628 results about "Correction algorithm" patented technology

The invention discloses an artificial board surface defect intelligent detection method and system based on machine vision, and particularly relates to the technical field of artificial board surface defect detection. An artificial board surface image is collected, image preprocessing, feature extraction, defect segmentation and classification recognition are carried out through a deep learning algorithm, a multi-scale convolutional neural network is adopted to carry out feature extraction on the image, a shallow convolutional layer captures small defect details, a deep convolutional layer recognizes global features of large defects, and a multi-scale convolutional neural network is adopted to carry out feature extraction on the image. Accurate defect segmentation is carried out through a Mask R-CNN model, a redundant frame is removed in combination with a non-maximum suppression algorithm, the classification problem of adjacent defects is corrected by using an error correction algorithm in combination with the spatial relationship and morphological characteristics of the defects, and defect information is fed back to a production line control system in real time; and defective products are automatically removed or production process parameters are automatically adjusted, so that the automation level of a production line is effectively improved, the product quality is optimized, and human intervention and production cost are reduced.

The invention discloses a screening efficiency online monitoring method, device and system based on a vibrating screen, and relates to the technical field of vibrating screen efficiency monitoring, and the screening efficiency online monitoring method comprises the following steps: collecting working parameter data of the vibrating screen in real time through a sensor arranged on the vibrating screen; the accuracy and stability of screening efficiency are improved by monitoring key parameters such as humidity, particle size distribution and vibration frequency in real time and combining a screening effect model and a dynamic correction algorithm. The system automatically adjusts equipment parameters, equipment overload or screen damage is avoided, the material passing rate is optimized, and production losses are reduced. Meanwhile, manual intervention is reduced through an automatic control mechanism, the automation level of a production line is improved, human errors are reduced, the equipment fault and shutdown time is shortened, the production efficiency is improved, and the service life of the equipment is prolonged.

The invention relates to the technical field of digital twinning control, and discloses a composite structure prefabricated segment digital twinning-adaptive assembly correction method and system, and the method comprises the steps: reading BIM geometric model data, and constructing an assembly reference coordinate system and a digital twinning geometry of a prefabricated segment; composite data are collected in real time, and end tooth groove boundary feature point cloud is extracted; matching the actually measured point cloud with the digital twinborn geometry through a dynamic point cloud registration technology, calculating a six-degree-of-freedom pose error vector, and generating a predicted total pose error vector in combination with a pre-constructed pose drift prediction model; based on the error vector, a mechanical fine adjustment jack is driven to execute position and posture adjustment until the position and posture are converged, and then tooth groove precise meshing and mechanical locking are completed; according to the method, through the synergistic effect of the dynamic mapping of the digital twinborn model and the self-adaptive correction algorithm, high-precision dynamic correction of the multi-combination structure segment assembly process is achieved on the premise that an original mechanical connection structure is not changed.

The invention discloses a territorial space planning index dynamic evaluation model, and particularly relates to the field of territorial planning, and the model comprises a multi-source heterogeneous data fusion module, an index system construction module, a dynamic adjustment model construction module, a scheme evaluation and optimization module, and a model application and dynamic updating module. According to the method, various spatial data are integrated through a multi-source heterogeneous data fusion technology, and the data quality is ensured by adopting a three-stage cleaning assembly line and an intelligent correction algorithm; the method comprises the following steps: constructing an index system based on six core contents of territorial space planning, and determining an index weight by applying a Delphi method and an analytic hierarchy process to realize adaptive adjustment of an index threshold; the model adopts time sequence analysis, regression analysis and a machine learning algorithm to predict a development trend, a dynamic feedback mechanism is established in combination with a system dynamics method, and an evaluation scheme is quantified through an analytic hierarchy process; the model performance is evaluated by adopting indexes such as accuracy rate and precision rate, and parameters are continuously optimized based on an evaluation result; finally, the model is integrated to a service system to realize real-time monitoring, and dynamic updating is realized through three mechanisms of data driving, performance triggering and regular maintenance.

The invention relates to the technical field of three-dimensional imaging, and discloses a high-precision underwater structured light three-dimensional point cloud imaging method and system, and the method comprises the following steps: deploying a plurality of sensor nodes, monitoring the temperature, salinity and turbidity key parameters of a water body in real time, selecting proper light source wavelength and power according to the collected data, and carrying out the three-dimensional point cloud imaging of the underwater structured light. It is ensured that the optimal signal-to-noise ratio is obtained under the current condition, and meanwhile the exposure time and gain setting of the camera are initialized. According to the method, the optimal signal-to-noise ratio is obtained by monitoring water body conditions in real time and dynamically adjusting light source characteristics, the optimal structured light pattern is predicted by using a machine learning model, scattering noise is removed, advanced registration and fusion algorithms and an accurate phase unwrapping method are applied, and a high-quality comprehensive structured light pattern is generated. Point cloud gaps are filled through a geometric correction algorithm and an interpolation / filtering algorithm, so that a reconstruction model is closer to the surface of a real object.

The invention relates to a method and system for monitoring the running state of a water pump unit based on the digital twin technology, and belongs to the technical field of pipeline pump monitoring. The method comprises the steps that a digital twin model fusing equipment parameters and environment sensing data is constructed, and multi-dimensional running data such as dynamic pressure, flow, electric energy quality and temperature are collected in real time in combination with the Internet of Things technology; a multi-dimensional correction algorithm is innovatively provided, fluid flow velocity calculation is corrected through pipe opening structure parameters, lift calculation is dynamically corrected through a temperature compensation algorithm and a mechanical wear factor, and dynamic compensation is conducted on the water pump efficiency by introducing a harmonic loss calibration mechanism. Based on a two-channel machine learning model, a fault incidence matrix is constructed by improving a weighted K-means algorithm to realize equipment health state clustering, an efficiency attenuation trend is predicted in combination with an LSTM network, static characteristics and dynamic time sequence data are fused by using an attention mechanism, and finally an accurate fault diagnosis report and a maintenance strategy are generated.

The invention discloses a large model scheduling multi-agent power grid fault coping strategy knowledge graph extension method and system, and the method comprises the steps: carrying out the modeling of task distribution as a mixed integer programming problem, and achieving the solving through a relaxation-correction algorithm; extracting and normalizing a core task of knowledge, and outputting structured data; summarizing the obtained text segments, and prompting a large language model to retain key entities, relationships and domain-specific terms; using an LLM-based named entity recognition technology, combining with prompt and dictionary / ontology filtering in the power dispatching field, recognizing related entities in a text, and normalizing the related entities into a standard form in a knowledge graph; detecting logic contradictions between the newly extracted triples and existing relationships in the knowledge graph, and classifying and solving the contradictions by utilizing debate prompts based on LLM (Logistics Library Model); summarizing the plurality of verification signals, and calculating the global confidence, definition and correlation score of each triple;

The present invention relates to the technical field of railway inspection, and relates in particular to a railway foreign-object intrusion detection method and system based on deep learning. The method comprises: S10, acquiring image data to undergo detection; and S20, inputting the image data into a trained attention semantic segmentation network to obtain a foreign-object detection result. The attention semantic segmentation network is obtained by first training a preset attention semantic segmentation network architecture using a preconfigured data set, and then re-training the attention semantic segmentation network on specified image data using an adaptive correction algorithm. A backbone network architecture is obtained by inserting a specified attention mechanism at a specified position within a pre-selected residual neural network, and using three parallel dilated convolutions as an initial convolutional layer in the residual neural network. A dual-branch decoder combines an edge recognition branch and a semantic segmentation branch. The method is applicable to various scenarios, achieves improved detection accuracy and maintains a lightweight design.

The invention discloses a PLC-based optical fiber winding tension control method and system, and the method comprises the steps: obtaining the remaining length of an optical fiber according to a set and real-time winding length through a length decreasing model, switching a winding unit to an ending preparation state and triggering a generator to decelerate when the remaining length is smaller than a threshold value, and obtaining the position information of a winding mechanism through a photoelectric sensor after the ending speed is reached; a motor speed correction coefficient is obtained through a position loop speed correction algorithm according to the angle difference and the target angle difference, and a control instruction is generated. And if the angle reaches or exceeds the target angle, starting the neural network prediction model to obtain a shutdown advance, and outputting a shutdown signal by the PLC according to the shutdown advance and a preset path model to stop the motor with low vibration. The PLC is utilized to integrate various sensing and intelligent algorithms, high-precision tension control in the full-winding process is achieved, the problems that in the traditional technology, tension control is not stable, and problems are prone to occurring at the end are effectively solved, the optical fiber winding quality and efficiency are improved, and the method adapts to various types of optical fibers and technological requirements.

The invention discloses a circuit breaker service life prediction system and method based on multi-source heterogeneous data fusion and dynamic weight correction, and the method comprises the steps: collecting the current, pressure, displacement, main loop current, voltage, contact temperature, environment temperature and partial discharge of a circuit breaker, and carrying out the multi-source heterogeneous data fusion algorithm, thereby achieving the prediction of the service life of the circuit breaker. Carrying out fusion analysis on the collected data; a dynamic weight correction algorithm is executed according to an analysis result, and different types of fault weights are corrected and adjusted, so that the life prediction model is more accurate; and executing a fault identification algorithm after each action of the circuit breaker, and calculating the service life of the circuit breaker according to the weight and the health index corresponding to each fault. According to the invention, the prediction accuracy of the service life of the short-circuiter is greatly improved, and the use safety of power equipment is improved.

The invention relates to the technical field of the manufacturing industry, and discloses a high-integration variable-focus laser engraving control method and system, and the method comprises the steps: carrying out the high-frequency data collection of the surface of a workpiece through a sensor array, and obtaining an initial surface state data set; and analyzing surface tiny differences by using the feature extraction model, and determining a dynamic trend distribution diagram. A real-time monitoring technique is employed to track the surface change, and focus position correction is triggered when it is detected that the height fluctuation exceeds a threshold. And adjusting the focus position according to a deviation correction algorithm, and driving an execution mechanism to carry out real-time correction. And the processing effect is analyzed and evaluated through data comparison, a parameter optimization instruction is generated, and the dynamic adjustment control model is called to finely adjust operation parameters. And finally outputting a processing quality evaluation result, and generating a complete log file. Real-time precision monitoring and dynamic adjustment in the complex surface machining process are achieved, the machining precision and stability are improved, and the high-end manufacturing standard requirement is met.

The invention relates to the field of sensors, and discloses an anti-electromagnetic interference dynamic compensation method for a gas sensor of power equipment, which comprises the following steps: monitoring and collecting sensor data, uploading the sensor data to a central processing unit for data fusion, and carrying out volatility analysis on the fused sensor data, so as to obtain an anti-electromagnetic interference dynamic compensation result. The change relation of the fluctuation of the sensor data along with time is monitored in real time, and the fluctuation abnormal state of the sensor data is analyzed and detected by combining a fluctuation analysis result with real-time environmental parameters, so that the monitoring duration and the monitoring frequency of the sensor data are dynamically adjusted; analyzing the contribution degree of the electromagnetic interference to the error generated by the gas sensor, detecting whether the error interferes with the measurement state of the gas sensor, and correcting the gas concentration measurement value by using a temperature and humidity-gas response coupling correction algorithm of Kalman filtering to realize the dynamic compensation of the gas concentration measurement value. And the accuracy of the corrected gas concentration estimated value is improved.

The invention discloses a new energy station intelligent micrometeorological monitoring device suitable for extreme weather and a power prediction method, and relates to the technical field of new energy station meteorological monitoring and power prediction. The device integrates a plurality of meteorological sensors, combines satellite meteorological data and ground meteorological station data, improves meteorological monitoring accuracy through data fusion, and gives 8-15-day long-time scale weather forecast. According to the method, based on meteorological data collected in real time and historical meteorological data of a new energy station, a long-short term memory network model and a segmentation multi-target risk method are adopted to carry out risk assessment on extreme weather, and an extreme weather early warning signal is given. A deep neural network is utilized to establish a power nonlinear prediction model of a new energy station, and a higher-precision power prediction result is given through a random forest dynamic correction algorithm. The meteorological and generated power prediction accuracy of the new energy station is improved, and the operation safety of the new energy station is improved.

The invention discloses a BIM-based hydropower station full life cycle design, construction, operation and maintenance integrated control system, and the system comprises an intelligent sensing layer which integrates 5G + Beidou positioning, a LoRa gateway and a sensor, collects multi-source heterogeneous data in real time, and transmits the multi-source heterogeneous data to a digital twinborn layer after the multi-source heterogeneous data is filtered by an edge node; a digital twinborn layer: constructing a parameterized model library through laser point cloud and BIM automatic registration, integrating a geological parameter dynamic correction algorithm, mapping a construction period stress field in real time, and updating model parameters based on sensing data self-evolution; the intelligent decision-making layer performs equipment fault prediction by using an LSTM neural network, optimizes multi-machine load distribution in combination with an improved PSO algorithm, and automatically adjusts a start-stop strategy when the load fluctuates; and the security execution layer is used for triggering equipment operation after virtual twinborn deduction verification through a block chain evidence storage instruction, realizing virtual-real dual verification in combination with an industrial firewall, and finally feeding back a running state to the sensing layer to calibrate and update a model, and supporting intelligent decision.

The invention discloses an AI-based regulation and control abnormal data identification and correction method, system and device and a medium, and belongs to the technical field of power system regulation and control, and the method comprises the steps: obtaining multi-source data of a main distribution network, carrying out the preprocessing, and constructing a power grid topology knowledge graph; the multi-source data comprises full voltage class model data, real-time measurement data and equipment state data. According to the method, a three-layer detection system of a statistics-density mixed detection layer, a graph neural network association detection layer and a time sequence Transform detection layer based on an LOF algorithm and a 3 sigma principle is constructed, deep association analysis of multi-source data is realized in combination with a power grid topology knowledge graph, and an intelligent correction module is used for matching a differentiation correction algorithm for different types of anomalies; the correction weight is dynamically adjusted based on reinforcement learning, and the problems that a traditional method depends on artificial experience to set a threshold value, the recognition capacity of a single algorithm is limited, a correction strategy lacks adaptability, and multi-source data association analysis cannot be achieved are solved.

The invention discloses a risk identification method for pediatric intensive care unit nursing, which relates to the technical field of medical early warning, and comprises the following steps: collecting multi-modal data, and carrying out time synchronization on the collected data; data features are extracted based on the data after time synchronization, fusion processing is carried out through linear weighting, an optimization target is determined, effective fusion of multi-modal data is achieved, and fused high-quality data is obtained; constructing a risk assessment model, calculating a risk value for each time point in combination with historical data, obtaining an overall risk value through weighted average processing, judging a risk level, and optimizing prediction accuracy through a self-correction algorithm; the state of a patient is monitored in real time, the early warning level is set, abnormal conditions are early warned according to the corrected risk value variation, feedback of medical staff is collected, and the system performance is continuously optimized. The method has remarkable clinical application value, and can effectively improve early warning accuracy, optimize medical resource allocation, reduce medical risks and improve overall medical quality.

The invention discloses a building semantic segmentation method for correcting an unsupervised domain adaptive pseudo tag by using a large segmentation model. The method comprises the following steps: collecting a source domain building data set with similar building characteristics with a target domain building data set; generating an initial target domain pseudo tag and an SAM segmentation result based on the source domain building data set and the target domain building data set; correcting the initial target domain pseudo-tag based on a tag correction algorithm by using an SAM segmentation result to obtain a final target domain pseudo-tag; a Deeplabv3 + semantic segmentation network is constructed; training a Deeplabv3 + semantic segmentation network based on the final target domain pseudo tag and the target domain image; and inputting a to-be-detected image to the trained Deeplabv3 + semantic segmentation network, and outputting the position and boundary information of the building in the to-be-detected image to realize building semantic segmentation. According to the invention, full-automatic high-precision building segmentation on the remote sensing image is realized.

The invention discloses a power equipment state evaluation method, system and device for intelligent monitoring and a storage medium, and belongs to the technical field of power equipment state evaluation, and the method comprises the steps: building a multi-physics field coupling calculation model based on the working characteristic parameters of power equipment, and simulating the behaviors of the power equipment in different working states; outputting an expected output value, comparing the expected output value with the physical field signal after time alignment, and calculating a deviation to obtain an initial prediction residual error; according to the initial prediction residual error, adjusting model parameters of the multi-physics coupling calculation model through an incremental correction algorithm, correcting the state estimation value of the power equipment, and updating the initial prediction residual error; based on the updated prediction residual error, analyzing an abnormal state of the power equipment through a residual error value, and extracting a fault feature; and identifying the fault type of the power equipment. According to the invention, through multi-physics field coupling calculation and an incremental correction algorithm, the accuracy and real-time performance of power equipment state evaluation are significantly improved.

The invention relates to the technical field of ecological system service function evaluation. The invention relates to a frequency offset correction algorithm in HPLC (High Performance Liquid Chromatography) + HRF (High Radio Frequency) dual-mode communication. The method comprises the following steps: S1, collecting phase sequence data from an HPLC communication link, collecting frequency offset estimation data and corresponding confidence data from an HRF communication link, and adaptively adjusting a preset detection threshold value of the phase sequence data according to the confidence data; s2, performing pulse abnormal sample point detection and elimination on the phase sequence data according to the adjusted preset detection threshold value; according to the method, the detection threshold is dynamically adjusted through the confidence data, so that the anomaly detection process can be matched with the current interference environment in real time, the integrity of the phase sequence data is maintained, and the problem of insufficient adaptability of a fixed detection threshold in an electromagnetic interference dynamic change scene is solved; and a more reliable phase data basis is provided for subsequent interpolation compensation and frequency offset estimation, so that the frequency offset correction precision of the dual-mode communication system is improved.

The invention discloses a numerical control machine tool machining track real-time correction method based on multi-source visual perception, and the method comprises the following steps: S1, synchronously collecting a multi-source real-time image sequence of a machined workpiece in a machining process through a plurality of visual sensors disposed at different visual angles of a machining region of a numerical control machine tool; s2, performing preprocessing and feature extraction on the multi-source real-time image sequence, and fusing to obtain real-time three-dimensional coordinate information of key feature points, corresponding to the theoretical processing track model, on a processing workpiece; s3, the real-time three-dimensional coordinates of the key feature points are compared with theoretical coordinates of corresponding points in a pre-stored numerical control machine tool theoretical machining track model, and the track deviation between the actual machining track and the theoretical machining track is obtained through calculation; and S4, based on the trajectory deviation, generating a trajectory correction value through a preset correction algorithm. The method can better correct the machining track of the numerical control machine tool in real time.

The invention relates to the technical field of power system monitoring, in particular to a power system parameter dynamic verification method, system and device based on digital twinning and a storage medium. Constructing a digital twinborn model of the power system, establishing a state mapping relation between a physical system and a digital model, realizing bidirectional dynamic mapping between the physical system and the digital model through a digital mapping mechanism, and obtaining system operation characteristics; constructing a parameter verification model based on the operation characteristics, obtaining a verification objective function by fusing physical constraints and data driving, and dynamically adjusting parameter values to generate parameter correction values by evaluating model errors in real time by adopting an adaptive parameter correction algorithm; based on a dynamic optimization mechanism of multi-objective optimization, an optimization strategy is adaptively adjusted according to the running state, and the verification process is continuously optimized; and performing multi-dimensional evaluation on the verification result through the verification precision evaluation system to generate a parameter verification result. The technical problems that a traditional method is inaccurate in modeling, low in response speed and poor in adaptability are effectively solved.

The invention discloses cutting equipment for pipeline prefabrication, belongs to the technical field of machining, and is mainly used for solving the problems of insufficient cutting precision and poor machining stability caused by clamping force fluctuation, pipeline ovality abnormity and dynamic path offset in the pipeline cutting process. According to the equipment, high-precision cutting is achieved through cooperative work of the rack, the clamping mechanism, the cutting executing mechanism, the driving motor and the control module, the clamping mechanism is provided with a pressure sensor to monitor fluctuation of clamping force, and the driving motor is a servo motor with an absolute value encoder to precisely control the rotating angle of a pipeline. The cutting executing mechanism feeds back the displacement track of the cutting machine in real time through a displacement sensor. Synchronous control logic is arranged in the control module, when the fluctuation of the clamping force exceeds the limit, the rotating speed of the servo motor is automatically reduced, the cutting machine is paused, meanwhile, a path correction algorithm is configured, and the proportional relation between the axial moving speed and the radial feeding amount of the cutting machine is calculated in real time based on encoder angle data. And the notch precision and the machining efficiency can be obviously improved.

The invention discloses a quantum measurement and control data processing method and device, and belongs to the technical field of data processing.The quantum measurement and control data processing method and device comprises the following specific steps that firstly, wavelet packet decomposition is conducted on an original measurement signal through a quantum noise feature extraction module, and a quantum noise feature extraction module is obtained; carrying out matched filtering with a preset quantum noise fingerprint database; and 2, performing quantum state reconstruction on the filtered signal by adopting a compressed sensing algorithm, wherein a measurement matrix is dynamically generated according to the Hamiltonian of a quantum system. Through combination of wavelet packet decomposition (time-frequency locality) and a dynamic measurement matrix (Hamiltonian relevance), compared with a traditional Fourier method, quantum state reconstruction errors are greatly reduced, time sequence dependence is captured through an LSTM network, physical legality is guaranteed through a quantum main equation, control parameter speed is optimized, 2ns-level time sequence precision is achieved through an FPGA segmentation correction algorithm, and the method has the advantages of being simple in structure, convenient to operate and high in practicability. And the real-time requirement of the quantum error correction code is met.

The invention discloses a movement detection and feedback intervention system for obese patients, and relates to the technical field of movement intervention for obese patients, the system comprises a data acquisition module, a personalized movement scheme generation module, a movement monitoring module and an interaction module; according to the method, artificial intelligence and big data analysis technologies are deeply fused, multivariate data of a patient is comprehensively and accurately collected, a personalized exercise scheme meeting individual requirements can be generated, the pertinence of the exercise scheme is improved, and during exercise monitoring, the multi-modal data fusion technology is applied, so that the accuracy of exercise monitoring is improved. Various physiological parameters and exercise environment data of a patient are collected in an omnibearing mode, the state of the patient is accurately controlled in real time, once monitoring data deviates from a preset target, parameters such as exercise items, strength, time and frequency in an exercise scheme can be rapidly and dynamically adjusted according to a real-time deviation correction algorithm, the safety of the exercise process is practically guaranteed, and the exercise efficiency is improved. The goal of losing weight is effectively promoted.

The invention discloses a multi-turn memory optomagnetic encoder hybrid system, belongs to the technical field of position and angle measurement encoders, and aims at solving the problems that an existing optomagnetic hybrid encoder is complex in signal processing, long in axial length, insufficient in precision stability and prone to losing the number of turns and the position. The system comprises an optical encoder, a gallium arsenide magnetic signal acquisition sensor, a multi-antipode unequal magnetic ring, a processor, a hardware circuit, an ultra-low power consumption circuit, a signal acquisition module, a signal processing module and the like, integrates self-developed magnetic angle analysis, turn number memory correction and other algorithms, and adopts dual-redundancy design and intelligent signal mode switching. The method simplifies the signal processing flow, shortens the axial length, improves the measurement precision and stability, enhances the environmental adaptability and fault-tolerant capability, prolongs the endurance, and is suitable for the position angle measurement in a high-precision and severe environment.

The invention relates to the technical field of catenary fault detection, and discloses a catenary fault detection method and system based on image recognition. According to the method, multi-spectral image acquisition equipment is deployed along a contact network, and a multi-angle image sequence of contact network components is acquired in real time; performing space-time alignment processing on the image sequence to generate a standardized image data set; constructing a catenary part segmentation model based on the data set, and identifying key parts by means of a deep semantic segmentation technology; extracting a geometric feature map and a deformation feature map, and establishing a component state feature library; analyzing a topological relation among the component characteristics through a space correlation module, and generating a fault correlation graph; updating the atlas by adopting a dynamic feature correction algorithm, and outputting a fault probability distribution matrix; comparing the matrix with a preset fault threshold set, and generating a fault positioning instruction; and after multiple rounds of correction by the iterative optimization module, a final fault detection report is output and is displayed on the human-computer interaction platform in real time, and parameter configuration is supported.

The invention provides a laser ranging device based on an FPGA (Field Programmable Gate Array) and a single photon avalanche diode. The device is characterized in that the distance measuring device is composed of a field programmable logic gate array calculation core (1), a quenching circuit (2), a single photon avalanche diode (3), a pulse laser driving circuit (4), a pulse laser (5) and an optical lens system (6), and the field programmable logic gate array calculation core (1) comprises time digital conversion logic (11) and calculation logic (12). Wherein the time-to-digital conversion logic comprises correction algorithm logic (111) and carry chain logic (112). According to the invention, a single photon avalanche diode and a dark counting correction algorithm are combined to solve the problems of low precision, low linearity and short measurement distance of the existing distance measurement device. The method can be used for detecting the distance by laser, and can be widely applied to the fields of automotive electronics, medical treatment, industrial robots, national defense, aerospace and the like.

The invention provides a bridge stay cable video multi-target identification tracking and vibration extraction method based on an unmanned aerial vehicle. The method comprises the following steps: 1, constructing a bridge stay cable refined inclined slender target detection model based on a YOLOv11 model; 2, providing a multi-target tracking algorithm fusing the inclined slender target detection model and a StrongSORT algorithm; step 3, improving a displacement extraction method combining SIFT / ORB feature point matching and a sub-pixel refinement technology; 4, designing an unmanned aerial vehicle motion correction algorithm based on variational mode decomposition and time-frequency domain combined screening; and 5, constructing a joint working modal analysis algorithm for realizing combination of a natural excitation technology and a random subspace recognition algorithm. According to the method, high-precision extraction of the vibration signals of the stay cable and identification of modal parameters of the stay cable are realized, and technical support is provided for health monitoring of the large-span cable-stayed bridge.

The invention relates to the technical field of multiphase flow parameter detection, and particularly discloses a vertical pipeline gas-liquid-solid dilute phase annular flow solid content image detection method, which comprises the following steps: S1, carrying out multi-view image acquisition; s2, preprocessing the acquired image; s21, dynamic background differencing is carried out, and gas-liquid interface reflection noise is removed; s22, non-local mean filtering denoising is adopted, noise is removed, and detail information of the image is reserved; s23, carrying out local Otsu segmentation, and segmenting the particles from the background; s3, analyzing the preprocessed image; s31, performing spatial correlation verification, and eliminating multi-view repeated particles; s32, time correlation analysis is carried out, and a displacement vector of the particles is calculated; s33, correcting a displacement error by using an optical flow correction algorithm; and S4, carrying out particle data statistics, and calculating the solid content. According to the detection method, dynamic interference can be effectively resisted, a single-visual-angle blind area is eliminated, repeated particle counting is accurately eliminated, and the solid rate measurement precision is improved.

The invention discloses a method for creating and correcting a large model fine tuning data set, and relates to the technical field of large models. The method comprises the following steps: converting received structured and unstructured texts into structured content units in a unified format; generating and distilling an initial question and answer pair by using a teacher model taking a pre-trained large language model as a core; screening out high-quality question and answer pairs according to a quantitative evaluation system containing correlation, accuracy and integrity dimensions; performing diversified optimization on high-quality question and answer pairs by applying a generalization strategy and multi-round distillation, and performing dynamic regulation and control based on language feature difference to avoid content homogenization; and finally, guiding the teacher model to generate ternary structure data containing a thinking chain, and executing a verification and self-correction algorithm so as to generate a final fine tuning data set containing a verified reasoning process.