74 results about "Micro defects" patented technology

The invention relates to the technical field of composite material detection, in particular to a fiber-reinforced thermosetting thermoplastic composite structure interface quality detection method and system.The method comprises the steps that vibration excitation is applied to a target area of the surface of a to-be-detected sample, and the amplitude of the excitation is lower than the damage critical value of a weak bonding interface; a dynamic response signal perpendicular to the interface is collected in a vibration field non-diffusion area of the excitation point of the target area; performing frequency domain transformation on the dynamic response signal, and separating a fundamental frequency component and at least one high-order harmonic frequency component; according to the nonlinear enhancement effect of the high-order harmonic frequency component relative to the fundamental wave frequency component, whether the weak bonding defect exists in the target area or not is judged. According to the method, the technical bottleneck that the dynamic evolution process of the weak bonding defect of the composite material interface is difficult to capture in real time is effectively solved, sensitive response to the microdefect is realized through a nonlinear harmonic characteristic enhancement effect, and the recognition precision of traditional ultrasonic detection is improved.

The invention relates to the technical field of high-end equipment, and discloses a real-time micro-defect intelligent control method and system for precision metal parts, and the method comprises the steps: obtaining historical defect detection data, and carrying out the preprocessing of the data to obtain a structured defect data set; determining a defect statistical mode through time sequence analysis and clustering analysis; generating a geometric feature set, a statistical feature set and an associated data set based on the defect statistical mode, and storing by adopting a dynamic retrieval database to form a queriable defect feature library; new defects are detected and matched with the feature library to judge types and potential positions; predicting time points and position coordinates of defects in combination with process parameters; adjusting process parameters according to a prediction result to obtain optimal configuration; optimizing and updating the defect feature library in combination with new defect information; and continuously monitoring and dynamically adjusting process parameters based on the improved feature library, and generating stable production control parameters. According to the method, real-time prediction and dynamic control of microdefects in the precise metal part production process can be achieved, and the product yield is increased.

The invention discloses a preparation method of a MgO-MgA12O4-ZrO2 nano ceramic eutectic composite material. The preparation method comprises the following steps: S1, preparing powder; s2, flash firing preparation; and S3, high-temperature melting preparation. According to the invention, two technologies of flash sintering and high-temperature melting are combined, the problems of high sintering temperature, easy growth of crystal grains, difficult densification and the like of the traditional eutectic ceramic are solved, and an excellent composite material with a nanoscale eutectic structure is finally obtained; a nanocrystalline blank obtained through flash sintering is crushed and then subjected to high-temperature melting, microdefects and air holes possibly existing in the flash sintering blank are eliminated, the chemical components and the microstructure of a final product are more uniform, and compared with traditional selective laser melting, the method is more convenient and faster to implement and lower in cost.

The invention belongs to the technical field of PCB detection, particularly relates to a PCB ceramic substrate appearance laser measuring instrument, and aims to solve the problem that tiny cracks hidden on a substrate are difficult to find when the substrate is not stressed in the actual detection process, the following scheme is provided: the PCB ceramic substrate appearance laser measuring instrument comprises a counterweight seat, support rods are fixed on the upper surface of the counterweight seat close to four corners, and the support rods are fixed on the upper surface of the counterweight seat; the top ends of the four supporting rods are fixedly provided with the same workbench, the middle of the workbench is rotationally connected with a vertical supporting pipe, the top end of the supporting pipe is fixedly provided with an arch-shaped supporting plate, and the two ends of the arch-shaped supporting plate are provided with bearing modules which are of the same structure and are symmetrical to each other respectively. According to the invention, the ceramic plate body can be extruded before surface defect detection, so that the residual stress of the ceramic plate body can be amplified to a certain extent, some potential surface micro-defects are forced to be amplified, a laser probe can find the surface micro-defects more easily, and potential quality defects of the ceramic plate body can be detected.

The invention relates to the technical field of processing, testing or inspection of aircraft parts, and provides an accelerated test method for aging of an aluminum alloy aircraft plate. The method comprises the following steps: carrying out a cyclic alternating test of ultraviolet irradiation and solution soaking on an aluminum alloy sample with an epoxy coating, and simulating a multi-factor coupling aging process in an aviation service environment. And regularly observing and recording micro-defects such as micropores, cracks and spalling on the surface of the coating in a test period, and evaluating the development of the micro-defects of the coating through qualitative and quantitative analysis. And after the test is finished, measuring the residual thickness of the sample, and establishing a relation model between the coating thickness degradation and the change of microdefect development along with the aging time in combination with the microdefect development state. According to the method, the actual aviation service environment can be simulated, the ultraviolet aging and corrosion coupling effect is considered, and accurate prediction and accelerated simulation of the aging behavior of the aluminum alloy aircraft plate with the epoxy coating are achieved.

The invention discloses an aluminum profile surface defect detection method and a storage medium, and the method comprises the following steps: S1, image preprocessing and data set construction; s2, improving the design of a YOLOv8 network structure; s3, model training and deployment; the image quality is optimized through wavelet transform denoising and histogram equalization, and the feature identification degree of the tiny defects is enhanced; a C2fFaster DCNv3 module and an ADFR module are adopted to enhance the multi-scale feature extraction and fusion capability; the newly added P2 layer detection head is specially adapted to tiny defects; the focus degree of a difficult sample is increased through weighting of a Focaler-IoU loss function, and the problems of missing detection and false detection of tiny defects are effectively solved; the Faster Block branch reduces the calculated amount, the TensorRT reasoning accelerates and optimizes the deployment efficiency, the real-time detection requirement of industrial production is met on the premise that the detection precision is guaranteed, and the dual purposes of high precision and high speed are achieved.

The invention relates to the field of precision optical detection, and discloses a device and a method for automatically identifying microcracks on the surface of quartz glass based on machine vision, and the device comprises a multi-dimensional image sensing module, a main control module and an AI identification module. The method utilizes a polarized light orthogonal imaging principle to adjust a polarization angle to be orthogonal with an inspection angle so as to suppress specular reflection interference on the surface of the quartz glass. An acquired image is enhanced by adaptive Gaussian filtering and then is input into an improved U-Net model integrated with a fine crack attention mechanism, and the mechanism combines channel and space attention to accurately capture micro crack features while suppressing noise. The system performs fitting on a crack skeleton based on a least square method so as to calculate geometric parameters, and intelligent retrieval and clustering of defects are realized by using cosine similarity and a K-means algorithm. The method effectively solves the problem that the surface microdefect of the transparent high-reflective material is difficult to detect, the detection precision and the automation level are remarkably improved, and the method is suitable for quality control of precise optical elements.

The invention relates to a novel small-sized receiving coil group electromagnetic ultrasonic sensor and a method for high-resolution detection of submillimeter-level tiny defects. The sensor comprises a permanent magnet, an exciting coil and a receiving coil group, the receiving coil group comprises at least two independent preset small coils; the receiving coil group is arranged above the effective receiving area of the exciting coil; the permanent magnet is placed above the effective receiving area of the receiving coil assembly, and the magnetic pole direction is perpendicular to the plane of the receiving coil assembly. The invention aims to solve the technical problems that an existing electromagnetic ultrasonic sensor is low in signal-to-noise ratio, depends on complex signal processing and is difficult to give consideration to high-resolution and high-reliability detection when detecting submillimeter-level tiny defects, and provides a small-sized receiving coil group electromagnetic ultrasonic sensor which is compact in structure and high in signal-to-noise ratio and a detection method. Under the condition of not depending on complex post-processing, high-resolution recognition of submillimeter-level tiny defects in the metal material can be achieved.

The application discloses an industrial product micro defect visual detection system and method, belongs to the product detection technical field, obtains the multi-source image data of the target industrial product under multiple detection angles and illumination conditions, and establishes an image information matrix; the image is subjected to region segmentation and texture enhancement, and a local texture direction inconsistency degree parameter is extracted; the pixel ratio is normalized and analyzed under different spectral channels, and a multispectral reflectance anomaly index is calculated; a deep convolution recognition model is constructed; the model is used for reasoning on the image, and a defect judgment result and a confidence score are output; whether it is a defect area is determined based on a dynamic threshold mechanism, and a detection report containing defect position information and a visualized heat map is output; the application realizes multi-dimensional fusion recognition of texture structure disturbance and spectral response anomaly, effectively improves micro defect recognition precision, has high robustness, automation and engineering practicability, and is suitable for high-precision quality control requirements in various industrial scenes.

The invention discloses an industrial aluminum material forming device and a forming method thereof, and belongs to the technical field of aluminum material extrusion forming. The device comprises a forming machine body, two side plates are fixedly connected to the top of the forming machine body, connecting plates are arranged at the tops of the two side plates, a pre-impurity-removing mechanism is jointly arranged between the two connecting plates, and an extrusion assembly, a forming assembly, a shearing assembly and a temperature control assembly are fixedly installed at the top of the forming machine body; when the extrusion assembly pushes the aluminum bar to penetrate through the mounting ring, the servo motor drives the gear to rotate, and the annular gear ring drives the mounting base and the first grinding brush to rotate in the circumferential direction through meshing. The mounting seat drives the first grinding brush to radially move under the action of the inclined surface of the guide plate and is always attached to the surface of the aluminum bar by matching with the synchronous adaptation mechanism. Under rotation and self-adaptive pressing, the first grinding brush evenly grinds the outer wall of the aluminum bar, oxide skin, oil stains and microdefects are effectively removed, a clean metal surface is obtained, and a high-quality blank is provided for follow-up extrusion forming.

The present application relates to a kind of electrochromic film defect analysis method and device, equipment, medium, it is related to electrochromic field, including: sample is carried out in darkroom and is marked defect by fading observation, obtain marked sample;The defect in the obtained marked sample is observed using optical microscope and determines defect category, obtain marked sample;The defect of marked sample is profiled, microanalysis, component analysis, obtains defect profile parameter, defect microtopography, defect component;The nature of defect is judged based on at least one of defect profile parameter, defect microtopography and defect component.The defect analysis method provided in the present application, by optimizing the defect analysis process for photochromic film, realize system, efficiently, accurately to the surface of electrochromic film Complex and varied macro, micro defect (such as scratch, halo point, pinhole-like white spot, white spot, blue dot, abnormal shape and color point etc.) Positioning, identification, characterization and root cause analysis.

The present application relates to the technical field of industrial vision detection, and discloses a hydraulic side plate surface micro defect identification method and system, the method comprising collecting a double light source image sequence, performing pixel level gray difference extraction, purifying defect feature maps, performing multi-scale Gaussian Laplace convolution and morphological screening to construct defect feature partial order lattices, calculating gray and spatial reference offset vectors to generate a composite drift correction operator to perform adaptive correction on the feature maps, inputting a convolutional neural network to determine the defect category and trigger a sorting instruction; the present application separates oil film artifacts by using optical response difference and quantifies imaging drift by combining partial order lattice structure, solves the problem of recognition precision decay caused by cutting fluid reflection interference, long-term operation light decay and micro vibration of the equipment, and realizes high-precision detection of micro defects under production line conditions without calibration.

The application discloses a kind of based on multi-mode spectral tomography technology's light-reflecting steel pipe weld defect identification method, propose a kind of multi-band reflection data acquisition, denoising and normalization, multi-region feature distribution modeling, sparse reconstruction separates abnormal response, multi-scale contrast weighted enhancement, saliency layered imaging and defect classification discriminant detection process.The method is by automatic partition, environment and material dynamic acquisition, adaptive parameter optimization and so on Measure, realize effectively suppress high light interference, significantly enhance the resolution of defect and background, improve the detection sensitivity and positioning accuracy of small and micro defects.

The invention relates to a homogenizing composite pouring method of a high-beryllium aluminum alloy, which adopts a beryllium-based alloy as a main material, and through a double-flow smelting process, the main material is smelted in a flow A, and auxiliary materials such as an aluminum ingot and a Y-TiC-TiB2 / Al composite modifier are smelted in a flow B, so that the cooperation of interface modification and grain refinement is realized. And then, the A-flow melt and the B-flow melt are synchronously and compositely poured into a preheated casting mold according to the proportion, rapid solidification is achieved through extrusion casting, and finally hot isostatic pressing treatment is conducted. According to the method, beryllium-aluminum segregation, beryllium phase sedimentation and formation of microdefects are effectively inhibited, and the prepared beryllium-aluminum alloy is uniform in component, compact in structure, excellent in mechanical property and wider in application range.

The application provides a method and system for detecting PV and PI regions of a silicon wafer, and belongs to the technical field of defect detection of silicon wafers. The method for detecting the PV and PI regions of the silicon wafer specifically comprises the following steps: firstly, performing copper decoration treatment on the surface of a silicon wafer sample to mark vacancy-type defects of the silicon wafer sample, so as to obtain a to-be-detected silicon wafer sample; then, immersing the to-be-detected silicon wafer sample, a counter electrode and a reference electrode in a hydrofluoric acid electrolyte, so that the to-be-detected silicon wafer sample serves as a working electrode and the working electrode, the counter electrode and the reference electrode jointly construct a three-electrode electrochemical system; subsequently, applying a constant anodic bias to the working electrode and keeping the constant anodic bias for a predetermined time, so that the marked vacancy-type defect region and the unmarked gap-type defect region are subjected to differential corrosion, the PV and PI regions are exposed, and the originally invisible micro defects are converted into macro and micro morphology differences which are clearly distinguishable.

A kind of detection method of high-precision flexible plate for intelligent networked vehicle, according to design data, flexible core plate is formed by processing, and the flexible core plate includes pad;Then the first socket is welded to the pad, forming a welded plate;Form a test adapter plate including rigid area and flexible area;The second socket is provided in the flexible area, and the test pad is provided in the rigid area;Second socket and test pad are electrically connected;The test adapter plate and the welded plate are connected, including the second socket is inserted into the first socket, then the test pad is tested by using test machine, and the detection is completed;Through welding the first socket, the test adapter plate containing the second socket is manufactured, a stable current loop and a voltage test loop are established, the identification accuracy of the line gap, micro short circuit, line young and other micro defects of the line is improved, the transmission stability of the line is guaranteed, and the problems such as traditional low resistance test due to high pad degree, easy to appear probe contact, miscontact or poor contact and other problems are effectively avoided.

This invention discloses a visual inspection system for surface defects in industrial parts, relating to the field of parts quality control technology. The system adopts a four-layer architecture: a thermoelectric coupling field control layer, a texture-encoded defect development and acquisition layer, a cross-scale linkage modeling layer, and a process closed-loop control layer. The core components include a field-induced texture-encoded defect development unit, a texture-encoded defect cause mapping unit, a cross-scale closed-loop control unit, and a multi-dimensional collaborative verification unit. This invention utilizes the core technologies of active thermoelectric coupling field control + reuse of processing texture process codes, field-induced development of micro-defects, and cross-scale closed-loop control to solve the problems of existing detection methods, such as the inability to trace the cause of defects, high false negative rates for nanoscale micro-defects, disconnect between detection and process optimization, high system costs, and dependence on constant temperature environments. It achieves simultaneous and accurate completion of defect detection and cause tracing, significantly reduces the false negative rate of micro-defects, enables closed-loop control of the entire detection-process optimization process, reduces system costs, and adapts to non-constant temperature industrial scenarios.

The invention provides an intelligent identification system and method for micro defects of a cladding layer on the inner wall of a large metal pipeline based on an ultrasonic phased array, and relates to the technical field of energy storage power stations. In order to solve the problems that a cladding layer (with the thickness of 0.5-1.0 mm) on the inner wall of an injection-production pipe of a compressed air energy storage power station easily falls off to form micro defects (with the diameter smaller than or equal to 1 mm) after long-term service, traditional ultrasonic detection is greatly interfered by roughness, and the signal discrimination degree is poor, an ultrasonic phased array probe is used for scanning and collecting signals along the outer wall of a pipeline, and after wavelet noise reduction and Fourier time-frequency conversion, the cladding layer (with the thickness of 0.5-1.0 mm) on the inner wall of the injection-production pipe of the compressed air energy storage power station is obtained. And extracting a voiceprint feature vector by using a convolutional neural network, identifying the defect size in combination with a dynamic matching algorithm, and outputting a defect image through full-focus imaging. According to the method, the advantages of high precision of the ultrasonic phased array and AI intelligent identification are integrated, the micro cladding layer falling defect can be rapidly and accurately identified, the anti-interference capability is high, and a reliable detection means is provided for safe operation of an injection-production pipe.

The application provides a kind of metal material defect density measurement method, it is related to the technical field of metal material defect density quantitative analysis regulation material performance.This method includes six steps of establishing original data set, preprocessing original data, judging grain boundary existence, calculating elastic distortion tensor, calculating dislocation density and calculating dislocation density.Compared with the traditional microdefect organization analysis method, this method first introduces Lie algebra mathematical tool to quantitatively analyze the rotation characteristics and structure of dislocation such as Frank vector, breaks through the existing material defect density measurement method and the difficulty in accurately describing and calculating the rotation vector, proposes a differential calculation method for the rotation vector, and quantifies the microdefects such as dislocation and dislocation by combining the EBSD data of the metal material to be measured, which provides an important reference for analyzing the influence of crystal defects such as dislocation on material performance.

The application relates to the technical field of steel fatigue resistance, and discloses a high-hardness steel fatigue resistance system and method based on ultrasonic and shot blasting composite reinforcement. The method comprises the following steps: collecting macroscopic geometric appearance, microscopic structure and dynamic stress and strain data of a target steel piece through a multi-modal sensing system, and fusing and constructing a multi-physical field digital twin model; based on the fatigue damage feature distribution in the model, a reinforcement decision tree model is sorted, and a priority supercritical value triggers parameter configuration; through a self-adaptive neural network algorithm, ultrasonic field regulation parameters and shot blasting trajectory planning parameters are synchronously calculated; the parameters are injected into the digital twin model to simulate material response and energy transmission, and data such as microscopic defect closure state is output, and parameter correction signals are generated through multidimensional consistency inspection. The method realizes accurate regulation and control of composite reinforcement, and effectively improves the fatigue resistance of high-hardness steel.

The invention discloses a method for local micro-corrosion of an equiaxial high-temperature alloy part before fluorescent penetrant detection. According to the method, after the local part of the equiaxial high-temperature alloy part is subjected to micro-corrosion, a false plugging layer of the part is removed, and an opening with a tiny defect is opened, so that the open defects of all the surfaces can be detected through subsequent penetration detection; meanwhile, unnecessary corrosion is reduced, the surface of a part which does not need to be corroded is protected, the micro-corrosion efficiency is improved, the micro-corrosion cost before penetration detection is reduced, the problem that in the fluorescent penetration detection process of the equiaxial high-temperature alloy workpiece, due to a surface false plugging layer, missed detection of micro defects is caused is solved, and the defects existing in a traditional overall corrosion method are overcome.

A detection method of a high-precision flexible board for an intelligent networked automobile comprises the steps that processing is carried out according to design data to form a flexible core board, and the flexible core board comprises a bonding pad; welding a first socket on the bonding pad to form a welding plate; manufacturing and forming a test adapter plate comprising a rigid area and a flexible area; the flexible area is provided with a second socket, and the rigid area is provided with a test bonding pad; the second socket is electrically connected with the test bonding pad; conducting connection is carried out on the test adapter plate and the welding plate, specifically, the second socket is correspondingly inserted into the first socket, then a test machine is used for carrying out electrical test on the test bonding pad, and detection is completed; by welding a first socket, manufacturing a test adapter plate comprising a second socket and establishing a stable current loop and a voltage test loop, the identification precision of line gaps, micro short circuits, line gaps and other micro defects of lines is improved, the transmission stability of the lines is guaranteed, and the problem that a traditional low-resistance test is poor in reliability due to high bonding pad degree is effectively solved. And the problems of contact failure, wrong contact or poor contact and the like of the contact pin are easily caused.

The present application belongs to the field of integrated circuit package testing, and particularly relates to a TGV defect detection method based on PSO-XGBoost. A defect model is established by taking the cavity, open circuit and composite defects in TGV as the research object, and the S parameters of different defects are extracted based on the established defect model; the TGV cavity defect radius, TGV open circuit defect length and TGV composite defect combination form are taken as the data classification basis, and the labels are set; a PSO-XGBoost classification model is established, the particle swarm optimization (PSO) algorithm is used to optimize the extreme gradient boosting (XGBoost), and the best hyperparameter is found based on the PSO-XGBoost algorithm; the S parameters of different defects are classified and processed by using the PSO-XGBoost model, the defect type is predicted, and the TGV defect detection accuracy is improved. The detection method avoids the secondary damage of TGV in the defect detection process, solves the problems that the traditional TGV detection method is difficult to accurately detect the micro defect and composite defect and has low detection efficiency, and has certain reference value for optimizing the design and manufacturing process of TGV, improving the TGV defect and improving the three-dimensional package yield.

1. The name of the design product: wafer macro and micro defect detection equipment. 2. The use of the design product: for wafer defect detection. 3. The design points of the design product: in shape. 4. The picture or photo that best shows the design points: perspective view 1.

The invention relates to the technical field of biomimetic materials, polymer composite materials and polymer fibers, and discloses a preparation method of biomimetic assembled high-performance artificial spider silk fibers, which comprises the following steps: preparing primary gel fibers containing rigid network nodes and flexible polymer chains; stretching the primary gel fiber in a closed environment in which the temperature and the solvent vapor pressure are controlled at the same time; according to the environment, solvent molecules preferentially swell the flexible polymer chain, the rigid network nodes are kept stable, and the applied drawing force drives the flexible polymer chain to slide and axially orientate among the rigid network nodes. According to the method, the specific stretching environment is constructed, internal physical conflicts of stretching gain and structural damage are remodeled into a synergistic process, the situation that microdefects are generated at a rigid-flexible interface by stretching force is avoided, and a bionic rigid-flexible composite structure which is difficult to achieve in the prior art and has high integrity is obtained for final fibers.

The application discloses a waterproof performance detection device for leather lining fabric and belongs to the technical field of fabric waterproof detection. The bottom of the fabric is attached to the pressing ring in the pressing assembly, the recessed part of the fabric is deepened, the purpose of fabric stretching is achieved, the driving mechanism drives the swing top material assembly to work, and then drives the multiple pressing heads at the bottom to regularly reciprocate up and down. Meanwhile, the second spring of the pressing assembly can be self-adaptingly stretched and contracted according to the reciprocating force of the pressing head, the pressing ring is always closely attached to the fabric, the fabric is prevented from being displaced in the dynamic movement process, the fabric detection area is partially immersed in the detection medium, the wet sensitive sensor continuously works, whether there is moisture penetration in the fabric is monitored in real time, the pressing head is used in cooperation with the pressing ring to simulate the repeated bending, extruding and pressing actions in the actual use of the fabric, the real stress scene of the attached fabric is simulated, the detection error is reduced, and the micro defects of the waterproof structure of the fabric can be accurately identified.

The application discloses a preparation method of an ultra-smooth organic film layer. The film coating is carried out on the surface of inorganic film layers such as silicon oxide and aluminum oxide, and the process comprises three process procedures of organic matter flashing, organic matter flowing and organic film layer polymerization. The organic film layer prepared by the method has the following effects: on the one hand, the inorganic film layer is planarized, and the roughness is reduced from 20 nm to within 3 nm; and on the other hand, the organic layer can compensate for microscopic defects such as pinholes and micro-cracks, prolong the path of gas molecules, and can prepare a high-barrier thin film. The organic diaphragm prepared by the application is obtained by flashing through the winding method on the surface of the oxide film layer, the thin film product has the ultra-smooth characteristic, the roughness is less than 3 nm by using an atomic force microscope, and can be applied in the fields of quantum dot display, photovoltaic, photoelectric material packaging and barrier film.

The embodiment of the invention provides a silicon wafer microdefect testing method and system. Comprises: acquiring a silicon wafer surface image; preprocessing the acquired surface image of the silicon wafer, extracting defect data, and performing feature analysis on the defect data; the method comprises the following steps: analyzing silicon wafer surface defects to obtain a defect surface area distribution monitoring index, a defect density integration index and a defect position deviation index; comprehensively analyzing the quality of the silicon wafer according to the analysis result of the defects to obtain a comprehensive evaluation coefficient of the quality of the silicon wafer, and evaluating the quality of the silicon wafer through the comprehensive evaluation coefficient of the quality of the silicon wafer; judging whether the quality of the silicon wafer is qualified or not according to a preset judgment standard; and generating unqualified prompt information for the silicon wafers with unqualified judgment results, generating a detection report, and sending the detection report and the unqualified prompt information to a tester terminal. The micro defects on the silicon wafer can be accurately captured, the requirement for high-precision detection is met, and automatic detection and defect recognition can be achieved easily.

The invention discloses a wafer defect detection method and system based on high-frequency perception and optical flow alignment, and relates to the technical field of semiconductor manufacturing. The method comprises the following steps: extracting multi-scale features of a reference image and a query image; performing frequency domain high-pass filtering and attention enhancement on each scale feature through a high-frequency sensing module so as to highlight high-frequency information of tiny defects; fusing multi-scale features by using a feature pyramid; correcting infinitesimal displacement between the images through an optical flow alignment module in a feature space to obtain an alignment feature; and finally, calculating the aligned feature difference and segmenting a defect region. According to the invention, the feature expression of tiny defects is effectively enhanced through the high-frequency sensing module, and the problem of missing detection of small targets is solved; the displacement interference is eliminated through the optical flow alignment module, the false detection rate is greatly reduced, and the wafer defect detection precision and robustness are remarkably improved.

The present application relates to the technical field of magnetic suspension compressor assembly, and specifically discloses a magnetic pollution source head prevention and control optimization method in the assembly process of a magnetic suspension compressor, comprising: in the pre-treatment stage of the magnetic suspension compressor assembly, using a self-adaptive genetic algorithm to dynamically optimize the thermal cycle parameters in a low-temperature environment, accurately inducing the ferromagnetic impurities hidden in the micro-defects of the compressor components to release in advance; combining wavelet packet transform and neural network algorithm to perform feature extraction and pattern recognition on the released impurity signals; the present application constructs a low-temperature thermal cycle process optimized by a self-adaptive genetic algorithm in the pre-treatment stage of assembly, accurately induces the hidden ferromagnetic impurities in micro-cracks to release in advance, and combines wavelet packet transform and neural network driven targeted pulse demagnetization to accurately remove the impurities before entering the assembly link, breaks through the limitation that traditional cleaning and demagnetization methods cannot reach the impurities in micro-defects, and eliminates the magnetic pollution hidden danger that cannot be found by initial detection from the source.