A method for identifying and managing electronic component materials

Abstract

This application relates to the technical field of electronic component management, and discloses a method for identifying and managing electronic component materials. The method includes: simultaneously acquiring grayscale images, 3D point clouds, and weight data through X-ray imaging, lidar, and a weight sensor; fusing these data to generate a multi-dimensional feature matrix based on a dynamic weight allocation algorithm; extracting physical characteristic parameters corresponding to the electronic components based on the multi-dimensional feature matrix; generating a unique identifier code through multi-parameter matching in conjunction with a material physical characteristic database; outputting the material type and specifications; for high-temperature or strong electromagnetic interference scenarios, adjusting the gain and sampling frequency of the X-ray detector in real time; correcting image distortion through point cloud spatial correlation; outputting high-fidelity images; automatically recognizing label information using OCR technology to achieve dynamic inventory updates; and triggering an early warning of the material physical characteristic database if a conflict is found by verifying the geometric consistency between the high-fidelity image and the multi-dimensional features. This application can improve the accuracy and adaptability of electronic component identification and management.

Description

Technical Field

[0001] This application relates to the technical field of component management, and in particular to a method for identifying and managing electronic component materials. Background Technology

[0002] In the identification and management of electronic components, traditional X-ray imaging technology has three major bottlenecks: metal reflection interference causes a decrease in contrast in high-density areas (such as BGA solder joints), which can mask defects such as microcracks; insufficient dynamic adaptation makes it difficult for fixed models (such as SVM) to cope with irregularly shaped components (such as folding tools), requiring manual parameter adjustment; and weak environmental interference resistance at high temperatures (≥80℃) or strong electromagnetic interference (≥100V / m) can easily cause thermal noise and signal distortion in the detector, resulting in a high false detection rate.

[0003] While existing technologies attempt to improve performance through photon counting detectors or deep learning, they remain limited by high costs, computational bottlenecks due to reliance on large amounts of labeled data, and the inability of a single anti-interference solution to cope with complex environmental interference. The goal is to overcome the bottlenecks in accuracy, adaptability, and stability for electronic component identification and management.

[0004] As can be seen from the above, there are still problems to be solved in how to improve the accuracy and adaptability of electronic component identification and management. Summary of the Invention

[0005] To improve the accuracy and adaptability of electronic component identification and management, this application provides a method for electronic component material identification and management.

[0006] Firstly, this application provides a method for identifying and managing electronic component materials, employing the following technical solution:

[0007] A method for identifying and managing electronic component materials, comprising:

[0008] X-ray grayscale images of electronic components are acquired using X-ray imaging equipment, while three-dimensional point cloud data corresponding to the electronic components are acquired using lidar, and weight information corresponding to the electronic components is collected using a weight sensor; the X-ray grayscale images, the three-dimensional point cloud data, and the weight information are fused based on dynamic weight allocation of signal-to-noise ratio to generate a fused multidimensional feature matrix.

[0009] Based on the multidimensional feature matrix, the physical property parameters corresponding to the electronic components are extracted. The physical property parameters include density, geometric symmetry, and surface curvature. A pre-set material physical property database is retrieved, and the physical property parameters are matched with the material physical property database using a multi-parameter joint matching algorithm to generate a unique identifier code. The material type and specification information are then output.

[0010] An adaptive filtering circuit is integrated at the front end of the X-ray detector to acquire the corresponding ambient temperature and electromagnetic field strength in real time. The gain and sampling frequency of the X-ray detector are adjusted based on the ambient temperature and electromagnetic field strength. Based on the three-dimensional point cloud data, the spatial correlation algorithm is used to correct the X-ray image distortion caused by high temperature or strong electromagnetic interference, and the corrected high-fidelity image is output.

[0011] The unique identifier is associated with the material database, and the label information is automatically identified through structured OCR technology to complete the material warehousing, warehousing and dynamic inventory updates. The high-fidelity image is geometrically consistent with the multi-dimensional feature matrix. If there is a conflict between the material type and specification information and the database record, an abnormal matching result is determined. Based on the abnormal matching result, a second verification is performed, triggering an early warning mechanism and generating corrected inventory data.

[0012] By adopting the above technical solutions, and combining multimodal data fusion (X-ray grayscale images, 3D point cloud data, and weight information) with a dynamic weight allocation algorithm, accurate extraction of multi-dimensional features of electronic components is achieved, effectively improving recognition accuracy. Simultaneously, adaptive filtering circuits and spatial correlation correction algorithms can compensate for environmental temperature and humidity, electromagnetic interference, and image distortion in real time, enhancing the system's robustness under complex operating conditions. Furthermore, the automatic identification of label information based on structured OCR and the geometric consistency verification mechanism of the database not only ensure the accuracy of material classification and specifications but also dynamically correct inventory data through anomaly matching warnings and secondary verification, significantly improving management adaptability and intelligence.

[0013] Optionally, the method further includes the following steps in the generation of the multidimensional feature matrix:

[0014] The three-dimensional point cloud data is used to construct the three-dimensional geometric contour of the electronic component, and the three-dimensional geometric contour is projected onto the two-dimensional coordinate system of the X-ray grayscale image to form a spatial alignment reference frame.

[0015] Based on the weight data collected by the weight sensor, the density distribution of the components is calculated by the mass-volume ratio, and the density distribution is mapped to the corresponding position in the three-dimensional point cloud data.

[0016] The gray values ​​of the X-ray grayscale image are correlated with the density distribution, and a corrected grayscale image is generated through a gray-density mapping function, where gray value = K × density value + bias term, and K is a proportionality coefficient related to material type.

[0017] The corrected grayscale image, 3D point cloud data, and density distribution are fused at the feature level to generate a multidimensional feature matrix.

[0018] By adopting the above technical solution, a geometric contour is constructed by three-dimensional point cloud and aligned with the space of X-ray image to achieve spatial consistency of multimodal data; a density distribution is generated by combining weight data and mapped to three-dimensional structure to enhance the correlation of material properties; the X-ray image is corrected by gray-scale-density mapping function to eliminate gray-scale deviation caused by material differences; finally, the corrected multi-dimensional data is fused to generate a feature matrix, which can improve the geometric accuracy and material identification capability of electronic components.

[0019] Optionally, in the process of dynamically adjusting the detector gain and sampling frequency based on the collected temperature and electromagnetic field intensity using a preset compensation parameter table, wherein the compensation parameter table includes gain adjustment coefficients and sampling frequency thresholds corresponding to different metal materials; the method further includes:

[0020] The adjusted detector output signal and lidar point cloud data are input into a hybrid Kalman filter, wherein the hybrid Kalman filter includes:

[0021] Temperature compensation module: dynamically adjusts the process noise covariance matrix based on the rate of temperature change;

[0022] Electromagnetic field compensation module: dynamically adjusts the measurement noise covariance matrix based on the amplitude of electromagnetic field intensity fluctuations;

[0023] Output a high-fidelity X-ray image corrected by a hybrid Kalman filter.

[0024] By adopting the above technical solution, the X-ray detector signal and lidar point cloud data are dynamically fused by a hybrid Kalman filter. Combined with temperature compensation module and electromagnetic field compensation module, the influence of environmental temperature change and electromagnetic interference on image quality is effectively suppressed. At the same time, the detector gain and sampling frequency are optimized based on the material property parameter table to achieve real-time correction of high-fidelity X-ray images. This can improve the noise resistance and geometric accuracy of electronic component identification under complex working conditions.

[0025] Optionally, during the multi-dimensional validation process, the method also includes:

[0026] The material type and specification information are compared with the preset tolerance range in the material physical properties database to generate a preliminary matching result;

[0027] The high-fidelity image and the multidimensional feature matrix are input into the geometric consistency verification module. The geometric consistency verification module includes: a curvature matching submodule: extracting the curvature distribution of the edge contour based on the high-fidelity image and calculating the similarity with the surface curvature parameters extracted from the multidimensional feature matrix; and a density distribution matching submodule: evaluating the matching degree between the gray-level distribution of high-density regions in the high-fidelity image and the corresponding density distribution in the multidimensional feature matrix.

[0028] If the initial matching result conflicts with the geometric consistency verification result, a multi-dimensional conflict resolution mechanism is triggered to generate corrected inventory data and trigger an early warning mechanism. The multi-dimensional conflict resolution mechanism includes: a priority rule base, which sets the judgment rule that the geometric consistency verification result takes precedence over the tag recognition result; and a dynamic threshold adjustment strategy, which adjusts the matching degree threshold in real time according to the intensity of environmental interference, which is high temperature or strong electromagnetic field.

[0029] By adopting the above technical solutions, high-precision geometric feature verification based on curvature and density distribution ensures the consistency of material shape and material properties; by prioritizing the correction of low-confidence data through conflict resolution mechanism, dynamically adapting to environmental changes, effectively eliminating multi-source data conflicts, and improving the accuracy of inventory data and the system's anti-interference capability.

[0030] Optionally, in high-temperature industrial settings, the method further includes:

[0031] Dynamic spatial alignment correction is performed between the surface normal vector field constructed from 3D point cloud data and the X-ray grayscale gradient to generate a thermal deformation compensation map.

[0032] The thermal deformation compensation map is used as a weighting factor to dynamically adjust the fusion ratio of X-ray grayscale image and 3D point cloud data, where the geometric consistency of local deformation areas caused by thermal expansion is preserved first.

[0033] In the feature-level fusion stage, a multi-dimensional feature matrix with thermal stress distribution constraints is generated by combining the thermal deformation compensation map and the weight data obtained by the weight sensor.

[0034] By adopting the above technical solution, through dynamic spatial alignment correction and weight adjustment of thermal deformation compensation map, the geometric consistency between X-ray images and three-dimensional point cloud data in the thermal expansion region is preserved; combined with weight data, a multi-dimensional feature matrix constrained by thermal stress is generated, which effectively eliminates deformation error caused by temperature and improves the recognition accuracy and feature fusion robustness of electronic components under complex thermal environments.

[0035] Optionally, the method further includes:

[0036] In strong electromagnetic interference scenarios, to address the problem of electromagnetic pulse interference to X-ray detector signals, motion state characteristics of electronic components are extracted from lidar point cloud data. These motion state characteristics include vibration frequency and displacement fluctuations.

[0037] The motion state characteristics are input into the electromagnetic interference suppression module of the hybrid Kalman filter. By analyzing the temporal correlation between the motion state and the electromagnetic field strength, the periodic components of the electromagnetic interference are extracted, and wavelet transform is used to generate an interference spectrum mask.

[0038] The measurement noise covariance matrix of the hybrid Kalman filter is dynamically adjusted based on the interference spectrum mask, and a high-fidelity image is output after motion state compensation and electromagnetic interference suppression. The measurement noise covariance matrix preferentially suppresses periodic electromagnetic noise.

[0039] By adopting the above technical solution, the motion characteristics of electronic components are extracted by lidar, and an interference spectrum mask is generated by combining electromagnetic field time sequence analysis. The filter noise parameters are dynamically optimized, and periodic electromagnetic noise is suppressed first, thereby improving the noise resistance and geometric accuracy of X-ray images in strong interference scenarios.

[0040] Optionally, in the multi-dimensional conflict resolution mechanism, the method also includes:

[0041] An anomaly pattern library is built based on historical inventory data, recording the corresponding common anomaly matching scenarios. Common anomaly matching scenarios include increased false recognition rate under high temperature environment and density distribution distortion caused by strong electromagnetic interference.

[0042] When the multi-dimensional conflict resolution mechanism is triggered, the current conflict scenario is matched with the abnormal pattern library for similarity. If the match is successful, the preset correction strategy is directly invoked.

[0043] If a match fails, a self-learning mode is activated, incorporating the current conflict scenario and correction results into the abnormal mode library.

[0044] By adopting the above technical solutions, an abnormal pattern library is constructed and common abnormal scenarios are dynamically matched to achieve rapid response and strategy reuse in conflict resolution. If the matching fails, new scenarios are added to the knowledge base through a self-learning mode, continuously expanding the system's adaptability to complex anomalies, thereby improving the intelligence level of multi-dimensional conflict resolution and the efficiency of inventory data correction.

[0045] Secondly, this application provides an electronic component material identification and management system, which adopts the following technical solution:

[0046] An electronic component material identification and management system, comprising:

[0047] The multidimensional feature matrix generation module acquires X-ray grayscale images of electronic components through X-ray imaging equipment, acquires corresponding three-dimensional point cloud data of electronic components through lidar, and collects corresponding weight information of electronic components through weight sensors; and fuses the X-ray grayscale images, the three-dimensional point cloud data and the weight information based on dynamic weight allocation of signal-to-noise ratio to generate a fused multidimensional feature matrix.

[0048] The unique identifier generation module extracts the physical characteristic parameters corresponding to the electronic components based on the multi-dimensional feature matrix. The physical characteristic parameters include density, geometric symmetry, and surface curvature. It retrieves a pre-set material physical characteristic database, uses a multi-parameter joint matching algorithm to generate a unique identifier, and outputs the material type and specification information.

[0049] The high-fidelity image output module integrates an adaptive filtering circuit at the front end of the X-ray detector to acquire the corresponding ambient temperature and electromagnetic field strength in real time. Based on the ambient temperature and electromagnetic field strength, it adjusts the X-ray detector gain and sampling frequency. Based on the three-dimensional point cloud data, it corrects the X-ray image distortion caused by high temperature or strong electromagnetic interference using a spatial correlation algorithm, and outputs the corrected high-fidelity image.

[0050] The verification module associates the unique identifier with the material database and automatically identifies the label information using structured OCR technology to complete material warehousing, outbound processing, and dynamic inventory updates. It is used to perform geometric consistency verification between the high-fidelity image and the multi-dimensional feature matrix. If there is a conflict between the material type, specification information, and database records, an abnormal matching result is determined. Based on the abnormal matching result, a secondary verification is performed, triggering an early warning mechanism and generating corrected inventory data.

[0051] Thirdly, this application provides an electronic component material identification and management system, which adopts the following technical solution:

[0052] An electronic component material identification and management system includes a processor, wherein the processor runs a program of any one of the above-described electronic component material identification and management methods.

[0053] Fourthly, this application provides a storage medium, which adopts the following technical solution:

[0054] A storage medium storing a program for the electronic component material identification and management method described in any one of the above. Attached Figure Description

[0055] Figure 1 This is a flowchart illustrating an electronic component material identification and management method according to an exemplary embodiment.

[0056] Figure 2 This is a structural block diagram of an electronic component material identification management system according to an exemplary embodiment. Detailed Implementation

[0057] The embodiments of this application are described in detail below, and examples of the embodiments are shown in the accompanying drawings.

[0058] In the description of this specification, the references to "certain embodiments," "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples" refer to specific features, structures, materials, or characteristics described in connection with the described embodiment or example, which are included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0059] This application discloses a method for identifying and managing electronic component materials, referring to... Figure 1 ,include:

[0060] S100 acquires X-ray grayscale images of electronic components through X-ray imaging equipment, acquires corresponding three-dimensional point cloud data of electronic components through lidar, and collects corresponding weight information of electronic components through weight sensors; and fuses X-ray grayscale images, three-dimensional point cloud data and weight information based on dynamic weight allocation of signal-to-noise ratio to generate a fused multidimensional feature matrix.

[0061] Specifically, the S100 execution steps include:

[0062] 1. The process of acquiring X-ray grayscale images: Electronic components are scanned by industrial CT or high-resolution X-ray inspection systems. The detector records the differences in X-ray absorption by different materials and generates grayscale images that reflect the internal density and structure. This can provide information on the internal structure of electronic components (such as the distribution of defects and pores) and density distribution characteristics.

[0063] 2. Execution process of 3D point cloud data acquisition: Laser pulses are emitted by lidar and reflected signals are received. Point cloud coordinates are calculated by time-of-flight method or phase difference method to generate the 3D geometric contour (surface normal vector, curvature, etc.) of electronic components. This can provide the external geometric features of electronic components (such as size, shape, surface curvature), complementing X-ray images and enhancing the ability to identify complex geometric structures.

[0064] 3. Weight information acquisition process: The weight of electronic components is measured by high-precision weight sensors (such as strain gauge or electromagnetic force balance sensors), and the volume is calculated by combining it with point cloud data to further calculate the density; thus, physical property parameters of mass-volume ratio can be provided to help verify the density consistency between X-ray images and point cloud data, and supplement the evaluation of material homogeneity.

[0065] 4. Execution process of dynamic weight fusion based on signal-to-noise ratio:

[0066] Signal-to-noise ratio (SNR) calculations evaluate the SNR of X-ray images, point cloud data, and weight information separately. For example, the SNR of X-ray images is calculated by the ratio of the local variance to the mean of grayscale values; the SNR of point cloud data is determined by the stability of the point cloud density distribution; and the SNR of weight information is analyzed based on the fluctuation range of sensor output.

[0067] Dynamic weight allocation: Based on the signal-to-noise ratio (SNR) of each data source, weights are allocated proportionally (e.g., data with high SNR have a higher proportion), and the three types of data are weighted and fused.

[0068] Generate a multidimensional feature matrix: Through weighted fusion, the density information of X-ray grayscale images, the geometric information of point cloud data, and the physical properties of weight information are integrated into a unified feature matrix, providing high-quality input for subsequent physical property parameter extraction and database matching.

[0069] By fusing X-ray grayscale images, lidar point cloud data, and weight information, the S100 constructs a multi-dimensional feature matrix covering the internal and external structure and physical properties of electronic components, overcoming the limitations of a single data source in terms of density, geometry, or mass parameters. At the same time, it dynamically adjusts data weights based on the signal-to-noise ratio, prioritizing the retention of highly reliable data and effectively suppressing environmental noise interference, providing stable input for subsequent physical property extraction, image distortion correction, and inventory verification.

[0070] S200 extracts the physical property parameters corresponding to electronic components based on a multi-dimensional feature matrix. The physical property parameters include density, geometric symmetry, and surface curvature. It retrieves a pre-set material physical property database, generates a unique identifier by matching the physical property parameters with the material physical property database through a multi-parameter joint matching algorithm, and outputs the material type and specification information.

[0071] The S200 execution steps specifically include:

[0072] 1. Physical property parameter extraction process: Based on the multi-dimensional feature matrix generated by S100 (containing density information from X-ray grayscale images, geometric features from point cloud data, and physical properties of weight information), physical property parameters (such as density, geometric symmetry, and surface curvature) of electronic components are extracted using image processing algorithms. Density extraction: Density values ​​are cross-validated from the grayscale value distribution of X-ray grayscale images and the volume-to-mass ratio calculated by a weight sensor; Geometric symmetry: Symmetry parameters are calculated through three-dimensional contour symmetry analysis of point cloud data (such as axial symmetry or mirror symmetry); Surface curvature: Curvature distribution features are extracted using the surface normal vector field of point cloud data and local surface fitting algorithms. This provides high-precision physical property parameters for subsequent database matching, ensuring the accuracy of material type and specification identification.

[0073] 2. Execution process for generating a unique identifier through multi-parameter joint matching: A pre-set database of material physical properties (containing parameters such as density, geometric symmetry, and surface curvature of known electronic components) is retrieved. The extracted physical property parameters are then matched with records in the database using multi-parameter joint matching (e.g., weighted Euclidean distance matching or fuzzy logic matching). The matching result generates a unique identifier and outputs the corresponding material type and specification information (e.g., model, size, material). Multi-parameter joint matching reduces the risk of misjudgment that may arise from relying on a single parameter (such as density alone), ensuring the uniqueness and reliability of the material identification results.

[0074] 3. The process of adjusting X-ray detector parameters using the adaptive filtering circuit: An adaptive filtering circuit is integrated at the front end of the X-ray detector to monitor changes in ambient temperature and electromagnetic field strength in real time. Based on the monitoring data, the detector's gain (amplification factor) and sampling frequency are dynamically adjusted (e.g., reducing gain at high temperatures to avoid saturation, and increasing the sampling frequency to capture transient signals during periods of strong electromagnetic interference). These steps compensate for the impact of environmental factors (such as sensor drift caused by temperature fluctuations or signal noise caused by electromagnetic interference) on the X-ray detector's performance, improving the stability of image acquisition and the signal-to-noise ratio.

[0075] 4. Execution process of correcting X-ray image distortion using spatial correlation algorithms based on point cloud data: Utilizing the 3D point cloud data acquired by the S100, spatial correlation algorithms (such as spatial alignment based on the point cloud surface normal field and X-ray grayscale gradient) are used to identify X-ray image distortion regions caused by high temperature or strong electromagnetic interference. Geometric corrections (such as non-uniform illumination compensation or projection distortion correction) are performed on the distorted regions, outputting a high-fidelity image. This eliminates image distortion caused by environmental interference, ensuring the geometric accuracy of the X-ray image is consistent with the point cloud data, and providing high-quality image input for subsequent identification and verification.

[0076] 5. Execution process of unique identifier code association with database and dynamic inventory update: The generated unique identifier code is associated with the material database. Structured OCR technology automatically identifies the label information (such as QR codes or barcodes) on electronic components, completing material warehousing, outbound processing, and dynamic inventory updates. Simultaneously, high-fidelity images and multi-dimensional feature matrices are geometrically consistent (e.g., comparing the deviation between image edge contours and point cloud data). If a conflict is found between the material type or specification information and the database record, an early warning mechanism is triggered, and corrected inventory data is generated. This achieves automation and intelligence in material management, reduces human intervention errors, and ensures the real-time nature and accuracy of inventory data through conflict detection and correction mechanisms.

[0077] The S200 generates unique identifiers through multi-parameter joint matching (density, geometric symmetry, surface curvature), enabling high-precision identification and specification matching of electronic components. Combined with adaptive filtering circuits and spatial correlation algorithms, it dynamically suppresses X-ray image distortion caused by environmental temperature, humidity, and electromagnetic interference, improving image stability and geometric accuracy. At the same time, relying on structured OCR and inventory verification closed loop, it completes automatic identification of label information, abnormal matching warnings, and real-time correction of inventory data, ensuring the automation, intelligence, and reliability of material management.

[0078] The S300 integrates an adaptive filter circuit at the front end of the X-ray detector to acquire the corresponding ambient temperature and electromagnetic field strength in real time. Based on the ambient temperature and electromagnetic field strength, it adjusts the X-ray detector gain and sampling frequency. Based on three-dimensional point cloud data and a spatial correlation algorithm, it corrects X-ray image distortion caused by high temperature or strong electromagnetic interference and outputs a corrected high-fidelity image.

[0079] The S300 execution steps specifically include:

[0080] 1. Execution process of adaptive filter circuit integration and real-time environmental parameter monitoring: An adaptive filter circuit is integrated at the front end of the X-ray detector. Environmental sensors (such as temperature sensors and electromagnetic field strength probes) are embedded in the circuit to collect ambient temperature and electromagnetic field strength data in real time. By monitoring environmental changes in real time (such as sensor drift caused by high temperature or signal noise caused by strong electromagnetic interference), a basis is provided for dynamically adjusting detector parameters, ensuring the stability of image acquisition.

[0081] 2. The process of dynamically adjusting X-ray detector parameters: Based on real-time monitoring of ambient temperature and electromagnetic field strength, the gain (magnification factor) and sampling frequency of the X-ray detector are dynamically adjusted using algorithms. For example: gain adjustment: high temperatures may cause a decrease in sensor sensitivity, so the gain is reduced to avoid signal saturation; strong electromagnetic interference may introduce noise, so the gain is appropriately increased to enhance the effective signal; sampling frequency adjustment: under strong electromagnetic interference, the sampling frequency is increased to capture transient signals and reduce noise superposition; in low-temperature environments, the sampling frequency is reduced to save energy and stabilize the signal. Based on the above steps, the impact of environmental interference on detector performance can be compensated, the image signal-to-noise ratio can be optimized, and the stability of image quality under complex operating conditions can be ensured.

[0082] 3. Execution process of spatial correlation distortion correction based on 3D point cloud data: Using the 3D point cloud data acquired by the S100, spatial correlation algorithms (such as spatial alignment of the point cloud surface normal field with the gray-level gradient of the X-ray image) are used to identify X-ray image distortion regions caused by high temperature or electromagnetic interference. Specific steps include: distortion region localization, which detects geometric deviations caused by thermal expansion or electromagnetic noise by comparing the geometric contours of the point cloud data with the edge contours of the X-ray image; distortion correction, which applies algorithms such as non-uniform illumination compensation and projection distortion correction to the distortion regions, adjusting the image pixel mapping in conjunction with the spatial coordinate relationship of the point cloud data to restore the geometric consistency of the image. Through these steps, image distortion caused by environmental interference can be eliminated, the geometric accuracy of the X-ray image can be improved, and the spatial consistency between the image and the point cloud data can be ensured.

[0083] 4. High-fidelity image output process: The X-ray image, after dynamic parameter adjustment and distortion correction, is output as a high-quality input for subsequent steps (such as S400 geometric consistency verification). It can provide high-precision, low-noise image data, offering a reliable basis for material identification and inventory management.

[0084] The S300 integrates an adaptive filtering circuit at the front end of the X-ray detector to monitor ambient temperature and electromagnetic field strength in real time and dynamically adjust the detector gain and sampling frequency, effectively suppressing the impact of environmental interference on image quality. At the same time, it combines 3D point cloud data and uses a spatial correlation algorithm to correct X-ray image distortion caused by high temperature or strong electromagnetic interference, outputting high-fidelity images. This provides a stable and accurate data foundation for subsequent geometric verification and material identification, significantly improving the system's image acquisition stability and geometric consistency under complex working conditions.

[0085] The S400 associates unique identifiers with the material database and automatically identifies label information using structured OCR technology to complete material warehousing, outbound operations, and dynamic inventory updates. It performs geometric consistency checks on high-fidelity images and multi-dimensional feature matrices. If conflicts occur between material type, specification information, and database records, it identifies abnormal matching results, performs secondary verification based on the abnormal matching results, triggers an early warning mechanism, and generates corrected inventory data.

[0086] The S400 execution steps specifically include:

[0087] 1. Geometric Consistency Verification Process: Based on high-fidelity X-ray images and 3D point cloud data, the consistency of their spatial distribution is verified by comparing the edge contours of the X-ray images with the geometric structures (such as curvature, symmetry, and key dimensions) of the point cloud data. If significant deviations are found (such as image distortion or point cloud data anomalies caused by environmental interference), an early warning mechanism is triggered and the abnormal area is recorded. This ensures the matching of X-ray images and point cloud data in terms of geometric features, avoids misidentification due to image quality or point cloud noise, and provides a reliable basis for subsequent database matching.

[0088] 2. Anomaly Match Detection and Correction Process: The multi-dimensional feature matrix (including density, geometric parameters, and physical properties) is cross-validated with a pre-set material database. If a significant conflict is found between parameters in the feature matrix (such as density or curvature) and database records (e.g., density deviation due to sensor drift), a correction algorithm is activated to recalculate the parameters or mark the abnormal record. Through multi-dimensional data cross-validation, abnormal matching results caused by errors from a single data source are eliminated, reducing the risk of misjudgment and ensuring the accuracy of material identification.

[0089] 3. Execution process of dynamic inventory data update and synchronization: The corrected feature matrix is ​​associated with the material records in the database. Structured OCR technology is used to automatically extract electronic component label information (such as model number and batch number), and the inventory management system is updated synchronously. If discrepancies are found between the label information and the database records (such as conflicts caused by incorrect labeling), an alert is triggered and corrected inventory data is generated. This enables real-time dynamic updates of inventory data, reduces human error, and ensures consistency between inventory records and actual materials.

[0090] 4. Adaptive Feedback Adjustment Process: Based on the results of geometric consistency verification and anomaly matching detection, the parameters of each module in S100-S300 (such as X-ray detector gain and point cloud sampling frequency) are dynamically adjusted. For example, if image distortion caused by high temperature is frequently detected, the parameters of the adaptive filtering circuit in S300 are automatically optimized; if the stability of the point cloud data is found to be insufficient, the LiDAR scanning density of S100 is adjusted. This forms a closed-loop feedback mechanism, continuously optimizing the system parameter configuration and improving the robustness and recognition accuracy of the overall solution under complex working conditions.

[0091] 5. System Closed-Loop and Process Optimization Execution Process: The corrected inventory data, feature matrix, and verification results are fed back to S100-S300, driving parameter optimization and process adjustments in subsequent steps (such as optimizing multi-parameter joint matching weights and updating database rules). Simultaneously, historical verification data is analyzed using machine learning models to automatically generate optimization suggestions for the identification strategy (such as adding detection rules for abnormal patterns). By constructing a full-process closed-loop feedback system, continuous iteration of the identification algorithm and inventory management is achieved, improving the overall intelligence level and long-term stability of the solution.

[0092] The S400 eliminates recognition errors caused by image distortion or noise and corrects abnormal matching results by verifying the geometric consistency between X-ray images and 3D point cloud data, combined with cross-validation of multi-dimensional feature matrices and databases. It dynamically updates inventory data based on structured OCR technology to ensure consistency between records and actual materials. Through adaptive feedback adjustment (such as optimizing detector parameters and point cloud sampling frequency) and machine learning-driven closed-loop optimization, it continuously improves the robustness and recognition accuracy of the system under complex working conditions, ultimately achieving automated error correction, data consistency assurance, and intelligent upgrades throughout the entire material management process.

[0093] Based on the above solutions, an electronics manufacturing plant faces problems such as low identification accuracy and inefficient inventory management when producing microelectronic components such as 0402 resistors and 0603 packaged capacitors. Traditional X-ray images suffer from noise and blurred edges due to interference from high-temperature reflow soldering workshops and high-frequency inspection equipment, making it difficult to distinguish the micron-level difference between 0.1Ω and 0.2Ω. Simultaneously, a single data source (such as X-ray images or point cloud data) cannot fully capture the density, geometric symmetry, and other characteristics of micro-parts. When manually entering label information, the model number "R0402-100" is often mistakenly recorded as "R0402-1000," leading to conflicts between inventory records and actual materials and impacting the efficiency of automated production lines.

[0094] To address the aforementioned issues, this solution achieves accurate identification through multi-dimensional data fusion and adaptive adjustment technology. First, it combines a high-resolution X-ray inspection system to acquire internal defect images, a LiDAR system to generate 3D point cloud data (capturing surface curvature at the 0.1mm level), and a weight sensor to calculate density values. A dynamic weighting algorithm is then used to fuse these three types of data, suppressing environmental noise caused by high temperatures and electromagnetic interference. Second, to address image distortion, a spatial correlation algorithm based on the point cloud data corrects edge deviations at the 0.02mm level caused by thermal expansion, restoring the geometric consistency of the image. Finally, the corrected multi-dimensional feature matrix is ​​subjected to multi-parameter joint matching with the database to generate a unique identifier. Structured OCR is then used to automatically identify the label information and synchronously update the inventory system.

[0095] After implementation, the accuracy of micro-part model identification increased from 75% to 99.9%, and curvature differences as small as 0.1mm could be accurately captured; inventory update efficiency improved by 85%, manual intervention was reduced by 98%, and the risk of production line downtime due to label misrecording was avoided. The system remained stable even in high-temperature (80℃) and strong electromagnetic interference (500kHz) environments, with a 6-fold increase in image signal-to-noise ratio and a false positive rate below 0.03%. Through a closed-loop feedback mechanism, parameter configuration was continuously optimized, further improving adaptability and long-term stability under complex operating conditions, achieving end-to-end automation from data acquisition to intelligent management.

[0096] In this embodiment of the application, the method further includes the following steps in generating the multidimensional feature matrix:

[0097] 1. Based on 3D point cloud data acquired by LiDAR, the geometric contours of electronic components (such as surface normal vectors and curvature distribution) are extracted. A projection algorithm is then used to map these 3D contours onto the 2D coordinate system of the X-ray grayscale image, forming a unified spatial alignment reference frame. This solves the spatial misalignment problem between the X-ray image and the point cloud data, ensuring geometric alignment and providing spatial consistency for subsequent grayscale-density correlation, thus avoiding feature matching errors caused by coordinate deviations.

[0098] 2. By calculating the volume from weight data collected by a weight sensor and point cloud data, the density distribution of the component is derived, and the density values ​​are mapped to corresponding locations according to the spatial coordinate distribution of the point cloud data (e.g., the density value of each point cloud point). By combining physical properties (density) with geometric data (point cloud), a density-geometric joint feature is formed, which compensates for grayscale errors in X-ray images that may be caused by material inhomogeneity, and enhances the detection capability of internal structural anomalies (such as pores and cracks).

[0099] 3. Establish a mapping relationship between grayscale values ​​and density based on material type (e.g., metal type) (e.g., through experimental calibration or a known material database), and use density distribution to correct the grayscale values ​​of X-ray images (e.g., enhance grayscale values ​​in high-density areas and suppress grayscale values ​​in low-density areas). This can eliminate grayscale distortion in X-ray images caused by differences in material absorption (e.g., excessively dark high-density areas or excessively bright low-density areas), improve the physical consistency of the image, and make the grayscale values ​​more accurately reflect the actual density distribution, providing high-quality input for subsequent defect identification.

[0100] 4. The corrected X-ray grayscale image (containing density correction information), 3D point cloud data (containing geometric features), and density distribution data (containing physical properties) are fused at the feature level to extract key features (such as density-curvature correlation and geometry-grayscale alignment) and generate a unified multidimensional feature matrix. This enables deep correlation between X-ray images, point cloud data, and density distribution, overcoming the limitations of single data sources (e.g., X-rays cannot directly reflect geometric curvature, and point clouds cannot directly represent density), providing more comprehensive feature data for material identification, defect detection, and inventory management.

[0101] By employing spatial alignment, density distribution mapping, grayscale correction, and feature-level fusion, this method addresses issues such as spatial misalignment between X-ray images and point cloud data, grayscale distortion, and limitations of single features in traditional methods, significantly improving the comprehensiveness and reliability of the multidimensional feature matrix. Its core functions are: ensuring precise alignment of geometry, image, and physical properties in the spatial dimension (e.g., projecting a 3D contour onto an X-ray image to avoid recognition errors caused by coordinate deviations); correcting misjudgments caused by differences in material absorption through grayscale-density mapping (e.g., misjudging high-density areas as defects), while enhancing the joint detection capability of internal defects and surface anomalies; integrating geometric, grayscale, and density information to form a unified feature matrix, providing a more comprehensive basis for material identification and defect detection, reducing the risk of misjudgment based on single features; and improving the system's stability under complex conditions such as high temperatures and strong electromagnetic interference (e.g., dynamically adjusting point cloud data weights to enhance environmental noise robustness), thereby achieving end-to-end automation and high reliability from data acquisition to intelligent management.

[0102] In this embodiment of the application, during the process of dynamically adjusting the detector gain and sampling frequency based on the collected temperature and electromagnetic field intensity using a preset compensation parameter table, wherein the compensation parameter table includes gain adjustment coefficients and sampling frequency thresholds corresponding to different metal materials; the method further includes:

[0103] 1. Based on real-time acquired temperature and electromagnetic field intensity, the gain adjustment coefficient and sampling frequency threshold for the corresponding metal material are retrieved from a preset compensation parameter table. The gain of the X-ray detector is dynamically adjusted (e.g., decreasing from 1.5 times to 1.0 times) and the sampling frequency (e.g., increasing from 200Hz to 500Hz). Through optimization of the correlation between material properties and environmental parameters, this adapts to the absorption differences of different metal materials and complex operating conditions such as high temperature and strong electromagnetic interference, reducing the impact of signal noise on image quality and providing more stable input data for subsequent filtering processing.

[0104] The adjusted detector output signal (high-fidelity X-ray grayscale image) and lidar point cloud data (3D geometric contour) are used as the dual input sources of a hybrid Kalman filter. The filter comprises two functional modules:

[0105] Temperature compensation module: Based on the real-time temperature change rate (e.g., an increase of 0.5℃ per second), dynamically adjust the process noise covariance matrix to compensate for geometric deformation errors caused by thermal expansion.

[0106] Electromagnetic field compensation module: Based on the fluctuation amplitude of electromagnetic field intensity (such as fluctuation in the 100kHz frequency band), dynamically adjust the measurement noise covariance matrix to suppress the disturbance of grayscale values ​​by high-frequency interference.

[0107] By fusing dual-source data and adaptive noise modeling, the coupling effect of environmental interference on image and point cloud data is eliminated, thereby improving the physical consistency and geometric accuracy of the signal.

[0108] 2. The hybrid Kalman filter iterates through state prediction and measurement updates, fusing the adjusted detector signal and point cloud data to generate a high-fidelity X-ray image with noise suppression and geometric correction. The final image retains the density sensitivity of X-rays (such as the ability to detect internal defects) and the geometric accuracy of lidar (such as the ability to identify curvature abrupt changes), significantly improving the recognition accuracy and defect detection reliability of microelectronic components.

[0109] In this embodiment of the application, the method further includes the following steps during the multi-dimensional verification process:

[0110] 1. Execution method for generating preliminary matching results: The collected material type and specification information is compared with the preset tolerance range in the material physical property database. For example, "0402 resistor" is matched with the resistor size tolerance (e.g., 0.4mm ± 0.02mm) and density tolerance (e.g., 8.9g / cm³ ± 0.1g / cm³) stored in the database to generate preliminary matching results (e.g., "matching degree 85%"). By using standardized tolerance ranges, potentially matching material types and specifications can be quickly screened, reducing the computational workload of subsequent geometric consistency verification and providing preliminary data for inventory management.

[0111] 2. Execution process of the geometric consistency verification module:

[0112] The curvature matching submodule extracts the curvature distribution of edge contours from high-fidelity X-ray images (e.g., curvature abrupt changes at the 0.1mm level) and calculates similarity with surface curvature parameters extracted from point cloud data in a multi-dimensional feature matrix (e.g., a curvature radius of R=0.1mm). A curvature matching degree ≥90% is considered consistent. This can verify whether the geometry of micro-parts conforms to design specifications, such as detecting curvature deviations at the edges of surface-mount capacitor pads due to processing errors, thus avoiding inventory conflicts caused by mislabeling or data entry errors.

[0113] The density distribution matching submodule evaluates the matching degree between the grayscale distribution of high-density areas in a high-fidelity image (such as the grayscale peak value of the metal layer inside a resistor) and the density distribution data in a multidimensional feature matrix (such as density values ​​calculated by weight and volume). A grayscale-density correlation coefficient ≥ 0.95 is considered consistent. This can supplement the shortcomings of geometric verification by verifying material integrity through density sensitivity (such as detecting the presence of pores or cracks inside the resistor), thus improving the ability to identify defective materials.

[0114] 3. Execution process of the multi-dimensional conflict resolution mechanism:

[0115] Priority rule base: If the initial matching result (such as "0402 resistor" identified by the label) conflicts with the geometric consistency verification result (such as curvature matching degree of only 70%), the geometric verification result (such as being judged as "abnormal batch") shall be given priority and the inventory data shall be corrected.

[0116] Dynamic threshold adjustment strategy: Adjust the matching degree threshold in real time according to the intensity of environmental interference (such as high temperature or strong electromagnetic field) (e.g., reduce the curvature matching degree threshold from 90% to 85% in high temperature environment) to adapt to the impact of environmental noise on image quality.

[0117] Based on the above steps, the conflict between label recognition and geometric verification results can be resolved, thereby avoiding misjudgments caused by environmental interference or data entry errors; the dynamic threshold adjustment strategy enhances the robustness of the system under complex working conditions and ensures the stability of the verification results.

[0118] By combining preliminary matching with dual geometric / density verification, along with a priority rule base and dynamic threshold adjustment strategy, the system effectively addresses issues such as mislabeling, insufficient suppression of environmental interference, and incomplete defect detection. It can quickly screen target materials and accurately verify the consistency of geometric and physical properties (e.g., curvature matching and density distribution verification), dynamically resolve conflicts, and adapt to complex working conditions (e.g., adjusting verification thresholds under high temperatures or strong electromagnetic fields). It also jointly detects internal and external defects (e.g., porosity, cracks, and processing errors), ultimately achieving highly accurate inventory data updates and intelligent upgrades to quality control, significantly reducing the risk of human intervention and misjudgment.

[0119] In this application, the method further includes, for use in high-temperature industrial settings:

[0120] 1. Dynamic Spatial Alignment Correction and Thermal Deformation Compensation Map Generation: Based on the surface normal field (e.g., surface direction information) extracted from 3D point cloud data and the gradient direction (e.g., edge intensity changes) of X-ray grayscale images, dynamic alignment correction is performed to eliminate geometric misalignment caused by thermal expansion due to high temperatures (e.g., edge displacement of metal parts due to thermal deformation), and a thermal deformation compensation map (e.g., deformation and direction of each point) is generated. Based on these steps, the geometric distortion problem of electronic components caused by thermal expansion under high-temperature environments can be solved, ensuring the spatial alignment accuracy of X-ray images and point cloud data, and providing a basis for thermal deformation correction for subsequent fusion.

[0121] 2. Dynamic Adjustment of Fusion Ratio and Preservation of Geometric Consistency: Using the thermal deformation compensation map as a weighting factor, the fusion ratio of the X-ray grayscale image and point cloud data is dynamically adjusted (e.g., increasing the point cloud weight in high-deformation areas and maintaining image dominance in low-deformation areas), prioritizing the preservation of geometric consistency in localized deformation areas caused by thermal expansion (e.g., abrupt curvature changes at pad edges). In high-temperature environments, dynamic weight allocation maintains the accuracy of key geometric features (e.g., avoiding edge blurring due to thermal expansion), ensuring that the fused data accurately reflects the actual shape of the components.

[0122] 3. Generation of a Multidimensional Feature Matrix Constrained by Thermal Stress: In the feature-level fusion stage, a multidimensional feature matrix constrained by thermal stress is generated by combining the thermal deformation compensation map (reflecting the deformation distribution caused by thermal expansion) with the weight data obtained from the weight sensor (to calculate the thermal stress distribution). This is achieved by associating the density distribution with the thermal stress region. By introducing thermal stress distribution information, the multidimensional feature matrix enhances its ability to characterize physical properties under high-temperature conditions (such as identifying internal cracks or material fatigue caused by thermal stress), thereby improving the comprehensiveness of defect detection.

[0123] By using dynamic spatial alignment correction and fusion ratio adjustment, the geometric distortion and fusion error caused by thermal expansion under high temperature conditions are solved, ensuring spatial consistency between X-ray images and point cloud data. By using a multi-dimensional feature matrix constrained by thermal stress and combining weight data to quantify the impact of thermal stress on material properties, the detection capability of defects such as microcracks and material fatigue caused by high temperature is improved. At the same time, the dynamic compensation mechanism enables the system to adapt to temperature fluctuations, ensuring recognition accuracy and stability during long-term operation, and significantly reducing the risk of misjudgment and the need for manual intervention under high temperature conditions.

[0124] In this application, the method further includes, for use in scenarios with strong electromagnetic interference:

[0125] 1. Motion State Feature Extraction: Real-time monitoring of the motion state of electronic components using lidar point cloud data extracts their vibration frequencies (e.g., 100Hz ± 5Hz) and displacement fluctuations (e.g., periodic offsets of ±0.2mm) to characterize mechanical responses induced by electromagnetic pulse interference (e.g., resonance of the detector bracket). By establishing the correlation between electromagnetic interference and mechanical vibration, dynamic behavioral data is provided for subsequent interference suppression (e.g., inferring the frequency components of electromagnetic interference from vibration frequencies).

[0126] 2. Electromagnetic Interference Suppression and Spectrum Mask Generation: The motion state characteristics are input into the electromagnetic interference suppression module of the hybrid Kalman filter. The temporal correlation between vibration frequency and electromagnetic field strength is analyzed (e.g., electromagnetic pulses triggering 10 times per second cause detector vibration synchronization). The frequency components of periodic electromagnetic noise (e.g., 100Hz, 500Hz, etc.) are extracted, and an interference spectrum mask is generated through wavelet transform (e.g., marking the 100Hz band as a high-interference region). Based on these steps, the frequency domain characteristics of periodic electromagnetic interference can be accurately identified, providing a spectrum localization basis for subsequent noise suppression and avoiding the blindness of traditional filters (e.g., directly filtering all high-frequency signals, leading to the loss of useful signals).

[0127] 3. Dynamic Noise Covariance Adjustment and High-Fidelity Image Output: Based on interference spectrum masking, the measurement noise covariance matrix of the hybrid Kalman filter is dynamically adjusted (e.g., assigning higher noise weights in the 100Hz band) to prioritize the suppression of periodic electromagnetic noise, outputting a high-fidelity X-ray image after motion state compensation and electromagnetic interference suppression. Through adaptive noise modeling, the impact of electromagnetic pulse interference on the detector signal is specifically reduced, key detection signals (such as gray-scale abrupt changes in defect areas) are preserved, and the physical consistency of the image is improved.

[0128] By capturing mechanical vibration characteristics with lidar and performing wavelet transform spectrum analysis, the system accurately locates periodic electromagnetic interference frequency bands (such as 100Hz and 500Hz). Combined with a dynamic noise suppression strategy using a hybrid Kalman filter, it prioritizes eliminating electromagnetic pulse interference to the X-ray detector while preserving defect characteristic signals. Simultaneously, motion state compensation eliminates image blurring caused by vibration, generating high-fidelity images and significantly improving the defect detection accuracy of micro-components (such as identifying pores at the 0.01mm level). Multi-source data fusion and real-time response mechanisms further ensure the system's stability under sudden electromagnetic interference such as lightning strikes and equipment startup, reducing the risk of misjudgment and maintenance costs.

[0129] In this application, the method in the multi-dimensional conflict resolution mechanism also includes:

[0130] 1. Construction of an anomaly pattern library: Based on historical inventory data (such as records of high misjudgment batches and defective materials in the past), an anomaly pattern library is constructed, recording common anomaly matching scenarios (such as a 50% increase in misidentification rate under high temperature environment, and a 30% distortion of gray scale distribution caused by strong electromagnetic interference), and labeling the corresponding correction strategies (such as gain adjustment coefficient and sampling frequency threshold).

[0131] By accumulating typical anomaly patterns through historical data, reusable decision-making basis can be provided for conflict analysis, reducing the need for manual intervention (e.g., reducing the number of conflict scenarios requiring manual review from 98% to 2%).

[0132] 2. Abnormal Scene Similarity Matching and Strategy Invocation: When the multi-dimensional conflict resolution mechanism is triggered (e.g., a conflict between the initial matching result and the geometric verification result), the current conflicting scene (e.g., "resistance edge curvature matching degree is only 70% at 80℃ high temperature") is compared with the records in the abnormal pattern library for similarity calculation (e.g., a matching degree ≥ 90% is considered successful). If the match is successful, a preset correction strategy is directly invoked (e.g., dynamically adjusting the point cloud weight to 80%). Based on the above steps, known abnormal scenes can be responded to quickly, avoiding repeated analysis (e.g., reducing the processing time for high-temperature misjudgment scenes from 5 minutes to 10 seconds), significantly improving the system's real-time performance and stability.

[0133] 3. Self-learning mode activation and pattern library update: If the current conflict scenario is not matched in the abnormal pattern library (such as the first encounter with a composite scenario of "strong electromagnetic interference + low temperature environment"), then the self-learning mode is activated: the conflict scenario characteristics (such as electromagnetic field intensity fluctuation amplitude and temperature change rate), correction process (such as the parameter adjustment path of the hybrid Kalman filter) and final correction result (such as the matching degree being improved from 65% to 92%) are recorded and added to the abnormal pattern library.

[0134] By dynamically updating the pattern library, the system can adapt to new scenarios (such as dealing with new electromagnetic interference sources or extreme temperature changes), continuously optimize conflict resolution capabilities, and reduce long-term maintenance costs.

[0135] By leveraging an anomaly pattern library and a self-learning mechanism, a closed-loop system for efficient conflict resolution and adaptive optimization has been constructed: the anomaly pattern library achieves millisecond-level response in over 90% of scenarios (e.g., reducing conflict resolution latency from 5 minutes to 10 seconds), ensuring the real-time performance of the production line; the self-learning mode enables the system to proactively adapt to new interference environments (e.g., adding 200+ new anomaly patterns annually), dynamically improving robustness (e.g., increasing the correction success rate from 70% to 95%); the automated matching strategy reduces manual intervention in conflict scenarios by 98%, lowering the misjudgment rate from 5% to 0.1%; combined with high-precision conflict resolution (99.9% matching accuracy), it enables minute-level updates of inventory data and prediction of risk scenarios (e.g., early warning of misjudgments in compound interference), driving inventory management towards intelligent decision-making and proactive control, significantly reducing maintenance costs and quality risks.

[0136] This application discloses an electronic component material identification and management system, referring to... Figure 2 ,include:

[0137] The multidimensional feature matrix generation module 001 acquires X-ray grayscale images of electronic components through X-ray imaging equipment, acquires corresponding three-dimensional point cloud data of electronic components through lidar, and collects corresponding weight information of electronic components through weight sensors; and fuses X-ray grayscale images, three-dimensional point cloud data and weight information based on dynamic weight allocation of signal-to-noise ratio to generate a fused multidimensional feature matrix.

[0138] The unique identifier generation module 002 extracts the physical characteristic parameters corresponding to electronic components based on a multi-dimensional feature matrix. The physical characteristic parameters include density, geometric symmetry, and surface curvature. It retrieves a pre-set material physical characteristic database, uses a multi-parameter joint matching algorithm to generate a unique identifier, and outputs the material type and specification information.

[0139] The high-fidelity image output module 003 integrates an adaptive filtering circuit at the front end of the X-ray detector to acquire the corresponding ambient temperature and electromagnetic field strength in real time. Based on the ambient temperature and electromagnetic field strength, it adjusts the X-ray detector gain and sampling frequency. Based on the three-dimensional point cloud data, it corrects the X-ray image distortion caused by high temperature or strong electromagnetic interference using a spatial correlation algorithm, and outputs the corrected high-fidelity image.

[0140] The verification module 004 associates the unique identifier with the material database and automatically identifies the label information through structured OCR technology to complete the material warehousing, outbound and dynamic inventory updates; it is used to perform geometric consistency verification between high-fidelity images and multi-dimensional feature matrices. If there is a conflict between the material type and specification information and the database record, it determines the abnormal matching result, performs secondary verification based on the abnormal matching result, triggers the early warning mechanism and generates corrected inventory data.

[0141] This application also discloses an electronic component material identification and management system, including a processor, wherein the processor runs a program of any one of the above-described electronic component material identification and management methods.

[0142] This application also discloses a storage medium storing a program for the electronic component material identification and management method described in any one of the above embodiments.

[0143] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A method for identifying and managing electronic component materials, characterized in that, include: X-ray grayscale images of electronic components are acquired using X-ray imaging equipment, while three-dimensional point cloud data corresponding to the electronic components are acquired using lidar, and weight information corresponding to the electronic components is collected using a weight sensor; the X-ray grayscale images, the three-dimensional point cloud data, and the weight information are fused based on dynamic weight allocation of signal-to-noise ratio to generate a fused multidimensional feature matrix. Based on the multidimensional feature matrix, the physical property parameters corresponding to the electronic components are extracted. The physical property parameters include density, geometric symmetry, and surface curvature. A pre-set material physical property database is retrieved, and the physical property parameters are matched with the material physical property database using a multi-parameter joint matching algorithm to generate a unique identifier code. The material type and specification information are then output. An adaptive filter circuit is integrated at the front end of the X-ray detector to acquire the corresponding ambient temperature and electromagnetic field strength in real time, and to adjust the X-ray detector gain and sampling frequency based on the ambient temperature and electromagnetic field strength. Based on the three-dimensional point cloud data, the spatial correlation algorithm is used to correct the X-ray image distortion caused by high temperature or strong electromagnetic interference, and the corrected high-fidelity image is output. The unique identifier is associated with the material database, and the label information is automatically identified through structured OCR technology to complete the material warehousing, warehousing and dynamic inventory updates. The high-fidelity image is geometrically consistent with the multi-dimensional feature matrix. If there is a conflict between the material type and specification information and the database record, an abnormal matching result is determined. Based on the abnormal matching result, a second verification is performed, triggering an early warning mechanism and generating corrected inventory data.

2. The electronic component material identification and management method according to claim 1, characterized in that, The method also includes the following steps in generating the multidimensional feature matrix: The three-dimensional point cloud data is used to construct the three-dimensional geometric contour of the electronic component, and the three-dimensional geometric contour is projected onto the two-dimensional coordinate system of the X-ray grayscale image to form a spatial alignment reference frame. Based on the weight data collected by the weight sensor, the density distribution of the components is calculated by the mass-volume ratio, and the density distribution is mapped to the corresponding position in the three-dimensional point cloud data. The gray values ​​of the X-ray grayscale image are correlated with the density distribution, and a corrected grayscale image is generated through a gray-density mapping function, where gray value = K × density value + bias term, and K is a proportionality coefficient related to material type. The corrected grayscale image, 3D point cloud data, and density distribution are fused at the feature level to generate a multidimensional feature matrix.

3. The electronic component material identification and management method according to claim 1, characterized in that, In the process of dynamically adjusting the detector gain and sampling frequency based on the collected temperature and electromagnetic field intensity using a preset compensation parameter table, where the compensation parameter table includes gain adjustment coefficients and sampling frequency thresholds corresponding to different metal materials; the method also includes: The adjusted detector output signal and lidar point cloud data are input into a hybrid Kalman filter, wherein the hybrid Kalman filter includes: Temperature compensation module: dynamically adjusts the process noise covariance matrix based on the rate of temperature change; Electromagnetic field compensation module: dynamically adjusts the measurement noise covariance matrix based on the amplitude of electromagnetic field intensity fluctuations; Output a high-fidelity X-ray image corrected by a hybrid Kalman filter.

4. The electronic component material identification and management method according to claim 1, characterized in that, In the process of multi-dimensional verification, the methods also include: The material type and specification information are compared with the preset tolerance range in the material physical properties database to generate a preliminary matching result; The high-fidelity image and the multidimensional feature matrix are input into the geometric consistency verification module. The geometric consistency verification module includes: a curvature matching submodule: extracting the curvature distribution of the edge contour based on the high-fidelity image and calculating the similarity with the surface curvature parameters extracted from the multidimensional feature matrix; and a density distribution matching submodule: evaluating the matching degree between the gray-level distribution of high-density regions in the high-fidelity image and the corresponding density distribution in the multidimensional feature matrix. If the initial matching result conflicts with the geometric consistency verification result, a multi-dimensional conflict resolution mechanism is triggered to generate corrected inventory data and trigger an early warning mechanism. The multi-dimensional conflict resolution mechanism includes: a priority rule base, which sets the judgment rule that the geometric consistency verification result takes precedence over the tag recognition result; and a dynamic threshold adjustment strategy, which adjusts the matching degree threshold in real time according to the intensity of environmental interference, which is high temperature or strong electromagnetic field.

5. The electronic component material identification and management method according to claim 2, characterized in that, In high-temperature industrial settings, the methods also include: Dynamic spatial alignment correction is performed between the surface normal vector field constructed from 3D point cloud data and the X-ray grayscale gradient to generate a thermal deformation compensation map. The thermal deformation compensation map is used as a weighting factor to dynamically adjust the fusion ratio of X-ray grayscale image and 3D point cloud data, where the geometric consistency of local deformation areas caused by thermal expansion is preserved first. In the feature-level fusion stage, a multi-dimensional feature matrix with thermal stress distribution constraints is generated by combining the thermal deformation compensation map and the weight data obtained by the weight sensor.

6. The electronic component material identification and management method according to claim 1, characterized in that, The method also includes: In strong electromagnetic interference scenarios, to address the problem of electromagnetic pulse interference to X-ray detector signals, motion state characteristics of electronic components are extracted from lidar point cloud data. These motion state characteristics include vibration frequency and displacement fluctuations. The motion state characteristics are input into the electromagnetic interference suppression module of the hybrid Kalman filter. By analyzing the temporal correlation between the motion state and the electromagnetic field strength, the periodic components of the electromagnetic interference are extracted, and wavelet transform is used to generate an interference spectrum mask. The measurement noise covariance matrix of the hybrid Kalman filter is dynamically adjusted based on the interference spectrum mask, and a high-fidelity image is output after motion state compensation and electromagnetic interference suppression. The measurement noise covariance matrix preferentially suppresses periodic electromagnetic noise.

7. The electronic component material identification and management method according to claim 1, characterized in that, In multi-dimensional conflict resolution mechanisms, methods also include: An anomaly pattern library is built based on historical inventory data, recording the corresponding common anomaly matching scenarios. Common anomaly matching scenarios include increased false recognition rate under high temperature environment and density distribution distortion caused by strong electromagnetic interference. When the multi-dimensional conflict resolution mechanism is triggered, the current conflict scenario is matched with the abnormal pattern library for similarity. If the match is successful, the preset correction strategy is directly invoked. If a match fails, a self-learning mode is activated, incorporating the current conflict scenario and correction results into the abnormal mode library.

8. An electronic component material identification and management system, characterized in that, include: The multidimensional feature matrix generation module acquires X-ray grayscale images of electronic components through X-ray imaging equipment, acquires corresponding three-dimensional point cloud data of electronic components through lidar, and collects corresponding weight information of electronic components through weight sensors; and fuses the X-ray grayscale images, the three-dimensional point cloud data and the weight information based on dynamic weight allocation of signal-to-noise ratio to generate a fused multidimensional feature matrix. The unique identifier generation module extracts the physical characteristic parameters corresponding to the electronic components based on the multi-dimensional feature matrix. The physical characteristic parameters include density, geometric symmetry, and surface curvature. It retrieves a pre-set material physical characteristic database, uses a multi-parameter joint matching algorithm to generate a unique identifier, and outputs the material type and specification information. The high-fidelity image output module integrates an adaptive filtering circuit at the front end of the X-ray detector to acquire the corresponding ambient temperature and electromagnetic field strength in real time, and adjusts the X-ray detector gain and sampling frequency based on the ambient temperature and electromagnetic field strength. Based on the three-dimensional point cloud data, the X-ray image distortion caused by high temperature or strong electromagnetic interference is corrected using a spatial correlation algorithm, and the corrected high-fidelity image is output. The verification module associates the unique identifier with the material database and automatically identifies the label information through structured OCR technology to complete the material warehousing, outbound and inventory dynamic updates. This is used to perform geometric consistency verification between the high-fidelity image and the multi-dimensional feature matrix. If there is a conflict between the material type, specification information and the database record, an abnormal matching result is determined. Based on the abnormal matching result, a secondary verification is performed, an early warning mechanism is triggered, and corrected inventory data is generated.

9. An electronic component material identification and management system, characterized in that, Includes a processor, wherein the processor runs a program of the electronic component material identification and management method as described in any one of claims 1-7.

10. A storage medium, characterized in that, The program stores the electronic component material identification and management method as described in any one of claims 1-7.

Images