. The implementation method comprises the following steps: 1, constructing a camouflage target detection network; 2, obtaining visible light and infrared dual-mode features for the network subjected to feature extraction; 3, performing multi-layer feature fusion on the visible light image features and the infrared image features; therefore, bimodal complementary information extraction is realized; 4, constructing an edge sensing network for performing detail processing on the edge of the image so as to obtain an edge prediction result; 5, training the camouflage target detection network to obtain training parameters of the camouflage target detection network; optimizing training parameters of the camouflage target detection network; 6, inputting the image pair into the training optimization camouflage target detection network to obtain a predicted camouflage target detection result; compared with the prior art, the accuracy of boundary fuzzy recognition and appearance similarity recognition of camouflage target detection is improved.

The invention discloses an image detection and recognition method based on data analysis, and relates to the technical field of image detection, and the method comprises the steps: determining an object recognition result corresponding to a to-be-detected weight ratio as a final recognition result if the to-be-detected weight ratio with the same number as a preliminary object recognition result is greater than or equal to a weight determination threshold value through statistics; if the weight ratio to be measured is smaller than a weight judgment threshold value and only different results which can be directly recognized exist in different object recognition results, image object recognition after error adjustment is carried out on a preset brightness simulation adjustment parameter set, or when a fuzzy recognition result exists in a difference result to be measured, image object recognition is carried out on the preset brightness simulation adjustment parameter set. If not, segmenting a to-be-recognized intercepted image from the to-be-detected image, and performing image object recognition after different degrees of error adjustment on brightness simulation of an edge object and a center object in the to-be-recognized intercepted image so as to judge a final recognition result. The image detection and identification method based on data analysis provided by the invention can improve the efficiency of image detection work.

The invention discloses a tower crane bolt axial force intelligent ultrasonic detection method and system based on multi-modal data fusion. Firstly, a multi-stage influence factor system is constructed based on a fuzzy hierarchical comprehensive evaluation theory, and a high-risk region identification model is established; and calculating a regional risk value based on the model, and determining a monitoring point distribution scheme. A heterogeneous network composed of a fiber grating strain sensor and an ultrasonic sensor is arranged in a selected area, and strain and ultrasonic data are synchronously collected. And inputting the multi-modal data into a deep belief network for feature fusion, and realizing health state dynamic evaluation through a fuzzy recognition model. And based on an evaluation result, performing fatigue life prediction by adopting an improved rain flow counting method and an accumulative damage theory. And finally, data fusion display and intelligent early warning are realized through the remote monitoring platform. According to the invention, full-chain intelligent health management from risk identification, real-time monitoring to life prediction is realized.

The invention relates to the field of transformers, in particular to a transformer state comprehensive judgment method and system based on a digital gas relay and a medium. The invention provides a weight adaptive fuzzy recognition comprehensive state evaluation method combining an industry score deduction criterion and a wooden barrel effect on the basis of data such as temperature, pressure, gas volume and gas production rate of a digital gas relay and online monitoring data such as oil chromatography, iron core grounding current, partial discharge and vibration. And early warning of major faults is realized. The action reliability of the digital gas relay is improved, and the alarming and tripping accuracy of the gas relay is improved.

The invention relates to the field of character and image recognition, in particular to a multi-mode intelligent dictation and correction system and method based on dynamic calibration, and the system comprises an acquisition module for obtaining a writing test image based on voice feedback and carrying out character recognition, a feature splitting module for constructing a feature observation group and labeling a structural unit, the layout analysis module selects a recombination structure by sliding a window frame and sets a fuzzy identification tag based on the prior probability of spatial distribution features, and the identification module compares contour features through region mapping to determine difference types. And the calibration module realizes context restoration and dynamic calibration by removing fuzzy blocks, performing mapping comparison with a sample database and replacing contour features. According to the method, the recognition accuracy of complex handwriting such as scribbling and stroke adhesion is improved, and on this basis, the local image content is calibrated through a replacement mechanism based on context restoration, so that the accuracy and reliability of the correction process are improved.

The present invention discloses a chip design and manufacturing optimization system and method based on edge reinforcement learning, which involves the field of data analysis and includes a scene recognition module, a strategy selection module, a decision correction module, and a cloud update module. The present invention realizes robust recognition of the equipment operating status through a state-aware feature channel and a confidence scoring mechanism, and introduces a fuzzy recognition and alarm mechanism to enhance adaptability; with the help of a dual-index strategy library and an impact gradient transmission mechanism, it realizes inter-process linkage optimization and improves process synergy; through the dynamic upload of state quadruple and cloud-based oscillation identification, the system can automatically enter a cool-down period lock under policy anomalies and energy consumption fluctuations, effectively suppressing the spread of anomalies. At the same time, lightweight edge deployment and reinforcement learning mechanisms enable it to have high response, low communication and self-learning capabilities, suitable for large-scale complex chip manufacturing environments.

The application discloses a hybrid vehicle energy management device, which comprises a driving style recognition module, a typical road condition recognition module and an energy management adaptive module; the driving style recognition module is used for acquiring driving style reference data in a preset time period, inputting the driving style reference data into a driving style fuzzy recognition model, and acquiring a current driving style output by the driving style fuzzy recognition model; the typical road condition recognition module is used for acquiring driving characteristic parameters of a vehicle in a preset time period, calculating driving characteristic similarities corresponding to various typical road conditions according to the driving characteristic parameters and various corresponding driving characteristic parameters of the various typical road conditions, and taking a typical road condition corresponding to a maximum driving characteristic similarity as a current road condition; and the energy management adaptive module is used for acquiring the current driving style and the current road condition, formulating an electric quantity balance point according to the current driving style and the current road condition, and selecting a proper vehicle driving mode according to a relationship between a current residual electric quantity and the electric quantity balance point.

This invention relates to a DNN fuzzy recognition model training method. The method involves extracting features from input sample objects using an object-oriented DNN fuzzy recognition model, mapping these features to obtain the object's classification probability distribution, and mapping these classification probabilities to membership degrees of fuzzy sets in a fuzzy sample model library. The model is trained by calculating a loss function value based on the classification probability distribution using sample annotations. After one generation of training, the model is adjusted by calculating a loss function value based on the fuzzy set proximity using sample annotations. If the result of either loss function calculation does not meet the preset loss value requirement, the samples in the fuzzy sample model library are randomly shuffled and reordered, and the calculation is iteratively repeated until the results of both loss function calculations meet the preset loss value requirement. This invention also relates to a method and apparatus for identifying hydrological and meteorological elements using a model trained using the above method. This method can simulate crew members automatically identifying hydrological and meteorological elements, improving the accuracy of identification and significantly reducing the workload of crew members.

The invention discloses a picture recognition management system based on video monitoring, and relates to the technical field of image management. The image processing unit is used for collecting image features of the basic data and carrying out gray processing on the image features, the image recognition management system based on video monitoring can enhance the image features by adjusting the foreground and background of a gray image, detail optimization can be carried out when character feature recognition blurring exists in a live broadcast image, and the image recognition efficiency is improved. The quality of the live broadcast picture is enhanced, and continuous adjustment is performed on the enhanced image features, so that the situation that the attention of the live broadcast room is not on a product and public opinions are caused due to sudden picture adjustment can be effectively avoided, a good live broadcast picture effect can be ensured, the attention of the live broadcast room can be reduced, and the user experience is improved. And the adjustment sequence of the plurality of image pictures is analyzed under the delimited rule, so that the pictures can be managed in a unified manner, and the value maximization of live broadcast picture optimization is ensured.

The invention relates to the related field of document data engines, in particular to a document data engine method and system, which realizes high-robustness field extraction capability by combining a template engine with an AI model, and adapts to multiple types of unstructured documents; field splitting / combination is intelligent, and manual intervention is reduced. Automatic mapping of standard fields and alignment of platform fields are achieved, the adaptation relation between an industry field standard library and multi-system fields is constructed, and the fields can be directly pushed to multiple service systems from original samples; field data quality and consistency guarantee capability are improved, and output fields are more accurate due to field-level difference comparison and format restoration functions; the AI completion and fuzzy recognition mechanism solves the problem that fields are not covered or uncertain in the traditional recognition technology. The system has four characteristics of standardization, rule driving, model assistance and dynamic updating, and the manual operation intensity and the business risk are remarkably reduced.

This application provides a digital grading method and apparatus for paper-based assignments. The grading method includes: performing adaptive partitioning on an assignment image containing student information to obtain a question area and an answer area; the answer area includes an objective filling area and a subjective writing area, wherein the subjective writing area includes a handwritten text area and a handwritten formula area; based on the content characteristics of the target area, calling a detection and recognition model that matches the content characteristics to obtain the recognition result of each answer area; for fuzzy recognition results of the subjective writing area, introducing the writing style characteristics of the current student information to correct the fuzzy recognition results, and updating the recognition results of the subjective writing area based on the correction results; comparing the final recognition results of each answer area with the assignment answers to generate and output the grading results. The grading method and apparatus of this application form a fully automated process from image input to grading result output, requiring no manual intervention and achieving effective automatic grading.

The invention relates to an elevator car scheduling method, device and system based on a spatial mode. The elevator car scheduling method based on the spatial pattern comprises the steps that target detection is conducted on an image when a car stops so as to extract object information, and the occupied capacity is estimated based on an object mobility rule; whether a historical scene is matched or not is judged in combination with the case library, if yes, a corresponding target function is directly selected, and if not, traffic flow prediction is conducted through object information of the lift car, lift car direction information and historical data, fuzzy recognition is conducted through the fuzzy rule library, and a current traffic mode is determined; then, an objective function is selected according to a use mode or a traffic mode of a historical case, optimization solution is carried out in combination with the current occupied capacity, and an optimal lift car scheduling scheme is generated. According to the elevator car scheduling method based on the spatial pattern, the problems of inaccurate scheduling and low efficiency caused by neglecting the capacity factor are effectively solved.

The invention relates to the technical field of self-healing and repairing assistance, and discloses a self-healing and repairing assistance system based on a four-start system, and the system comprises a four-dimensional data collection module which collects the initial four-dimensional data of a user; the four-dimensional data analysis module is used for carrying out independent scoring and correlation analysis on the initial four-dimensional data and determining the short plate dimension of the user; the self-healing repair scheme generation module is used for generating a self-healing repair scheme according to the short plate dimension and the intervention strategy library; the scheme feedback adjustment module is used for acquiring feedback data of the user for executing the self-healing repair scheme and dynamically adjusting the self-healing repair scheme; and the repair effect evaluation module is used for obtaining the final-period four-dimensional data of the user, comparing the final-period four-dimensional data with the initial four-dimensional data, and determining the repair evaluation effect of the self-healing repair scheme. The technical bottlenecks of multi-dimensional cutting, fuzzy recognition, strategy stiffness, fault assessment and the like in traditional growth intervention are solved, and the functions of active prevention, accurate repair and system collaboration are achieved.

Methods and systems are described herein for identifying the location and nature of any blur within one or more images received as a user communication and generating an appropriate correction. The system utilizes a first machine learning model, which is trained to identify blurred components of inputted images and determine whether the blurred components are located in portions of the inputted images comprising textual information. The system may apply a corrective action selected by the first machine learning model, which may comprise stitching blurred images together to a sharp product image and / or some other method appropriate for rectifying images received.

The invention discloses a hydraulic structure safety state evaluation method based on multi-level fuzzy recognition, and relates to the technical field of hydraulic engineering safety monitoring, and the method comprises the following steps: pre-constructing a multi-level evaluation index system which comprises a target layer, a criterion layer and an index layer; by introducing trapezoidal and triangular membership functions, the limitation that a traditional accurate mathematical method cannot describe the continuous safety level is broken through, and the evaluation result is made to better conform to the gradual change rule of underwater engineering damage development. Meanwhile, scattered local data is converted into a unified global safety level through fuzzy comprehensive operation, the data utilization rate and evaluation efficiency are greatly improved, a whole-process scientific support from risk identification, level judgment to maintenance scheme selection is provided for engineering managers, the probability of occurrence of safety accidents is remarkably reduced, and the engineering management efficiency is improved. Reliable technical guarantee is provided for safety management of the whole life cycle of the water conservancy project.

The application discloses a neurosurgery operation scene intelligent understanding method, which is realized through a trained neurosurgery operation scene intelligent understanding network; the network comprises a time coding layer and a space coding layer, and the output end of the space coding layer is connected with at least one fuzzy attention module; the time feature output by the time coding layer and the fuzzy space feature output by the fuzzy attention module sequentially pass through the cross-modal fusion operation of a decoding layer and the dimension reduction processing of a feedforward neural network, and output an operation stage recognition prediction result; the fuzzy attention module evaluates and quantifies the reliable degree of the space visual feature at each position in the feature sequence output by the space coding layer, dynamically adjusts the attention weight and gives the space visual feature to form a fuzzy space feature; the method effectively improves the recall rate and Jaccard coefficient while maintaining high accuracy and high precision, and proves that the method has the effect of significantly optimizing the fuzzy recognition and prediction accuracy.

This invention relates to the field of speech recognition technology, and discloses a speech recognition method, apparatus, device, storage medium, and computer program product. The method includes: constructing a hash mapping table; acquiring the speech to be recognized and recognizing it using a preset speech recognition model to obtain a conventional recognition result; performing precise and fuzzy recognition on the speech to be recognized based on the hash mapping table to obtain precise and fuzzy recognition results; and calculating the target speech recognition result based on the conventional, precise, and fuzzy recognition results. This invention solves the technical problem of poor speech recognition performance in specialized fields or specific industries by calculating the target speech recognition result using the conventional recognition result obtained from the preset speech recognition model and the precise and fuzzy recognition results obtained from the hash mapping table.

The invention relates to the technical field of voice recognition, in particular to an LED multi-dimensional information issuing method based on voice intention recognition, which comprises the following steps of: decoding an input voice signal to generate a decoded word graph, positioning a preset LED multi-dimensional information semantic slot position in the decoded word graph and extracting similar competition paths; and calculating to obtain a high-confidence ambiguity candidate set. When voice signals are processed, similar candidate word combinations can be extracted from an ambiguous recognition path by constructing a decoding word graph and positioning LED multi-dimensional information semantic slots, posterior probability differences are further calculated to form an ambiguous candidate set, and a clear distinguishing basis is introduced in a scene with similar recognition precision but semantic conflicts, so that the recognition accuracy of the voice signals is improved, and the recognition accuracy of the voice signals is improved. And fuzzy recognition caused by path divergence in the word graph is relieved. The clarified questions embedded with the competition words are synchronously presented in a voice and LED dual-channel form, so that the user feedback is more targeted, and the ambiguity clarification efficiency is improved.

This invention discloses a gas channeling early warning method for tight oil reservoirs, belonging to the field of intelligent early warning technology. The method includes acquiring static data, dynamic production data, and geological parameters of the target reservoir; identifying dominant channels and their distribution patterns in the horizontal and vertical directions based on the static data; calculating the gas channeling sensitivity index (GSI) based on the dynamic production data and geological parameters; and triggering a gas channeling early warning when the GSI exceeds a preset threshold. This invention establishes a comprehensive risk awareness foundation by integrating static geological, dynamic production, and engineering parameters. It innovatively adopts a two-level early warning logic combining "static fuzzy recognition evaluation" and "real-time early warning of the dynamic GSI index," achieving a seamless transition from potential risk screening to real-time anomaly capture. This enables precise location of high-risk layers at the fracturing cluster level, prediction of the probability of gas channeling in the next 7-15 days, and visual diagnosis of channeling directions, providing direct and quantitative decision-making basis for plugging agent selection and well location optimization.

This invention provides a short text clustering and fuzzy recognition algorithm for large-scale online subgraph sampling, comprising the following steps: Step S1, extraction and preprocessing of training samples; Step S2, construction of the neural network; Step S3, overall clustering prediction; Step S4, fuzzy sample recognition; Step S5, retraining of the neural network. This invention combines short text clustering with a large language model, which not only improves clustering accuracy but also enables the handling of clustering tasks with different themes and classification requirements, significantly reducing the manual cost of data annotation. Furthermore, this invention can annotate fuzzy samples for classification, using K-nearest neighbors combined with minimum spanning trees to assist subgraph sampling in the selection scheme. This utilizes sparse structure to reduce computational costs and exposes the fluctuations of boundary samples through spectral clustering, providing a more comprehensive perspective for fuzzy sample selection and improving the accuracy and interpretability of the clustering results.

The invention discloses a medical consumable abnormal data type-2 fuzzy screening method based on cross-domain collaboration and a heterogeneous map, and belongs to the technical field of medical information processing. According to the method, professional extraction is carried out on medical consumable data features through a medical consumable attribute feature cleaning and extracting system based on a language processing model. Introducing an abnormal collaborative reasoning model based on first-class transfer learning and deep domain confrontation, and carrying out preliminary study and judgment on the distribution of data features; and in combination with a built heterogeneous semantic knowledge graph information fusion framework based on a knowledge graph and a graph neural network, carrying out deep anomaly detection and correction attempt on a product generating primary anomaly. And after the repair is completed, performing deep discrimination on the corrected abnormal product through an identification and analysis model based on type-2 fuzziness. The method is suitable for multi-dimensional information fusion, knowledge enhancement modeling, multi-section uncertainty discrimination and high-precision and extensible medical consumable data anomaly intelligent screening application requirements in a medical consumable treatment scene.

The invention relates to the technical field of remote sensing image processing, in particular to a river surface feature recognition method based on multi-modal remote sensing data fusion. The method comprises the following steps: collecting a visible light image and laser point cloud data of a river channel region, generating a digital orthoimage map for the visible light image, and constructing a digital surface model after performing noise reduction processing on the point cloud data; normalizing the elevation value of the digital surface model, performing spatial registration with the digital orthoimage, and constructing four-channel fusion data containing three channels of red, green and blue and an elevation channel; and calculating a mean value and adding Gaussian noise disturbance to generate an initial weight of a fourth channel, and constructing an improved network adaptive to four-channel input through channel dimension splicing. The invention provides a four-channel adaptive DeepLabV3 + improved network, seamless adaptation from three channels to four channels is realized through a pre-training weight migration mechanism, and the problem of fuzzy recognition in a complex boundary region is solved.

The application relates to the technical field of data processing systems or methods specially applicable to supervision or prediction purposes, in particular to a fuzzy recognition reservoir flood control safety chain type discrimination system and method, which comprises a storage module, an upstream water quantity estimation module, a downstream water quantity difference value module and a reservoir redundancy analysis module, the upstream water quantity estimation module is configured with an estimation model, the downstream water quantity difference value module is configured with a calculation model, and the reservoir redundancy analysis module is configured with an analysis model. Through the upstream water quantity estimation module, the downstream water quantity difference value module and the reservoir redundancy analysis module, reservoir flood control safety early warning can be realized based on the change of upstream water quantity, the existing technology needs to be combined with multiple data information for estimation, meanwhile, the application can realize simple reservoir flood control early warning in chain type by relying on upstream water quantity data.

The present invention belongs to the field of artificial intelligence technology related to formation recognition, and specifically relates to a multi-agent formation recognition method based on interpolation and fuzzy recognition. The present invention includes a multi-agent formation entering a radar recognition area, the radar scanning and obtaining the coordinates of each agent in the recognition area to obtain a scattered point set; obtaining four vertices from the scattered point set; using a multi-point interpolation method, randomly adding interpolation points between the vertices and adding them to the scattered point set; constructing a formation based on the lines connecting the vertices; using a fuzzy recognition method to identify the constructed formation; multiple iterations, based on the last recognition result, predicting the next membership of each formation through a fitting function, and taking the formation corresponding to the maximum membership as the final recognition result. The present invention uses fuzzy recognition and uses membership functions as a measure of samples and templates, which can better reflect the overall characteristics of the pattern and has a strong ability to eliminate interference and noise in the sample.

The present invention discloses a centrifuge liquid flow detection system based on fuzzy recognition and classification technology, which relates to the technical field of centrifuge flow detection, and includes: collecting an image set, performing preprocessing and marking processing to obtain a segmented flow velocity; obtaining characteristic parameters; obtaining an optimized fuzzy model; obtaining the characteristic parameters of a real-time liquid image to perform flow detection on the real-time liquid. The present invention is used to solve the problem that in the existing methods for liquid flow detection, when the liquid surface is irregular or there is liquid fluctuation, the accuracy and stability of the data received by the sensor will be affected, resulting in a difference between the collected liquid flow and the actual liquid flow, thereby affecting the operation of the centrifuge equipment.

The invention relates to an OCR (Optical Character Recognition) fuzzy recognition method under medical kit binding belt detection and shielding, which relates to the technical field of computer processing, and comprises the following steps: S01, an OCR decoding camera collects continuous frames of gray level images when a medical kit arrives at a packaging station so as to obtain an image data set; s02, preprocessing the grayscale image to extract pixel features of a packing belt and pixel features of a medicine chest, and obtaining coordinates of the pixel features of the packing belt on the pixel features of the medicine chest through a predetermined algorithm; s03, the specific position of the packing belt on the medicine box is obtained through relative position coordinate conversion; and S04, the grayscale image is extracted again through OCR identification, and a label pattern shielded by the packing belt is obtained. According to the method, OCR identification and ribbon identification can be carried out at the same time, when the packing belt shields key information, fuzzy identification is adopted to match a database for reasoning information content, and therefore errors caused by follow-up action implementation of a packing belt cutting machine are reduced.

The invention relates to the technical field of commodity data management, in particular to a commodity sorting method based on a type-2 fuzzy set and online data. Obtaining online data and offline data of a commodity, and performing data processing and structured integration to obtain commodity associated data; constructing a type-2 fuzzy recognition model to recognize the commodity associated data, and converting the input data into type-2 fuzzy data defined by a membership function; constructing a logic fuzzy rule base, and performing fuzzy logic operation on the type-2 fuzzy data to obtain commodity fuzzy performance data; performing order reduction calculation on the commodity fuzzy performance data to obtain a commodity first-order fuzzy set; performing defuzzification calculation on the commodity first-order fuzzy set to obtain a commodity recommendation score; and sorting the commodities according to the commodity recommendation scores. According to the invention, sorting guidance is carried out on the commodities through commodity recommendation scores.