54results about How to "Avoid slow recognition" patented technology

The application discloses a lining cloth dyeing control method and system based on artificial intelligence, and is used for the control field, and the method comprises the following steps: real-time monitoring of multi-source data in the lining cloth dye vat, capturing the video stream inside the lining cloth dye vat; using a Gaussian mixture variational autoencoder model to extract features from the multi-source data, obtaining multi-source feature data; obtaining the changing moment of the fabric state in the video stream, and extracting the fabric state change feature data according to the observed fabric color change and the heterogeneity of the dye distribution in the video stream; fusion of multi-source feature data and fabric state change feature data; introducing a model predictive control layer, predicting the result to adjust the dyeing parameters in real time; establishing an intelligent feedback mechanism. The Gaussian mixture variational autoencoder model is trained and optimized through the small-batch stochastic gradient descent and back propagation algorithm, important features are learned and captured from the data.

The utility model relates to visual calibration target device technical field, specifically disclose a kind of visual calibration target device of steel trestle dynamic displacement monitoring, including target mechanism, the upper end side of the target mechanism is provided with wire arrangement mechanism, the front lower end of the target mechanism is provided with auxiliary flickering mechanism;The target mechanism includes mounting plate, suspension pole, target main body, wiring seat, lamp holder, outer ring reference lamp bead, intermediate layer reference lamp bead and inner ring reference lamp bead.This scheme can adapt to different monitoring distance and camera focal length by the reference lamp bead design of multiple circle layer distribution, ensure target quick identification and stable tracking by outer ring double-layer concentric circle lamp bead under long-distance large visual angle, provide high-precision positioning by intermediate layer polygon lamp bead in medium distance, and rely on snowflake inner ring lamp bead to realize ultimate calibration in nearest distance, only single target can cover the full-scale monitoring requirement from bridge overall shaking to local slight deformation.

The invention discloses a method and system for detecting the internal quality of a polyethylene pipeline hot-melt welding joint, and belongs to the technical field of pipeline quality detection.The method comprises the following steps that before a tensile test is carried out, ultrasonic quality detection is carried out on a casual inspection joint of a polyethylene pipeline set to be detected; dynamically determining an initial tensile rate based on the material grade and the material attribute parameter of the to-be-tested polyethylene pipeline group, and executing a first tensile test at the initial tensile rate; iteratively and dynamically adjusting the tensile rate of the tensile test of the next casual inspection joint based on the tensile test result of the previous casual inspection joint until the tensile test of each casual inspection joint is completed; dynamically executing fracture quality detection based on the fracture position and the fracture characteristic parameter of each spot check joint to obtain a fracture quality detection result of each first joint; and executing internal quality comprehensive detection based on the ultrasonic quality detection result, the fracture quality detection result and the tensile test result of each first joint.

The utility model provides a gated intelligent household system, and belongs to the technical field of intelligent household. Comprising a door control main control board, an intelligent door lock, an intelligent home linkage chip, an intelligent home gateway and a plurality of intelligent home sub-devices, the intelligent door lock is used for identifying unlocking operation of a user; the intelligent home linkage chip is arranged in the door control main control board, and the intelligent home linkage chip can receive an unlocking signal of the intelligent door lock and a signal of a shortcut key on a door control rear screen panel and generate a corresponding instruction data packet according to the received signal. According to the utility model, the intelligent household linkage chip is arranged, and the intelligent household linkage chip supports a wireless communication protocol of one of WIFI, Bluetooth or Zigbee, so that the intelligent household linkage chip can stably communicate with the intelligent household gateway, the linkage function of the door control and other household equipment is enriched, and the purpose of convenient and fast household automatic control is realized.

The utility model discloses a kind of special treatment tray of diabetes nursing personnel, including treatment tray body, several test tubes, disinfection cotton swab, disinfection cotton ball and disinfection cotton yarn, first baffle is equipped in treatment tray body, and treatment tray body is divided into dirt area and clean area in the upper and lower sides of first baffle.The utility model divides treatment tray body into dirt area and clean area by first baffle, places small tray and sponge body in clean area, places several test tubes in several circular insertion slot, disinfection cotton swab and disinfection cotton ball are placed in first recess and second recess respectively, and syringe and injection medicine bottle are placed in small tray, separate with second baffle, load seat is placed on the long side of dirt area with U type clamping groove clamping, it is convenient to place several disinfectant bottles in several circular load grooves, realize the classification and separate placement of medical supplies, convenient and fast identification and use, effectively isolate sterile and germ area, avoid cross infection.

The invention belongs to the technical field of transformer fault detection, and discloses a transformer fault diagnosis method and related device.The transformer fault diagnosis method comprises the steps that primary side power data and box surface vibration data of a to-be-diagnosed transformer are collected, and primary side data and vibration data are obtained; comparing the primary side data with a preset power data threshold value, and judging whether the primary side data is abnormal or not; if the abnormity judgment result of the primary side data is abnormal, outputting and obtaining an electrical operation state fault diagnosis result of the to-be-diagnosed transformer; if the abnormity judgment result of the primary side data is normal, performing feature extraction on the vibration data to obtain vibration feature data; inputting the vibration characteristic data into a pre-trained support vector machine model for fault diagnosis, and outputting a mechanical health state fault diagnosis result of the to-be-diagnosed transformer; according to the invention, through multi-dimensional data acquisition, a multi-level diagnosis process and an advanced data processing algorithm, accurate and rapid identification of the transformer fault is realized.

The invention belongs to the technical field of wheat molecular breeding, and particularly relates to an SNP molecular marker related to wheat grain hardness on a wheat chromosome. The SNP molecule is named as qHI.1B-1; the molecular marker contains a nucleotide sequence of which the polymorphism at the 393305782bp of the wheat 1B chromosome is A or G; the wheat with the polymorphic site genotype of GG has lower grain hardness compared with the wheat with the genotype of AA. Based on the particularity, the inventor carries out further BSR analysis on an F2 population constructed based on the two varieties, and the SNP site is obtained. Based on the site, a certain foundation can be laid for early and rapid detection and analysis of the wheat hardness character, the subsequent selection efficiency can be improved, the breeding period can be shortened, and wheat varieties with ideal grain hardness can be rapidly identified and cultivated.

The application discloses an engine fault diagnosis method based on a multilevel progressive detection principle, and has the characteristics that the method comprises the following steps: engine timing geometric deviation angle measurement, to complete mechanical layer detection; engine crankshaft and camshaft Hall pulse signal measurement and correction, to complete electrical signal angle layer detection; engine cold test speed-up working condition crankshaft angle domain / frequency domain comprehensive analysis, namely vibration acceleration layer detection; and artificial intelligence fault identification based on big data. The method provided by the application can progressively judge and identify faults through the mechanical layer, the electrical signal angle layer, timing belt tension or vibration frequency and vibration acceleration layer, so that common engine faults in the engine cold test and bench test process can be quickly identified. In the judgment of the vibration acceleration layer, the engine fault is quickly identified and pre-judged by combining the mechanical layer and the electrical signal angle layer and through an artificial intelligence fault judgment algorithm based on big data.

This invention provides a method, device, and medium for automatic identification and detection of extended power supplies for nodal seismographs, belonging to the field of seismic exploration. The method includes: confirming good contact at the extended power supply interface; obtaining the remaining capacity of the extended power supply; using a step-up / step-down DC-DC circuit to boost the output voltage of the extended power supply; and judging the connection status of the extended power supply in real time. This invention can quickly complete the automatic identification and detection of extended power supplies, effectively improve the success rate of assembly, ensure the reliability of the extended power supply, reduce the risk of rework due to operational errors, and improve construction efficiency.

The application discloses a kind of cloud edge end collaborative hot continuous rolling process hydraulic system oil leakage monitoring system, belong to industrial fault monitoring technical field, the system uses cloud edge end architecture;Wherein, end side is used to collect the working parameter of hot continuous rolling process finishing mill hydraulic screwdown system;Edge side is used to the working parameter of end side collection is preprocessed, and the working parameter after preprocessing is input into the detection model trained by cloud, utilizes the detection result output by the detection model, to detect whether the hot continuous rolling process finishing mill hydraulic screwdown system appears oil leakage fault;Wherein, detection model fuses convolutional neural network, long short-term memory network and multi-head attention mechanism;Cloud is used for the training and updating of detection model.The application uses deep learning fusion model to carry out feature extraction to multi-source time series signal, intelligently identifies to leakage type, and can realize the rapid early warning and visual display of hydraulic system oil leakage.

This utility model relates to the field of testing device technology and discloses a testing device for an automotive cockpit controller, comprising: a main cabinet, which is divided into multiple functional layers along a first direction by partitions, the functional layers being suitable for testing at least one function of the automotive cockpit controller; and a darkroom connected to the main cabinet, the darkroom being suitable for forming a dark space; the main cabinet includes: a first side panel, the first side panel being suitable for mounting the object under test for testing in the darkroom. By dividing the main cabinet into multiple functional layers along the first direction, the parallel testing capability of multiple functions of the automotive cockpit controller is realized. This not only optimizes space utilization but also significantly improves testing efficiency, allowing different functional modules to be tested simultaneously, greatly shortening the overall testing cycle. The connection design of the darkroom provides a controllable dark environment for optically related tests, ensuring the reliability and consistency of test data.

The invention belongs to the field of micro-plastic pollutant detection, and particularly relates to a method for identifying micro-plastic in blood based on a laser direct infrared imaging system, and the method mainly comprises the following steps: digesting the blood by using 68% concentrated nitric acid; filtering the digested mixed solution through a filter membrane; rinsing the filter membrane; micro plastic particles are enriched; concentrating and transferring the micro plastic particles; dropwise adding a micro-plastic sample on the glass window and drying; a laser direct infrared chemical imaging system is used for identifying the micro-plastic particles in the blood and the apparent characteristics of the micro-plastic particles. The method disclosed by the invention not only can efficiently and simply extract and collect the micro-plastics in the blood, but also can quickly and accurately identify and analyze the types and apparent characteristics of the micro-plastics, effectively makes up the defects of high detection cost, low detection efficiency and the like of a traditional detection method, has a great application potential in detection of the micro-plastics in the blood of the living body, and has a wide application prospect. The important support is provided for comprehensively analyzing the health risk of the micro-plastic to the life organism.

This utility model discloses an automated product quality inspection device, relating to the field of industrial automation. The device includes a defect inspection station, an infrared scanner, a robotic arm, and a computer. The robotic arm picks up parts transported to a designated location via a conveyor belt and places the product to be inspected on the defect inspection station. The defect inspection station performs a preliminary inspection of the product's exterior; after confirming no surface damage, the infrared scanner is activated to perform internal scanning inspection. The computer integrates and analyzes the detected data and outputs the results. This utility model, through an automated product quality inspection device, solves the problems of poor real-time performance, low accuracy, and poor compatibility in current inspection technologies, achieving efficient automated inspection and identification, improving inspection accuracy, and being compatible with various products.

The utility model relates to the technical field of transfer devices, and particularly discloses a PCB (printed circuit board) transfer device which comprises a placing frame, two partition plates are arranged in the placing frame, and transverse plates are arranged on the left side and the right side of the upper end face of the placing frame; the limiting mechanism comprises side plates fixedly connected to the upper end faces of the two transverse plates, a movable plate arranged on the inner side of the right side plate and a plurality of containing plates fixedly connected to the opposite sides of the left side transverse plate and the movable plate, the space for containing the PCBs is adjusted according to the different sizes of the PCBs, the PCBs of the same type and the same specification are separately placed in the same space, and therefore the PCBs of the same type and the same specification are placed in the same space. The PCB transferring device is simple in structure and convenient to use, the required PCB can be quickly recognized and positioned in the transferring process, the transferring efficiency is improved, the movable plate is pushed to move through the threaded rod till the distance between the left side plate and the movable plate is matched with the large-size PCB, then the large-size PCB can be limited and fixed, and the transferring stability is improved.

This application relates to a method, apparatus, equipment, and medium for identifying and locating leaks in public building restrooms. The method includes: extracting local time-frequency features from a two-dimensional feature matrix of a log-Melogram using a CNN feature extraction module and converting it into a one-dimensional time-series feature sequence; inputting this sequence into a Transformer encoder; capturing the global temporal dependencies of the one-dimensional time-series feature sequence through a multi-head self-attention sublayer; and processing this sequence through a feedforward neural network sublayer to determine a first output feature sequence; inputting this sequence into a Transformer decoder; modeling temporal causality through a masked multi-head self-attention sublayer; and refining the features by fusing global time-frequency features through a cross-attention sublayer to determine a second output feature sequence; and outputting a classification probability distribution by a Softmax layer; if the maximum classification probability is higher than a preset confidence threshold, then the category of the leaking device corresponding to the maximum classification probability is taken as the category of the leaking device in the restroom to be detected. This application enables early detection of leaks, preventing secondary problems caused by the expansion of leaks.

The utility model belongs to pipeline arrangement technical field, and disclose a kind of integrated pipeline arrangement device of assembly type laboratory, including bridge, bridge is provided with pipe fitting room in, several first sliding slots are provided in pipe fitting room, several first sliding blocks are slidably connected in first sliding slot, the upper surface of first sliding block is fixedly connected with clamping block, the both sides of clamping block are provided with clamping groove bridge is provided with wire room in, several second sliding slots are provided in wire room, the inner surface of second sliding slot is fixedly connected with mounting plate, the outer surface of mounting plate is fixedly connected with spring, the end of spring away from mounting plate is fixedly connected with second sliding block, the upper surface of second sliding block is fixedly connected with clamping plate, by the cooperation of pipe fitting room, first sliding slot, first sliding block, clamping block and clamping groove, every pipe fitting can be effectively fixed, can prevent pipe fitting from loosening, falling off due to vibration, impact and other reasons, reduce the risk of leakage, can effectively shorten installation time, and can help pipeline neat arrangement according to design specification.

This invention belongs to the field of UAV navigation and augmented reality technology, and relates to a passive geographic information overlay system and method based on cross-modal features. The system includes: an airborne visual acquisition module installed on the UAV platform for real-time acquisition of video frames and flight attitude data; an edge computing processing module for receiving video frames and performing geographic information calculations, calculating confidence levels, and outputting geographic feature information that meets a set confidence threshold as a preferred dataset; and an augmented reality display module for overlaying the geographic feature information from the preferred dataset onto the video image in an augmented reality manner. Its advantages include the ability to perform positioning matching and augmented display without relying on satellite signals in GNSS-denied environments, strong anti-interference capabilities, high positioning accuracy, and low overlay latency. It can overlay and present geographic features in UAV videos, facilitating users to quickly identify and understand the corresponding geographical location and environmental elements in the image.

The application discloses a kind of based on multi-core heterogeneous vehicle gateway monitoring method, device, equipment and medium, belong to vehicle gateway monitoring technical field.The method comprises: control application processing core continues to read the current reception total number of Ethernet heartbeat packet, compare with the reception total number of adjacent period, whether new Ethernet heartbeat packet cannot be sent to Ethernet switch chip is monitored, if yes, the accumulation update of exception count is completed and threshold discrimination is carried out, when determining that Ethernet switch chip exists exception, first hardware reset request is sent to real-time processing core, and reset control of Ethernet switch chip is independently completed by real-time processing core.The application can let application processing core focus on the reading and processing of service data, let real-time processing core focus on hardware control response, clearly divide the working boundary of dual-core, realize the rapid identification and autonomous repair of Ethernet switch chip exception, meet the practical application needs of the rapid fault repair of vehicle gateway.

This invention relates to a dedicated clutch chainring assembly for a mid-drive motor. This chainring assembly, used to control the speed of the mid-drive motor, includes a bottom bracket, a first crank, and a second crank. A first chainring is mounted on the bottom bracket, and a rotating component is concentrically mounted on the first chainring. A first spring is provided between the rotating component and the first chainring to generate rotational resistance and to reset the rotating component during rotation. The magnitude of the rotational resistance is positively correlated with the rotation angle of the rotating component. This invention modifies the chainring so that the teeth on the chainring can rotate, and achieves stepless control of the mid-drive motor based on pedaling force. It not only accurately and quickly identifies and responds to pedaling force, allowing the mid-drive motor to adjust accordingly and improve the riding experience, but also integrates sensors with the chainring, simplifying the vehicle's structure and size, making the vehicle more aesthetically pleasing.

The utility model relates to the technical field of RFID tags, and discloses an RFID archive electronic tag, which comprises a tag body, a fastening assembly is arranged at the bottom of the tag body, a protective assembly is arranged at the top of the tag body, the fastening assembly comprises a base band, the base band is adhered to the bottom of the tag body, and the protective assembly is arranged at the bottom of the tag body. A fiber net is arranged in the base band, the bottom of the base band is coated with an adhesive layer, the front end of the base band is connected with a pulling piece, and the label body and the base band are mutually bonded and fixed through adhesive. By means of the pull piece and the fiber net, the label body is convenient to take, the situation that the adhesion effect is affected due to the fact that the adhesive layer is polluted by the hand is avoided, and the tensile property of the label body is enhanced; through the grooves in the inner sides of the fluorescent strips, the bar code stickers and the number stickers can be conveniently pasted, the phenomenon that the stickers fall off due to scratching of external objects is reduced, the label body can be rapidly positioned under the dark light condition, and archives can be rapidly identified and searched conveniently.

This utility model discloses a temperature and humidity regulating laboratory specimen preservation device, belonging to the field of laboratory equipment technology. It includes a cabinet with rotating doors on both sides, and also includes a temperature sensor, a humidity sensor, a small compressor, an ultrasonic humidifier, a condenser dehumidifier, a heater, and a controller for intelligent control. This utility model, by arranging temperature, humidity, and light sensors inside the preservation cabinet, allows for comprehensive and accurate collection of temperature, humidity, and light intensity data within the cabinet. Combined with an intelligent control module, a temperature and humidity regulation module, and a light regulation module, it achieves precise temperature and humidity regulation, ensuring sufficient light intensity for the specimens and maintaining a suitable preservation environment. The controller's intelligent control module automatically controls the temperature and humidity regulation module according to preset temperature and humidity standard ranges, eliminating the need for frequent manual operation and improving the efficiency and stability of specimen preservation.

The application discloses a chromosome image recognition method and device, computer equipment and a storage medium, and relates to the technical field of chromosome image processing. The method comprises the following steps: after obtaining a plurality of independent chromosome monomer images, image feature extraction processing is performed on each chromosome monomer image, corresponding image features are obtained, the image features are subjected to optimization processing based on a white-box neural network ReduNet, corresponding image optimization features are obtained, and finally, the image optimization features are input into a pre-trained chromosome classifier to output corresponding chromosome monomer classification results. Through the vector optimization of the embedding representation after the image feature extraction network, the final chromosome classification recognition accuracy can be significantly improved, and through corresponding experiments, it is found that common evaluation indexes such as accuracy, precision and F value can reach more than 92.39%, and even reach 97.43%.

The invention discloses a dynamic insulation diagnosis method for a power battery system based on an adaptive threshold value, and relates to the technical field of data analysis. A dual-channel monitoring mechanism of a main channel and an auxiliary channel is introduced on the basis of traditional dynamic threshold calculation, and rapid recognition and response to sudden insulation faults are achieved. The auxiliary channel senses the change rate of the insulation resistance in real time and triggers a threshold freezing mechanism after identifying a jump event, so that the problem that a dynamic threshold covers a sudden fault risk due to smooth delay is effectively avoided; furthermore, by quantitatively analyzing the state change in the freezing period, calculating a state recovery index and taking the state recovery index as a judgment basis for judging whether the freezing state is quitted or not, the defects that in a traditional mechanism, a threshold value is not matched with the alarm frequency rhythm, and whether the system is recovered to be stable or not cannot be accurately judged are overcome; therefore, the accurate switching of the insulation diagnosis mechanism between the burst state and the stable state is realized, and the response speed of diagnosis, the judgment accuracy and the overall safety of system operation are remarkably improved.

The utility model relates to cable technical field especially relates to a kind of anticorrosive marine cable.Its technical scheme includes: cable body, the cable body includes multiple conductor body, the outside of each conductor body is provided with natural color insulating layer.The utility model is protected by setting up semiconductive conductor shielding layer and semiconductive insulation shielding layer to natural color insulating layer and color bar color separation insulation, by setting up isolation layer protection cable body internal structure, improve the fireproof and explosion-proof performance of cable body, by setting up inner protective layer, natural color insulating layer and color bar color separation insulation are not damaged by external factors, while preventing moisture and corrosive substance from invading, by setting up armoring layer, the mechanical strength and explosion-proof performance of cable body can be enhanced, by setting up outer protective layer, cable internal structure can be effectively protected from the corrosion of marine environment, while also having certain flame retardancy, the explosion-proof performance of cable can be improved.

This utility model discloses a road reinforcement device after road core sampling, including a reinforcement trolley. The rear end of the reinforcement trolley is equipped with an identification component, a moving component, and a compaction component for identifying and reinforcing the road core sampling site. The identification component is horizontally mounted at the rear end of the reinforcement trolley, and several identifiers are mounted on the bottom of its mounting frame. By setting the identification component and the moving component at the rear end of the reinforcement trolley in the road reinforcement device, when the identifiers identify the core sampling site on the road, they simultaneously control the horizontal moving mechanism, the lateral moving mechanism, and the height adjustment mechanism on the moving component to move the grouting nozzle mounted on the mounting plate to the inside of the core sampling site, thereby grouting the core sampling site to fill it with grout. Simultaneously, the compaction component is activated, causing the compaction roller on the compaction component to perform roller pressing reinforcement on the core sampling site as the reinforcement trolley moves, thus improving the reinforcement efficiency and quality of the road core sampling site.

The utility model belongs to the field of montmorillonite preparation inspection, particularly relates to a montmorillonite preparation inspection device, and aims to solve the problems that in the prior art, after various reagents and samples are put in, mixed liquid in a container is shaken up generally through a device or hand cranking, and in the process, the mixed liquid in the container easily generates excessive foam or splashes out of the solution, so that the operation is inconvenient. In order to solve the problems that in the prior art, due to the fact that a sample is placed on the base plate, the inspection process of the sample is affected, and the inspection accuracy of the sample is further affected, the following scheme is provided that the device comprises a base plate, a containing groove is formed in the top of the base plate, a container is placed on the containing groove, a controller is fixedly connected to one side of the top of the base plate, and a vertical block is fixedly connected to the top of the base plate; according to the montmorillonite preparation color identification device, rapid stirring and color identification of a montmorillonite preparation are achieved, the detection efficiency is greatly improved, excessive foam or solution splashing caused by shaking of a reagent is reduced, meanwhile, due to the design of a wiping plate and a wiping cotton ring, cleaning and maintenance of a stirring rod are facilitated, and the service life of the device is prolonged.

The application relates to the technical field of integrated power supply, in particular to an integrated power supply module assembly which comprises a shell, a power supply module, a circuit breaker and an interface socket. The power supply module is fixed to one side in the shell, the circuit breaker is located on the outer surface of a first panel, the incoming line end of the circuit breaker is connected with the power supply module; the interface socket is located on the outer surface of a second panel, the interface socket comprises a direct-current interface socket and an alternating-current interface socket, the direct-current interface socket is connected with the negative electrode of a power supply output end and the outgoing line end of the circuit breaker; the alternating-current interface socket is connected with a power supply input end and the outgoing line end of the circuit breaker. The power supply module, the circuit breaker and the input / output interface are highly integrated in a unified shell, and through the optimized space layout design, the miniaturization of the equipment is realized under the premise of ensuring the electrical safety gap.

The invention provides a multi-mode sensing anomaly detection and acousto-optic interaction method and system for process level equipment, and relates to the technical field of industrial automation, energy consumption fluctuation data and wireless communication quality data of a process level equipment cluster are collected, an original operation data set is formed by combining equipment identification, and equipment association identification is marked for the original operation data set; the method comprises the following steps: forming an associated data set, adjusting and verifying an initial filtering gain parameter of a self-adaptive filtering circuit, then determining a target filtering gain parameter, carrying out signal filtering on the associated data set, finally generating an equipment operation link, identifying abnormal equipment with parameters deviating from a normal range in the link, generating an equipment abnormity judgment result, and determining an alarm level. Corresponding alarm audio signals, optical signals and ultrasonic carriers are generated; the characteristic parameters are modulated to ultrasonic carrier waves, and directional acousto-optic prompting of an abnormal area is completed in combination with optical signals, so that accurate identification of implicit anomalies under complex working conditions is realized, and intelligence and precision of anomaly detection and acousto-optic prompting are realized.

The invention discloses a tailing pond intelligent identification system and method based on multi-agent collaboration, and the system comprises an image preprocessing module which is used for carrying out the standardization processing of an input original remote sensing image, and generating a standardized image slice; the multi-agent iteration collaborative reasoning module is connected with the image preprocessing module and is used for performing iteration identification, classification and boundary optimization on the image slices; and the final boundary extraction and output module is connected with the multi-agent iteration collaborative reasoning module and is used for converting the selected target set after iteration convergence into geographic coordinate information and outputting the geographic coordinate information. According to the method, cognitive reasoning of visual languages is realized, the robustness and interpretability of tailings pond recognition under complex backgrounds and fuzzy conditions are improved, rapid deployment and recognition of zero samples or few samples, self-adaption and high-precision positioning of target boundaries instead of blind iteration are realized, and accurate classification and efficient recognition of different types of tailings ponds are realized.

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.