226 results about "Thresholding algorithm" patented technology

The invention relates to a railway traction substation state monitoring method and system, equipment and a medium. The monitoring method comprises the following steps: acquiring real-time monitoring data of the equipment in a railway traction substation; performing data preprocessing on the real-time monitoring data to obtain a preprocessed monitoring data set, performing protocol identification, and converting heterogeneous data in the monitoring data set into structured data according to a preset protocol template library; based on the structured data, time-frequency domain characteristic parameters of the equipment are extracted, and a multi-dimensional characteristic matrix is constructed; inputting the multi-dimensional feature matrix into a pre-trained hybrid diagnosis model, and generating an equipment health degree score and a fault probability value; according to the health degree score and the fault probability value, generating an early warning instruction in combination with a dynamic threshold algorithm; and generating a priority maintenance strategy through a maintenance strategy optimization model based on the early warning instruction and the equipment maintenance resource constraint condition. According to the invention, accurate perception and intelligent decision making of the equipment state are realized in a multi-source heterogeneous data environment.

The invention discloses a monorail crane inspection robot intelligent test method based on data analysis, and relates to the technical field of intelligent detection, and the method comprises the following steps: synchronously collecting track images, point cloud and attitude data, carrying out time alignment and preprocessing, and outputting a standardized data packet; detecting an inspection target by using a YOLO detection network, and outputting a multi-scale feature vector in combination with a point cloud feature hierarchy extraction network and a time sequence convolutional network; predicting a fault development trend in combination with an improved A-star algorithm and a long short-term memory network, optimizing an inspection path through reinforcement learning, and outputting a maintenance decision scheme; the maintenance decision scheme is converted into a control instruction, the robot is driven to execute an inspection task and feed back the operation state in real time, incremental learning and point cloud reconstruction are combined, and a visual diagnosis report is output. According to the method, the dynamic threshold algorithm is adopted for self-adaptive analysis, and the key geometric indexes are calculated in combination with the cross-modal attention mechanism, so that the recognition capability of structural anomalies is improved.

The invention discloses a GIS partial discharge intelligent diagnosis system and method based on one-dimensional ultrahigh frequency signal analysis, and relates to the technical field of power electrical equipment intelligent monitoring, and the system comprises a signal collection and preprocessing module which is used for collecting ultrahigh frequency signals of GIS equipment and obtaining preprocessed signal data through a dynamic threshold algorithm; the discharge initial judgment module is used for performing multi-dimensional sequential judgment to eliminate interference discharge data so as to obtain effective discharge signal data; the feature extraction module is used for performing time domain kurtosis and pulse width analysis, frequency domain energy distribution analysis and time-frequency domain wavelet entropy calculation based on the multi-dimensional features of GIS partial discharge, and generating an optimized feature subset; and the type identification module is used for identifying the partial discharge type by using the integrated learning model to obtain a diagnosis result. According to the invention, the problem of unstable recognition accuracy caused by insufficient signal preprocessing, single feature representation and single classification algorithm in the prior art is solved.

The invention discloses an energy consumption management platform based on big data, and belongs to the technical field of energy consumption management, and the platform specifically comprises a building level energy consumption tracking module which collects energy consumption data of different levels of a building in real time, binds energy consumption equipment with a building space, maps sub-item energy consumption equipment to corresponding level nodes according to physical positions, and transmits the energy consumption data to the building level energy consumption tracking module; constructing an energy consumption flow direction topological graph according to a total-score table relation; the visual energy consumption display module is used for dynamically generating a multi-level energy consumption distribution thermodynamic diagram according to the topological graph, setting different color gradients according to the energy consumption loss rate, calculating a loss rate threshold value by means of a dynamic threshold value algorithm in combination with historical and environmental data, and marking abnormal nodes if the loss rate threshold value is exceeded; and the energy consumption repair positioning module is used for generating a diagnosis report containing fault equipment positioning and energy consumption waste reasons based on the historical monitoring data and the user operation log of the corresponding energy consumption equipment after detecting the abnormal node.

The invention discloses a power transmission line icing galloping risk early warning system and early warning method, and belongs to the technical field of icing galloping risk early warning, and the system comprises the following modules: a historical data processing module reconstructs historical monitoring data by using a long and short term memory auto-encoder, recognizes abnormal data through reconstruction errors, and sends the abnormal data to an early warning module; removing outliers based on a 3sigma-dynamic threshold algorithm, and then marking a space-time credibility weight; the terrain compensation module constructs a micro-terrain feature vector, calculates the similarity through a Siamese network, and compensates low-confidence data; the real-time data processing module obtains the icing thickness according to the monitoring data of the current time point and the previous time point; the data acquisition and analysis module fuses historical and real-time data and analyzes a galloping state; and the risk early warning module takes the historical data with the weight and the galloping state information as input, outputs a risk value through the prediction model, and triggers early warning. According to the system, through cooperative work of all the modules, the icing galloping risk is accurately warned in real time, and the safety and reliability of power grid operation are improved.

The invention discloses a CPLD-driven IGBT short-circuit current dynamic monitoring method and system, and relates to the technical field of power electronics, and the method comprises the specific steps of S100 real-time data acquisition, S200 dynamic threshold calculation, S300 short-circuit state judgment, S400 rapid protection execution, and S500 fault data recording and uploading. Accurate and prospective judgment of the short circuit state of the IGBT is achieved, two stress parameters including the direct-current bus voltage and the junction temperature serve as independent variables, when the bus voltage rises, the voltage stress borne by the IGBT in the turn-off period is increased, meanwhile, the current peak at the short circuit moment is more remarkable, early warning is conducted in advance by dynamically lowering the threshold value, and therefore the short circuit state of the IGBT is accurately judged. According to the mechanism, the monitoring system does not simply respond to the occurred overcurrent, but can pre-judge the safety boundary of the device under the current working condition, so that intervention is implemented before the current really reaches a dangerous peak value, and a protection blind area under a severe working condition is effectively avoided.

The invention relates to the technical field of battery management, and discloses a battery management system with an active monitoring function, which comprises a processor, a battery state monitoring module, an active equalization control module, a performance evaluation module and a fault early warning module. The battery state monitoring module collects various parameters of a battery at different stages, and generates an abnormal signal by using a dynamic threshold algorithm. The active equalization control module adjusts an equalization strategy based on a fuzzy logic algorithm, and a fuzzy rule base can be updated. And the performance evaluation module tracks and evaluates an equalization process, analyzes equalization efficiency and quality, and optimizes an equalization strategy accordingly. The fault early warning module displays abnormal signals in a grading mode and carries out early warning, and the processor is further connected with the aging feature extraction module and the thermal runaway prediction module to achieve battery life prediction and thermal runaway early warning. According to the invention, comprehensive monitoring, efficient equalization control, accurate performance evaluation and reliable fault early warning of the battery are realized, the safety of the battery is improved, and the service life of the battery is prolonged.

The invention provides a numerical control machining defect real-time detection method and system based on AI image recognition and a medium, and belongs to the technical field of intelligent manufacturing and machine vision crossing. The method comprises the following steps: collecting part surface image data according to a preset frequency, and separating a part from a background through noise reduction, contrast enhancement and a dynamic threshold algorithm of a data preprocessing module; inputting the processed image into a multi-modal feature fused AI detection model, and extracting and fusing multi-modal features to judge the type and position of a defect; and processing the image by using a parallel computing architecture, feeding back a detection result to a numerical control processing control system in real time, and visually displaying the detection result on a monitoring interface. According to the method, real-time and accurate detection and feedback control of numerical control machining defects are realized, and the machining quality and efficiency are effectively improved.

The invention provides a coastal water level monitoring method based on a dynamic threshold algorithm, a medium and a system, and belongs to the technical field of coastal water level monitoring. The coastal water level monitoring method based on the dynamic threshold algorithm comprises the steps that data are collected through a water level sensor array, and a water level basic matrix and a floating matrix are constructed; a dynamic warning value is calculated by applying a tidal kinetic equation and a hydrometeorological coupling equation set, efficient resource scheduling is realized by utilizing a calculation jump matrix, self-learning optimization and water level prediction are performed based on a hydrological space-time network model, and water level monitoring and prediction data are output. Early warning of different levels is triggered according to the fact that the water level reaches the minimum range or the maximum range of the dynamic warning value, warning information is pushed to the disaster management platform through the edge computing gateway, a comprehensive warning report is finally generated for emergency decision making, and intelligent monitoring and accurate early warning of the coastal water level are achieved.

A concrete damage real-time positioning method based on laser induction and dynamic wave velocity correction comprises the following steps: constructing a damage induction platform integrated with a high-energy pulse laser, combining with a three-dimensional mobile platform to realize accurate positioning of a laser focus, and arranging a phase demodulation type optical fiber acoustic emission sensor array to construct a full-coverage monitoring network; synchronous acquisition of high-frequency acoustic emission signals is realized by adopting a multi-channel coupling phase demodulation system, and noise suppression is completed in combination with a self-adaptive wavelet threshold algorithm; a dynamic wave velocity attenuation model is established based on space-time reference information, a propagation path is optimized through a Basic Theta * path planning algorithm, a wave velocity field is reconstructed by applying an ART algebraic iteration algorithm, and a thermodynamic diagram is generated; and finally, constructing a convolution-graph neural joint network model, and realizing accurate monitoring and positioning analysis of the concrete structure damage through a damage probability distribution generation and multi-scale verification module by using a multi-modal data fusion and transfer learning technology.

The invention relates to a landslide intelligent early-warning method based on an aging damage mechanism, and the method comprises the steps: building a landslide numerical model, and carrying out the assignment of the mechanical parameters of a rock-soil body; setting a numerical calculation initial condition and a boundary condition for elastic calculation, and calculating and generating a landslide initial ground stress model; calling the dynamic constitutive model, and performing numerical calculation through a creep command; embedding an improved tangent angle threshold algorithm; the system collects monitoring data in real time and automatically completes damage accumulation quantification based on the aging damage dynamic constitutive model; embedding the aging damage early warning model identification system into a computer system; and calculating and identifying the aging damage early warning model, and outputting an early warning result. According to the invention, through the dynamic damage model, the damage evolution rule of the rock-soil body under the long-term natural environment effect is accurately reflected, and the timeliness and accuracy of the early warning system are ensured; the corrected tangent angle lower limit value is adopted, so that the reliability of landslide early warning is improved, the landslide can be found and early warned earlier, and disaster prevention and reduction of the landslide are facilitated.

The invention relates to the technical field of industrial equipment monitoring, and discloses an industrial equipment state monitoring and fault early warning method and system based on big data, and the system comprises a data collection and fusion module, an intelligent diagnosis module and an early warning execution module. Through innovative design of a spatio-temporal feature fusion engine, the problem of spatio-temporal mismatch of vibration, temperature, current and other signals is solved, dynamic characteristics of millisecond-level vibration signals are captured through a long-short-term memory network, a spatial topological relation of equipment is modeled through a graph convolutional network, cross-scale alignment of multi-physical-quantity data is achieved, the limitation of physical significance isolation in a traditional method is broken through, and the method has a good application prospect. The composite fault recognition capability is improved, the diagnostic value of unstructured data such as sound emission is released, a full-life-cycle self-adaptive mechanism is constructed through a dynamic threshold algorithm, the working condition fluctuation tolerance is intelligently adjusted through a sensitivity coefficient, an early warning model continuously optimized along with the operation time is formed, the stiffness defect of a static threshold is changed, and the early warning effect is improved. And an accurate intelligent early warning system is established.

The invention relates to the technical field of state monitoring, in particular to an energy-saving air compressor running state monitoring system and method, and the system comprises a pressure slope analysis module, a deviation trend judgment module, an error fitting acquisition module, an error trend analysis module and an alarm triggering module. According to the method, a pressure range is calculated by dividing time slices, a slope ratio is generated, a dynamic threshold algorithm is combined to successively compare and capture an offset trend, frequency and current waveform data are synchronously acquired to match a timestamp to calculate a difference value, an error vector is generated in a standardized manner, and a periodic mean value is extracted to verify an incremental relationship to identify progressive anomaly. The cross validation slope ratio and error vector positioning wear or coordination abnormity, the dynamic threshold value is adjusted along with the mean value coefficient at the initial period to reduce fluctuation interference, the timestamp matching enhances the composite fault detection capability, and the range division operation quantifies the matching degree to provide an energy efficiency optimization basis. And a standardized error vector and mean value incremental mechanism improves the prediction precision and reduces false alarm and missing alarm.

The invention discloses a signal detrending method based on sliding window statistics and dynamic wave crest correction. The method comprises the step of realizing accurate extraction of signal trend and fluctuation characteristics through a double-window dynamic adjustment mechanism of a mean value window and a standard deviation window. A detection threshold value is dynamically adjusted by analyzing the change trend of the historical wave crest amplitude, and the real-time tracking capability of signal fluctuation is ensured; meanwhile, different parameter configurations are preset for different application scenes, the problem of missing detection or misjudgment generated when the amplitude of a fixed threshold algorithm changes suddenly is solved, and the wave crest recognition accuracy is improved. Based on a dynamic threshold algorithm, wave peak start and end points are automatically identified, a linear interpolation technology is adopted to calculate slope of each interval, fitting parameters are dynamically adjusted through residual analysis, and finally low-frequency trend terms and high-frequency characteristic components are effectively separated through difference operation of original signals and fitting signals. The problem of fitting deviation caused by non-uniform sampling can be effectively solved.

The invention relates to the technical field of hyperspectral image processing, and provides a deep-expanded spatial spectrum sparse memory hyperspectral sharpening fidelity method and system, and the method comprises the steps: carrying out the priori knowledge coding of a high-resolution hyperspectral image through a regularization technology, solving a target function of a hyperspectral panchromatic sharpening task through a semi-quadratic splitting method, and carrying out the processing of a hyperspectral image. A residual module is adopted to simulate a degradation operator, spectral fidelity and spatial fidelity sub-problems are solved through gradient descent iteration, and a sparse prior sub-problem is solved based on an iterative shrinkage threshold algorithm; the high-fidelity and high-resolution hyperspectral images are corrected through the spatial spectrum prior features, and the multi-stage reconstructed high-resolution hyperspectral images are fused through the cross-stage memory fusion network to obtain the high-resolution hyperspectral images. According to the method, the reconstruction precision of the image is improved, the calculation complexity is reduced, the extraction of the spatial spectrum combined prior features of the original image is realized, and lossless information transmission is realized.

The invention discloses a low-dose 3D spiral CT reconstruction method and system based on a context cold diffusion model, and the method comprises the following steps: converting a linear inverse problem of CT imaging into an optimization problem, and adding a regularization term; based on a near-end gradient descent framework, solving the optimization problem by adopting an iterative shrinkage threshold algorithm, replacing a near-end operator in the near-end gradient descent method with a pre-trained two-dimensional cold diffusion model, and replacing a regularization item with the pre-trained cold diffusion model; when the cold diffusion model is pre-trained, three-dimensional context information formed by a current slice and two adjacent slices is used as the input of the cold diffusion model; taking a low-dose two-dimensional CT image as input, and performing low-dose 3D spiral CT image reconstruction by adopting the trained network; by combining the physical imaging model and the deep learning prior, the reconstruction quality and efficiency of the low-dose CT are remarkably improved, the noise and artifacts in the low-dose CT image are remarkably reduced, and the quality of the low-dose CT image is remarkably improved.

The invention provides a supply chain cost anomaly AI positioning system based on multi-mode depth tracing, and relates to the technical field of cost anomaly tracing. According to the supply chain cost anomaly AI positioning system based on multi-modal depth tracing, a three-layer architecture is adopted, a dynamic threshold algorithm and a multi-level association map are combined, real-time detection and root positioning of cost anomaly are achieved, and the three-layer structure is data acquisition, dynamic modeling and visual tracing. The system has the advantages that dynamic adaptability is improved, reasonable fluctuation and abnormal risks are accurately distinguished, the system breaks through the rigid limitation of a traditional static rule, the judgment threshold value is dynamically adjusted by fusing market dynamics (such as bulk commodity prices and policy adjustment) and real-time service data, and a supplier-production-logistics-storage association graph is constructed, so that the risk of abnormal risks is accurately distinguished. And the conduction path of the cost abnormity in each link of the supply chain is visually displayed.

The invention discloses a sparse regularization direction of arrival estimation method based on a risk minimization principle, and belongs to the technical field of array signal processing and underwater acoustic signal processing. The method comprises the steps of receiving array signals and establishing an observation model; constructing a sparse representation and over-complete dictionary; establishing and initializing a regularization optimization model; carrying out adaptive weight updating and risk-driven parameter selection; after regularization parameters are determined, a fast iterative shrinkage threshold algorithm FISTA is adopted to carry out optimization solution, and dictionary refinement is carried out on the detected direction after each iteration convergence so as to reduce off-grid errors; and after a small amount of outer layer iteration is repeated, outputting a final DOA estimation result and corresponding power. According to the method, self-adaptive selection of regularization parameters and noise levels can be realized, and the problem of precision degradation under complex conditions of low signal-to-noise ratio, limited snapshot number, signal source correlation, power imbalance and the like is effectively solved without manual parameter adjustment, so that the robustness and practicability of estimation are remarkably improved.

The invention relates to the technical field of community management, in particular to a dynamic monitoring analysis method and system for vehicles in an intelligent community, and the method comprises the steps: collecting multi-source data through a sensor, and generating structured sensing data through a lightweight OCR algorithm; calculating the real-time weight of each sensor by using a self-adaptive dynamic weight algorithm according to the structured sensing data in combination with environmental parameters and equipment states, calculating fusion confidence through a feature confidence fusion formula, rejecting unqualified data, and generating high-precision sensing data; constructing a community digital twinning scene, performing anomaly judgment through a dynamic threshold algorithm, positioning a fault through a three-level mechanism, generating a configuration updating instruction, and generating system state data; and inputting the high-precision sensing data, the system state data and the historical vehicle violation records into the game theory model to calculate the vehicle violation probability, and generating a differentiated management strategy according to the vehicle violation probability. According to the scheme, precise management of community vehicles is realized through multi-source fusion and hierarchical association.

The invention provides a steel bridge automatic finite element modeling method fusing three-dimensional point cloud and intelligent graph recognition, and the method comprises a bridge key construction external information automatic extraction method based on a point cloud model, and the method comprises the steps: carrying out the principal component analysis and transformation of a bridge point cloud principal axis; and secondly, carrying out secondary segmentation on the bridge structural member by using an adaptive threshold algorithm and a region growing RANSAC algorithm, and finally, realizing information extraction of a key carrier surface through a minimization loss function and parameterization expression, and accurately extracting external key parameters. On the other hand, a structure internal information extraction method based on deep learning and drawings is provided. According to the method, geometric dimension measurement and automatic finite element modeling can be automatically completed, reliable digital twin model support is provided for bridge safety assessment, and the method has high engineering application value.

The invention belongs to the technical field of medical health monitoring, and provides a human health prediction method and system based on facial video physiological signal detection, and the method comprises the steps: collecting a facial video stream, and extracting time sequence physiological signals such as heart rate, HRV, respiratory rate and the like through an rPPG algorithm; constructing a graph database individual health portrait in combination with multi-scale time sequence alignment; adopting a dynamic threshold algorithm to detect instantaneous anomaly, and fusing nonlinear dynamics and waveform morphological characteristics to quantify anomaly; constructing a hybrid model, extracting space-time and high-order features, and modeling multi-parameter interaction; a prediction result is dynamically corrected based on a Bayesian algorithm, and health risk layering is realized through clustering; and outputting the visual health report. Through non-contact monitoring, multi-modal fusion and edge-cloud collaborative architecture, the problems that traditional equipment is low in compliance, non-contact technology is insufficient in precision and prediction is shallow are solved, dynamic health prediction and closed-loop management are achieved, and the system is suitable for scenes such as remote monitoring and chronic disease screening.

The invention discloses a credit risk prediction method based on data and dynamic feature optimization and Bagging integration, and the method comprises the following steps: S1, obtaining an initial data set, carrying out the data preprocessing, balancing the data set through an ATGA algorithm, and reducing the data noise; s2, using LightGBM as an agent model, calculating importance scores of all features, gradually screening key features, adaptively adjusting feature screening standards according to data distribution through a dynamic threshold algorithm, gradually screening feature subsets, and inputting the screened features into a subsequent model; and S3, constructing a TabPFN model by using a Bagging integration algorithm, introducing a dynamic weight distribution mechanism, adaptively adjusting the weight of each base model according to sample features, and finally realizing credit risk prediction through the TabPFN model. According to the method, the credit risk prediction process is comprehensively optimized in three aspects of data set optimization, feature screening optimization and model performance optimization, so that indexes such as the accuracy rate and the rate of return are improved.

The invention discloses a new energy automobile shock absorber state intelligent early warning method based on attitude monitoring, and relates to the field of state intelligent early warning, and the method comprises the steps: collecting multi-dimensional attitude data such as the vertical acceleration, the pitch angle speed and the roll angle speed of an automobile in real time, and carrying out the self-adaptive detection of an impact event through combining with a dynamic threshold algorithm; and normal driving vibration and abnormal impact working conditions can be accurately distinguished. In the feature extraction stage, multi-dimensional physical features such as a peak-to-peak value, a root-mean-square value, a damping ratio and the like are combined and coded, so that the dynamic attenuation characteristics of the shock absorber in an impact event can be comprehensively represented. By establishing a similar event feature database in a healthy state and performing multi-feature space deviation calculation by adopting a mahalanobis distance, the robustness of anomaly detection is remarkably improved. Therefore, the technical problems of poor environmental adaptability, low diagnosis precision and the like of the traditional method can be effectively solved.

The invention relates to the technical field of electrical performance testing, and discloses a multi-class nonferrous metal breakdown voltage testing device, system and method. The testing device comprises a testing groove provided with a plurality of independent clamping grooves, a temperature and humidity regulation and control system, an electrode module, a power module and a data acquisition module, and high-precision breakdown voltage testing of multiple samples in a constant environment is achieved through multi-channel synchronous testing, temperature and liquid level dynamic control, symmetrical electrode layout and high-speed data acquisition. The system is integrated with a central control unit, the boosting rate is automatically set according to sample parameters, multiple channels independently boost and synchronously collect breakdown data, and a double-threshold algorithm is adopted to judge breakdown events in real time. The method comprises the whole process of solution preparation, sample installation, electric field optimization, boost control and data judgment. Compared with the prior art, the testing efficiency and the measuring precision are improved, the environmental interference is reduced, and the method is suitable for batch detection and quality control of tantalum, niobium and oxide materials of the tantalum, the niobium and the oxide materials of the tantalum and the niobium.

The invention provides a wind power plant equipment health degree assessment method and system based on multi-source data fusion and adaptive dynamic modeling, and the method comprises the steps: collecting multi-source data, carrying out the preprocessing of the data, taking a timestamp as an index, carrying out the fusion of data from different sources through a data fusion technology, intercepting sequence data in a fixed time period p, and carrying out the segmentation of the sequence data in the fixed time period p; carrying out dimensionality reduction on the sequence data by using P CA, and extracting main components; training a dynamic LS < TM > sequence model, calculating a sliding step length q according to a deviation index, setting p = p + q, and returning to training; and outputting a prediction result. An adaptive dynamic LS < TM > model is constructed through multi-source data and a sliding time window with variable width, real-time judgment of fan faults is realized in combination with a dynamic threshold algorithm, data-driven equipment health degree evaluation and degradation trend prediction are supported, an attention mechanism is added, important moments in a time sequence are captured, the convergence speed is increased, and the prediction accuracy is improved. The accuracy of the model is improved, and generating capacity optimization and ecological compatibility decision making are supported.

The invention provides a quick image recognition method for the water quality condition in a water supply pipeline, and the method is characterized in that a high-definition camera with a light source is used for obtaining a color image in the underground water supply pipeline, the color image is converted into a gray image through an OpenCV method, the gray image is converted into a high-resolution image through a super-resolution method, and the high-resolution image is used for recognizing the water quality condition in the underground water supply pipeline. Carrying out segmentation and feature enhancement on a distribution region of suspended particulate matters in the high-resolution image by applying an average signal-to-noise ratio, a local self-adaptive thresholding algorithm and a denoising algorithm, extracting a region related to a water quality turbidity parameter, comparing a feature region containing the water quality turbidity parameter with a standard water quality turbidity grade image database, and obtaining a water quality turbidity grade image; and judging the turbidity grade of water in the in-service water supply pipeline. According to the invention, rapid identification and real-time monitoring of the pipeline water quality turbidity grade in the water supply network are effectively realized, and a new method is provided for rapid identification and condition monitoring of the water quality of the water supply network.

The invention discloses a continental lake basin grain mud layer sequence image recognition method, device and equipment, a medium and a product, and relates to the technical field of geological exploration. The method comprises the following steps of: after acquiring a smooth surface / slice sample grayscale image of a continental lake basin sedimentary rock core, respectively obtaining a global pixel optimal segmentation threshold value and a local pixel self-adaptive segmentation threshold value of each pixel by applying a maximum between-class variance method and a Sauvola local threshold value algorithm; the two thresholds are fused by taking the local contrast normalized value of each pixel as a weight coefficient to obtain a final segmentation threshold, pixel-by-pixel binarization processing is carried out on the grayscale image to obtain a binarized image, the percentage of the target pixel is counted line by line in the binarized image, and the percentage of the target pixel is calculated. A target pixel percentage trend line is formed according to a statistical result, and finally bright color and dark color texture mud layer segments on the trend line are identified, and quantitative geological parameters of the texture mud layer are obtained according to the identification result, so that the segmentation precision and robustness of the complex texture mud can be remarkably improved.

The application provides a wind farm equipment health degree evaluation method and system based on multi-source data fusion and adaptive dynamic modeling, comprising: collecting multi-source data, preprocessing the data, using a timestamp as an index, fusing data of different sources by using a data fusion technology, intercepting sequence data within a fixed time period p, reducing the dimension of the sequence data by using PCA, and extracting main components; training a dynamic LSTM sequence model, calculating a sliding step q according to a deviation index, setting p = p + q, returning to training; and outputting a prediction result. An adaptive dynamic LSTM model is constructed by using multi-source data and a sliding time window with variable width, real-time discrimination of fan faults is realized by combining a dynamic threshold algorithm, data-driven equipment health degree evaluation and degradation trend prediction are supported, an attention mechanism is added, important moments in the time sequence are captured, the convergence speed is accelerated, the accuracy of the model is improved, and power generation optimization and ecological compatibility decision-making are supported.

The invention, which relates to the technical field of industrial equipment state monitoring, discloses a belt conveyor fault detection and early warning system comprising a multi-mode sensing module, an edge calculation module, a feature fusion and analysis module, a risk level determination module, an intelligent early warning module and a visual feedback module. Multi-source data such as images, sound, tension, temperature and vibration in the running process of a conveyor are collected, features are extracted and preprocessed through edge calculation, and a structured feature vector is constructed; through deep fusion and fault identification model analysis, operation state identification and trend score generation, and in combination with a dynamic threshold algorithm, a risk level is determined, the method has a hierarchical response mechanism, supports user feedback driven model online learning and optimization, and improves the accuracy and adaptive ability of fault identification; the system has the advantages of being comprehensive in detection, timely in response and high in intelligent degree, and is suitable for various complex conveying environments.

The invention provides a multispectral visual perception electroslag remelting defect online identification method, which comprises the following steps: constructing a cross-modal perception network through a multispectral camera and an array sensor, and collecting slag bath morphology, multiphase flow parameters and slag system component data; feature fusion is carried out on the multi-source data, a multi-physics field coupling model is established, and a defect inoculation mechanism is analyzed; a hierarchical identification network is constructed based on the fusion features, and defect type determination is realized in combination with a dynamic threshold algorithm; a cross-modal sensing network is constructed to synchronously collect multi-source data, a multi-physics field coupling model is combined to analyze a defect formation mechanism, and a hierarchical identification network is adopted to realize dynamic threshold judgment, so that the technical problem that a traditional method cannot track multiphase flow dynamic and slag system component changes in real time is effectively solved, and the method is suitable for large-scale popularization and application. The method has the capability of capturing multiphase flow dynamic characteristics of the molten pool and real-time evolution of slag system components at the same time, and the identification accuracy of slag inclusion type defects is remarkably improved.