38results about How to "High recognition sensitivity" patented technology

The invention discloses an ultrasonic welding quality online detection system and method, and belongs to the technical field of laser non-contact measurement. According to the invention, the laser vibration measurement probe is rigidly coupled to the welding head to construct a follow-up-common-optical-path measurement unit, so that synchronous acquisition of vibration signals with zero mass loading is realized; further extracting an attenuation section after welding is finished from the signal, and inverting an equivalent modal damping ratio representing an interface energy dissipation state based on an exponential attenuation model; meanwhile, multi-dimensional physical indexes such as an energy transfer efficiency index Itrans and an input stability index Iin are calculated, and in combination with an energy absorption state index, collaborative fusion is carried out through a weighted fusion formula to obtain a comprehensive quality score; and finally, realizing quality attenuation early warning based on time sequence trend analysis of a scoring sequence. On-line sensing of the deep dynamic state in the welding process is achieved, the early defect recognition sensitivity and the system anti-interference capacity are remarkably improved, and the predictive maintenance capacity of quality degradation is achieved.

The invention provides a fire extinguisher tank coating process parameter collaborative optimization control method and system. The method comprises the following steps: acquiring infrared radiation data on the surface of a paint film of a fire extinguisher tank in a curing oven, and generating a temperature distribution matrix through thermal gradient analysis; quantifying temperature difference characteristics of different positions of the surface by using the matrix; a multispectral infrared thermal imager is annularly arranged in the curing oven, the optical scanning path of the multispectral infrared thermal imager is dynamically adjusted based on temperature difference characteristics, temperature abnormal points are marked, and a three-dimensional thermal field distortion atlas is constructed; analyzing temperature abnormal point distribution in the atlas, and generating a local cooling compensation vector; and converting the compensation vector into a heating unit power adjustment instruction by applying a sliding mode control mechanism, and adjusting output power to inhibit paint film deformation. Dynamic balance of the curing temperature of the paint film and temperature range compression are realized, and the deformation defect of the paint film is eliminated.

The application provides a cable insulation aging evaluation method based on high-voltage frequency domain dielectric response characteristic quantity, under the condition that the insulation electric field intensity is 0.5kV / mm-1kV / mm, an ultralow frequency sweep test signal is applied to the measured cable under three different voltages, and the dielectric loss tangent value frequency domain response curve corresponding to each voltage level is obtained; at least two dielectric characteristic quantities are calculated based on the frequency domain response curve, the dielectric characteristic quantities include a first characteristic quantity representing the discrete degree of the dielectric loss tangent value under different voltages, and a second characteristic quantity representing the frequency domain integral characteristic of the dielectric loss tangent value under the highest test voltage; based on a preset correlation evaluation model taking the cable insulation mechanical performance aging index as a benchmark, the insulation health state score of the measured cable is calculated by combining the dielectric characteristic quantities, and the insulation aging state is determined according to the health state score.

The application discloses a method for anti-electricity stealing based on abnormal connection of user single-phase electric meter, and relates to the technical field of electric power metering, which comprises the following steps: installing sampling units on the incoming line side and the outgoing line side of the user single-phase electric meter, synchronously collecting single-phase voltage, current, neutral line current and temperature and humidity signals of the meter box, and generating synchronous electrical data sequences through standardization and time stamp alignment. Single-phase voltage and current fundamental phase angles are calculated based on voltage and current instantaneous value sequences, and a phase angle difference map is constructed. The effective value of the neutral line current is calculated, and an abnormal flag sequence is generated in combination with a normal load reference range. The environmental temperature and humidity abnormal fluctuation period is extracted. The above-processed features are input into a connection mode recognition model to identify suspected electricity stealing connection mode categories, improve the accuracy of meter connection abnormality recognition, and reduce misjudgment. The method can effectively identify meter connection abnormalities and improve the accuracy of electricity stealing identification.

The application discloses a kind of detection methods based on engine cold test machine's cylinder bore scratch, it is related to engine detection field, comprising: acquisition measured engine under constant speed condition Torque data, extract the maximum, minimum and average of the torque of each cylinder;When torque data is out of limit, determine target cylinder, and calculate the difference between its maximum and minimum, obtain single cylinder torque span value;If the span value exceeds the determination limit, then trigger exhaust back pressure test link, acquisition target cylinder single cylinder exhaust peak pressure;If the peak pressure is lower than the peak pressure of normal cylinder, and difference reaches abnormal drop threshold, then determine that there is cylinder bore scratch.The present application effectively amplifies weak fault signal by combining torque span feature and exhaust back pressure feature, without disassembly, cylinder bore scratch and pure mechanical bearing defect can be accurately distinguished, and the sensitivity and accuracy of online detection of mass production line are improved.

The present application relates to the technical field of industrial vision detection, and discloses a green color printing surface defect detection method based on computer vision. The method comprises the following steps: collecting real-time image flow and performing multi-scale decomposition to generate a feature vector sequence; constructing a dynamic defect template set by using a generative adversarial network; encoding features by using a convolutional neural network, calculating a similarity matrix with the template set, and generating a defect matching index; adjusting the detection threshold value adaptively according to the defect matching index, and obtaining a defect probability distribution map through clustering analysis; identifying a defect aggregation area through space-time domain analysis, and starting a local re-inspection mechanism; updating a defect feature prototype library and optimizing network weights by using re-inspection data, and forming a self-learning closed loop; and finally performing full-process quality consistency inspection and outputting a report. The present application realizes dynamic adaptation and continuous optimization of the detection process.

The invention discloses an identification method and related equipment for rapidly identifying a multi-modulus seed magnetic island, and relates to the field of magnetic confinement fusion plasma diagnosis. Wherein at least one group is used for switching observed circumferential positions in different time sequence segments through a waveguide switch in the quasi-optical transmission module. The method comprises the following steps: synchronously acquiring electronic temperature signals of two groups of microwave measurement subsystems at the same radial channel in a time sequence segment corresponding to different annular phase differences, carrying out time averaging on the electronic temperature signals, and judging and identifying a seed magnetic island corresponding to an annular modulus based on the same radial temperature difference and normalization, and further extracting a flat region based on the distribution of the electron temperature along with the radial channel to determine the width and the boundary position of the magnetic island. According to the invention, the rapid distinguishing and high-precision positioning of a plurality of annular modulus seed magnetic islands can be realized without adding extra detection channels.

The invention discloses an electricity larceny prevention method based on abnormal metering wiring of a user single-phase ammeter, which relates to the technical field of electric power metering and comprises the following steps of: synchronously acquiring single-phase voltage, current, neutral current and meter box temperature and humidity signals by arranging sampling units on a wire inlet side and a wire outlet side of the user single-phase ammeter; a synchronized electrical data sequence is generated by standardized and timestamp alignment. Calculating a single-phase voltage and current fundamental wave phase angle based on the voltage and current instantaneous value sequence, and constructing a phase angle difference map; calculating a neutral current effective value and generating an abnormal mark sequence in combination with a normal load reference range; and extracting the abnormal fluctuation time period of the environment temperature and humidity. And inputting the processed features into a wiring mode identification model, and judging the type of a suspected electricity stealing wiring mode, thereby improving the metering wiring abnormity identification precision and reducing misjudgment. According to the method, metering wiring abnormity can be effectively identified, and the electricity larceny judgment accuracy is improved.

The invention provides a tumor identification method based on multi-modal information cooperation, and relates to the technical field of tumor identification. The method comprises the following steps: firstly, constructing a multi-modal cooperative characterization unit, synchronously receiving medical image and electrical impedance time sequence data, cooperatively mining macroscopic morphological association features and microscopic difference features of a tumor by using a multi-domain coupling characterization space technology, and outputting a multi-modal cooperative feature vector; then, an electrical impedance time sequence analysis unit is built, electrical impedance signal deep correlation features are extracted through time sequence feature multi-order analysis, and electrical impedance deep feature vectors are output; then designing a dynamic feature adaptation fusion unit, performing adaptive fusion and dimension recombination on the multi-modal collaborative feature vector and the electrical impedance deep feature vector based on a feature responsivity adaptation technology, and outputting a stabilized fusion feature vector; and finally, constructing a feature correlation judgment engine, quantifying the feature contribution degree based on a structural causal model and a multi-layer correlation verification technology, and outputting a tumor diagnosis result. The problems of insufficient multi-modal feature collaboration and explanatory deviation caused by black box mapping in the prior art are effectively solved, and the accuracy and logic robustness of tumor recognition are improved.

The application relates to the field of artificial intelligence and data fusion, and specifically discloses a kind of multi-modal data fusion method driven by graph neural network, which is cut into fixed time window after multi-modal observation data, and the time domain data is projected into frequency domain space by short-time Fourier transform and wavelet transform, so that the graph node represents the energy distribution of a specific frequency band rather than the sampling point of a discrete time point, avoiding the dilemma of inconsistent point-by-point sampling rate. In the frequency domain space, the graph topology structure of the same modal frequency adjacent correlation and the cross-modal resonance coupling correlation is constructed based on the physical priori, and the cross-band spectral graph attention mechanism is introduced to deeply fuse the multi-modal frequency domain features, so that the frequency domain features of different sampling rate modes can be directly interleaved and cooperated. The technical concept shifts from time domain microscopic alignment to frequency domain macroscopic wrapping paradigm transition, effectively eliminating the false features and gradient erasing problems caused by forced interpolation in the time domain, and improving the recognition sensitivity and diagnosis accuracy of early faults of industrial equipment.

The invention provides a detection system and a detection method for a decomposition product of perfluoroisobutyronitrile-nitrogen mixed gas, and belongs to the technical field of gas analysis and electric power insulation detection. The detection system for the decomposition product of the perfluoroisobutyronitrile-nitrogen mixed gas is provided with a sample introduction device, a first separation device, a second separation device, a flow path switching valve device and a detection device, and the first separation device comprises a first gas chromatographic column, a second switching valve and a third gas chromatographic column which are connected in series; the first separation device comprises a first gas chromatographic column and is used for separating a fluorocarbon compound in a gas sample, and the second separation device comprises a second gas chromatographic column and is used for separating the nitrile compound in the gas sample and realizing fine separation and quantitative detection of a nitrile decomposition product; the separated fluorocarbon compounds and nitrile compounds are accurately detected in the detection device, and the recognition accuracy and sensitivity of detection can be effectively improved.

This invention provides a power generation device based on the cold energy of a cryogenic propane tank, comprising a liquefied propane vaporizer, a propane heater, an intermediate medium vaporizer, a turbine generator set, and a dimethylamine circulating pump. The tube-side inlet of the liquefied propane vaporizer is connected to a cryogenic storage tank containing propane. The tube-side outlet of the liquefied propane vaporizer is connected to the tube-side inlet of the propane heater, and the tube-side outlet of the propane heater is connected to a propane pipeline network. Dimethylamine, after passing through the dimethylamine circulating pump, is connected to the tube-side inlet of the intermediate medium vaporizer. The tube-side outlet of the intermediate medium vaporizer is connected to the inlet of the turbine generator set, and the outlet of the turbine generator set is connected to the shell-side inlet of the liquefied propane vaporizer. The shell-side outlet of the liquefied propane vaporizer is connected to the inlet of the dimethylamine circulating pump. This invention achieves dual-sided cascaded energy recovery from cold to heat. Compared to traditional direct seawater heating schemes, the net power generation is increased.

The application provides a workbench production line control method and system based on environment state analysis, comprising the following steps: S1: arranging multiple environment sensors at each workbench, collecting multiple environment elements through the environment sensors, and uniformly modeling the collected environment data to obtain an environment data model; S2: performing feature extraction on the environment data model to extract state feature parameters corresponding to each environment element; S3: performing fusion calculation on the multiple environment elements according to the state feature parameters to obtain a fusion risk parameter representing the risk level of the current work station; S4: calculating the evolution trend of the risk level with time according to the environment data model and the fusion risk parameter to obtain a risk evolution result in a prediction period; S5: determining a corresponding risk grade according to the risk evolution result; and S6: triggering a corresponding control strategy according to the risk grade to control the running state of the workbench equipment, so that the safety performance of the workbench production line can be provided.

This invention relates to a method for patent retrieval and technology monitoring of multi-source heterogeneous data. It constructs a multimodal semantic representation including text, images, and time-series fragments, and generates a joint temporal representation vector by combining attention perturbation entropy and term evolution distance. A unified semantic space is established based on historical technology anchors, dynamically monitoring and quantifying semantic shifts to achieve real-time determination of drift trends. When substantial semantic drift occurs, local semantic correction is performed using gated contrastive learning and an anchor-aware adaptation layer to guide the orderly alignment of newly added modal features, ensuring the continuation and expansion of the original semantic topology. Simultaneously, an anchor decay and candidate new anchor generation mechanism is designed, combining cross-validation from patent citations, industry reports, and policy documents to achieve incremental expansion of the semantically stable anchor set. This scheme effectively improves the fusion accuracy, real-time adaptability, and dynamic topology update capability of multimodal patent data.

The invention belongs to the technical field of intelligent power grid terminal equipment, and particularly relates to an electric energy meter data acquisition method and system based on low-power-consumption task scheduling, and the method comprises the steps: firstly obtaining and normalizing a voltage sampling point sequence, adaptively adjusting the length of a sliding window according to data fluctuation, and outputting a normalized data sequence; traversing the sequence to calculate the physical gradient of the adjacent sampling points, introducing a weight amplification factor which is in nonlinear positive correlation with the physical gradient to construct a physical value entropy, and directionally amplifying the calculation weight of the high physical gradient data; then, a coupling evaluation logic containing an energy suppression item and a value guarantee item is constructed, a transmission decision index is calculated, and the transmission state of the communication module is controlled; and finally, counting a multi-time window decision index mean value, determining a next sampling frequency and completing a new round of acquisition. According to the method, the problem of missing detection of transient faults of the power grid is solved, and the cruising ability of equipment is effectively improved on the premise of guaranteeing zero loss of key fault data.

The present application relates to the technical field of centrifuge operation monitoring, in particular to a centrifuge data remote monitoring method and system based on a cloud platform. The method comprises: obtaining the vibration intensity, noise decibel value, temperature, actual speed of the screw, liquid flow at the separation liquid discharge outlet, and humidity of the material liquid in the cylindrical area and conical area of the centrifuge during the operation of the centrifuge; evaluating the centrifugal vibration intensity according to the vibration intensity and noise decibel value at each time; obtaining the abnormal factor at each time according to the distribution characteristics of the centrifugal vibration intensity in the local time period; correcting the abnormal factor by using the temperature and temperature change characteristics in the local time period; obtaining the separation abnormal score by combining the actual speed of the screw at each time, the liquid flow, the humidity difference of the material liquid in the cylindrical area and conical area of the drum, and the corrected abnormal factor; and determining whether to give a warning for the centrifuge abnormality according to the separation abnormal score. The present application improves the accuracy of the centrifuge monitoring result.

The application relates to the technical field of industrial equipment fault detection, and discloses a generator iron core stacking equipment fault detection method and system. The method comprises the following steps: acquiring an original vibration signal, synchronously calibrating the original vibration signal, and extracting a first vibration data set; performing time-frequency analysis and decomposition to obtain a second vibration data set; constructing a model to enhance a periodic weak impact signal, acquiring an enhanced signal feature set, and matching and quantifying a preset reference frequency template library to determine a potential fault signal distribution range; when the threshold value is exceeded, a fault feature data set is obtained through deep feature extraction, an hyperplane decision model is constructed based on the fault feature data set, a feature local evolution trend is analyzed, and a warning result is output. The method can accurately capture early fault signals, reduce the missed detection and misdiagnosis rate, and improve the equipment operation reliability.

This application relates to the field of stress optimization technology, specifically to a method for optimizing machining parameters to reduce residual stress in transmission system housings. The method includes: during high-speed cutting of the transmission system housing, performing dimensionality reduction processing on the multidimensional frequency domain features of cutting force signals, vibration signals, and temperature signals in each local time period; extracting independent potential fluctuation modes characterizing the dynamic changes of the thermal field; quantitatively characterizing the thermo-mechanical coupling response of the dynamic fluctuations of the thermal field; and quantifying the residual stress sensitivity value, which characterizes the degree of influence of the current machining state on residual stress. Using the cutting parameters as optimization variables and minimizing the predicted residual stress value, the machining parameters are optimized. During the solution process, the inertia weight of the optimization algorithm is dynamically adjusted based on the residual stress sensitivity value, and the optimal machining parameters for the transmission system housing are output. This application improves the optimization accuracy of the machining parameters for transmission system housings and effectively reduces the residual stress in the transmission system housing.

The invention provides a preparation method of a nanocapillary with an ion rectification effect and application of the nanocapillary, and particularly relates to application of the nanocapillary in improving the resolution ratio, the image quality and the scanning speed of a scanning ion conductance microscope (SICM). Ions are rectified and modified on the nano capillary to improve the resolution of the SICM, and meanwhile, the image quality and the scanning speed are improved well.

This invention relates to the field of aluminum formwork inspection technology, solving the problem of insufficient accuracy and precision in the inspection and sorting of high-precision aluminum formwork for aluminum buildings due to the difficulty in comprehensively covering complex surfaces and nodes. Specifically, it discloses an inspection and sorting device and method suitable for high-precision aluminum formwork. The device includes a feeding input mechanism, an inspection and sorting mechanism, an output mechanism, and a temporary storage mechanism. The feeding input mechanism is located between the output mechanism and the temporary storage mechanism, and the inspection and sorting mechanism is located above the feeding input mechanism, output mechanism, and temporary storage mechanism. The method includes S100, feeding positioning and preprocessing; S200, multi-dimensional inspection data acquisition; S300, algorithm inspection data processing; S400, sorting execution; and S500, inspection data recording and scheduling. This invention is used for the inspection and sorting of high-precision aluminum formwork in aluminum building structures, improving the inspection effect by increasing the inspection coverage and accuracy of complex surfaces.

The application discloses a kind of transaction flow data exception detection method and system, it is related to big data analysis technical field.The method includes: based on transaction flow data, the operation behavior data of target user in transaction process and the interactive behavior data after transaction completion are collected, and user operation behavior sequence and user interactive behavior sequence are formed by time stamp sorting;The sequence is input into the initial risk coefficient output by pre-training exception risk prediction model, and the model integrates individual behavior baseline and adaptive weight module;The group behavior baseline of similar user group is obtained, and the final risk coefficient is obtained by correcting initial risk coefficient in combination with the matching result of sequence and group baseline;Based on preset risk threshold, final risk coefficient is executed graded exception early warning operation.This application improves the accuracy and timeliness of transaction flow data exception detection by whole-process behavior sequence analysis, individual and group baseline double calibration and dynamic weight adjustment.

The present application relates to a unit multi-physical field coupling monitoring system based on a Relief-FCC algorithm, and belongs to the technical field of electric digital signal processing, aiming to solve the problems of ignoring multi-field coupling effect, high fault omission rate and extensive control strategy of existing pumped storage unit monitoring. The system collects unit electromagnetic field, temperature field and mechanical vibration field operation data through a sensor network, introduces an inter-field coupling coefficient FCC in the Relief feature weight to construct a Relief-FCC algorithm, and selects dominant coupling features; the inter-field coupling coefficient is used to complete multi-field coupling consistency cross-validation, distinguish unit body and sensor abnormalities, and complete adaptive correction; based on the dominant coupling features, fault identification, health assessment and failure root cause analysis are realized, and dynamic optimization control of the unit is simultaneously realized. The present application significantly improves the fault prediction accuracy of pumped storage units and ensures efficient and safe operation of the units.

The application discloses a fault identification method, device and equipment of a flexible alternating current (AC) feeder, and a medium, and belongs to the field of power system protection. The method is as follows: equivalent circuits of a flexible direct current (DC) system in different modes are topologically connected according to boundary conditions of single-phase ground faults to construct a multi-mode equivalent network model; a fault excitation is applied at a preset fault point of the multi-mode equivalent network model, a theoretical current response of a preset protection installation under the fault excitation is calculated, and theoretical transient characteristic values are extracted from the theoretical current response; electrical signals at the preset protection installation are collected in real time, when it is detected that the electrical signals meet preset starting conditions, an actual current response of the preset protection installation is acquired, and actual transient characteristic values are extracted from the actual current response; the actual transient characteristic values are compared with the theoretical transient characteristic values, and the fault of the flexible AC feeder is identified according to a comparison result. The application can improve the accuracy of fault identification of the flexible AC feeder.

The utility model belongs to the technical field of lithium ion batteries, and particularly relates to an auxiliary device for assisting in observing the state of a single diaphragm of a lithium battery, which comprises a bottom fixing plate, a diaphragm pressing mechanism and an LED (light-emitting diode) backlight plate, threaded holes are formed in the two sides of the bottom fixing plate and used for fixing the diaphragm pressing mechanism. The diaphragm pressing mechanism comprises a pneumatic pressing device, a cylinder fixing plate, a vertical plate, a pressing plate and a buffer layer; the vertical plate is connected with the bottom fixing plate through bolts, the air cylinder fixing plate is installed on the outer side of the vertical plate, the pneumatic pressing device is vertically fixed to the air cylinder fixing plate, the pressing plate is in threaded connection with a piston rod of the pneumatic pressing device, and the buffer layer is attached to the bottom face of the pressing plate. And the LED backlight plate is embedded into the groove of the bottom fixing plate. According to the utility model, through the collaborative design of the diaphragm pressing mechanism and the LED backlight plate, the detection precision and stability of the single diaphragm of the lithium battery are obviously improved.

The present application relates to the technical field of milling control system, and more particularly to a kind of EIGA milling collaborative control method and system based on multi-source data fusion, wherein the EIGA milling collaborative control method based on multi-source data fusion comprises: obtaining the sensor monitoring data of each sensor in EIGA milling process;For current process, extract the change characteristics of each sensor monitoring value;Determine the process fluctuation resistance factor of current process;Obtain the change information of upstream process key parameters, determine the production adaptation degree of current process;Determine the collaborative control compensation, and optimize the control parameter of current process based on collaborative control compensation.The present application extracts the change characteristics of each sensor monitoring value to represent the data fluctuation state under different sampling rates, eliminates the time dislocation between each dimension inside the feature vector, so that control decision can be based on the data reflecting the current real milling state, and improves the accuracy and reliability of collaborative control.

The invention relates to the technical field of oil product detection, in particular to a diesel oil Raman spectrum detection method and device based on a convolutional neural network, computer equipment and a computer program product. The method provided by the invention comprises the following steps: acquiring a Raman spectrum of diesel oil to be detected; the Raman spectrum is input into a pre-trained diesel oil detection model, a detection result output by the model is obtained, and the detection result represents the alcohol substance doping possibility of the to-be-detected diesel oil; wherein the diesel oil detection model comprises a multi-scale convolution branch structure, and the multi-scale convolution branch structure comprises three parallel convolution branches with different convolution kernel sizes. According to the method provided by the invention, through parallel execution of three convolution branches, feature information of various scales in the spectrum can be synchronously extracted, the recognition and extraction capability of the model on weak signal features is remarkably enhanced, and efficient and accurate recognition of alcohol doping behaviors in diesel oil is realized.

The application discloses a kind of based on multi-source inspection data fusion and correlation analysis water conservancy project management method and system, and relates to data processing technical field.The application is by layered coding to inspection data in uniform collection cycle and constructs three-layer envelope structure, realizes the spatio-temporal integration organization of multi-source inspection information, makes cross-source data obtain consistent expression, improves the accuracy and traceability of data fusion;Through the establishment of joint positioning point in time envelope layer and the allocation of collaborative labeling mark in space envelope layer, different source data form collaborative labeling items with synchronous semantics, thereby significantly improving the alignment capability and comparative analysis efficiency of inspection information;With the construction of step-by-step linkage correlation link, the continuous evolution of water conservancy structure state can be presented on time series, and the spatio-temporal characteristics can be captured synchronously by combining dynamic correlation paragraph generation, to enhance the identification sensitivity of abnormal change.

The invention relates to the technical field of sleep staging, in particular to a brain wave-based sleep staging model training method and system. Comprising the following steps: acquiring a sleep label sequence, identifying a switching position, clustering and constructing a conversion section, extracting signal analysis amplitude and rhythm, screening mutation fragments, extracting standard training fragments, organizing a tensor sequence, establishing a numbering relationship, dividing training batches, and generating a task list. According to the method, by recognizing the change points of the electroencephalogram tag and clustering the change points, a dense region of stage conversion can be positioned, the recognition of a critical state is enhanced, the trend judgment of the interval between the rhythm amplitude and the peak value is combined, stable and representative signals are screened out from the change, the sample effectiveness is improved, and unified processing of channels and duration is carried out on signal segments; standardization of input data is guaranteed, the time sequence capturing capacity of the model is enhanced in a serial number arrangement and sequence combination mode, data and labels are ensured to be consistent through unified division rules, and classification accuracy and generalization capacity are improved.

This invention belongs to the field of power line fault detection technology and discloses a method, system, equipment, and medium for diagnosing power distribution line faults. The method includes: acquiring initial fault signal data corresponding to the power distribution line, preprocessing the initial fault signal data, and outputting fault signal data; decomposing the fault signal data using joint decomposition technology to obtain intrinsic mode components (IMCs); introducing information mining technology to extract the distribution change feature vector of the fault signal and fusing it with the IMCs to obtain a change image; extracting local features of the change image through a convolutional neural network and combining it with a strategy function optimized by a Huffman tree for fault classification, and outputting the fault diagnosis result. This invention applies the Softmax strategy of Huffman trees, improving the efficiency and accuracy of judgment, enabling early identification and warning of fault signals in power distribution lines, and improving the efficiency of power distribution line fault diagnosis.

The present invention specifically relates to a method and system for detecting the surface quality of electronic components based on machine vision, belonging to the technical field of electronic component quality detection, including: based on the objective functions of minimum reconstruction error and the sparsest coefficients, learning basis functions through an iterative optimization algorithm, and obtaining a set of dedicated basis functions adapted to normal textures by combining discrimination screening. In the present invention, a database of high-purity qualified texture blocks is constructed to provide standardized training data for the learning of basis functions. The high-quality training data supports the optimization of dedicated basis functions, enabling the basis functions to accurately match the normal textures of components, improving the recognition sensitivity to subtle defects such as cracks and pits, laying a foundation for subsequent detection accuracy, and meeting the high requirements for data reliability in industrial batch detection.