An electronic component management system with serial number tracking

Abstract

The application discloses an electronic component management system with serial number tracking, and relates to the technical field of electronic component management.The system comprises a serial number identification generation module, a response data acquisition module, a defect candidate area positioning module, a defect type determination module, a risk assessment module and a traceability management module.The quality of components is classified into A-grade superior products, B-grade qualified products, C-grade degraded products and D-grade unqualified products according to risk indicators, so that fine quality classification is realized.The serial number of a component is bound to the serial number of a complete machine, so that when the complete machine fails, the associated component list can be reversely inquired, and the detection data and risk assessment results of each component can be called up, the possible problem source is quickly located in descending order of risk indicators, and a batch early warning mechanism is designed, so that when the abnormal proportion of components in the same batch exceeds a preset threshold, early warning is automatically triggered, and the risk prevention and control efficiency is improved.

Description

Technical Field

[0001] This invention belongs to the field of electronic component management technology, specifically an electronic component management system with serial number tracking. Background Technology

[0002] Electronic components are the foundational parts of the electronic information industry, and their quality and stability directly determine the reliability of the entire device. Currently, quality inspection and management in the manufacturing process of electronic components typically employs a batch management model for sampling inspection, relying on manual recording and general testing equipment.

[0003] However, existing technologies for evaluating the quality and tracking the flow of electronic components typically rely mainly on batch numbers for extensive management, generally ignoring the individual differences of individual components throughout the entire production, testing, and assembly process. This makes it difficult to accurately trace back to the original file of a specific component when the whole machine fails. At the same time, existing testing methods are limited in scope, often focusing only on surface appearance or static parameters, making it difficult to detect hidden defects with dynamic characteristics or microscopic physical field changes, such as microcracks in solder joints, internal pores, and delamination within the board. Optical indicators alone cannot accurately determine the potential failure risk. Based on this, an electronic component management system with serial number tracking is proposed. Summary of the Invention

[0004] The purpose of this invention is to provide an electronic component management system with serial number tracking to solve the problems mentioned in the background art.

[0005] An electronic component management system with serial number tracking includes: The serial number identification generation module generates a 32-bit serial number for the electronic component to be tested, solidifies the serial number onto the surface of the component to generate a serial number identification, and simultaneously creates an initial file in the database corresponding to the serial number by entering the basic information of the component. The response data acquisition module places electronic components with serial numbers on the testing platform, identifies the serial numbers and acquires component information. Based on the acquired component type code, it matches and loads the corresponding testing parameter configuration scheme from the preset parameter knowledge base. Based on the configuration scheme, it controls the testing platform to synchronously acquire response data of laser interference vibration signal, surface potential signal and acoustic emission signal. After the test is completed, the three types of data are bound to the serial number and stored in the database. The defect candidate region localization module extracts the vibration response detection data corresponding to the serial number from the database, selects multiple candidate reference regions on the surface of the component, calculates the coefficient of variation of the vibration data of the measurement points in each candidate reference region, selects the region with a coefficient of variation less than a preset threshold as the effective reference region, establishes the reference vibration response spectrum and standard deviation spectrum based on the vibration data of all measurement points in the effective reference region, calculates the degree of deviation between the vibration response spectrum of each measurement point and the reference spectrum as the vibration response anomaly, identifies and judges the anomaly points, and locates the defect candidate region through connected component analysis; The defect type determination module extracts vibration-related feature parameters from vibration response data, potential-related feature parameters from surface potential data, and acoustic emission-related feature parameters from acoustic emission event data for defect candidate regions, constructing a multi-physics field coupled feature vector. Based on hierarchical decision rules, the module determines the defect type. The first layer distinguishes between dynamically active defects and statically stable defects based on the relationship between acoustic emission event counts and a first preset threshold. The second layer further subdivides the defect type for dynamically active defects based on the relationship between potential contrast and frequency offset and their respective preset thresholds. The third layer further subdivides the defect type for statically stable defects based on the relationship between amplitude offset and potential standard deviation and their respective preset thresholds. The risk assessment module reads the defect type and characteristic parameters corresponding to the serial number from the database, calculates the geometric risk factor, vibration response risk factor, potential anomaly risk factor, and acoustic emission active risk factor respectively, and obtains the comprehensive defect risk index by weighting and summing the four risk factors according to preset weights. The defect risk level is determined according to the relationship between the comprehensive defect risk index and the preset risk threshold, and the overall quality is graded according to the risk level distribution of all defects on the component. The traceability management module executes corresponding processing procedures for components of different grades based on the quality grading results, and establishes full-process traceability management information based on serial numbers.

[0006] As a further aspect of the present invention: the specific method for generating a 32-bit serial number for the electronic component to be tested is as follows: The serial number consists of 32 digits, and its structure from left to right is as follows: digits 1 to 4 are the component type code, digits 5 to 10 are the production date code, digits 11 to 14 are the production line number, and digits 15 to 32 are the serial number. After the serial number is generated, the database is automatically queried. If the serial number already exists, the serial number is automatically incremented by 1 until an unused serial number is obtained.

[0007] As a further aspect of the present invention: after placing the electronic components with serial numbers on the detection platform, an identification timeout threshold is set. If the identification time exceeds the threshold, an alarm is triggered to prompt the components to be repositioned. The detection parameter configuration scheme includes ultrasonic excitation parameters, laser vibration measurement parameters, surface potential scanning parameters, and acoustic emission sampling parameters. Here, the identification timeout threshold is 0.3 seconds.

[0008] As a further aspect of the present invention, the specific method for identifying and determining anomalies is as follows: The total spectral deviation value Dspectrum between the measured spectrum and the reference spectrum is calculated. The Dspectrum is calculated as follows: for each frequency point, the absolute value of the difference between the measured spectrum and the reference spectrum is calculated, and then divided by the reference standard deviation to obtain the normalized deviation value for that frequency point. The normalized deviation values ​​of all frequency points are summed to obtain the Dspectrum. The main resonance peaks of the measured spectrum and the reference spectrum are identified, and the relative offsets of the resonance peak frequencies and amplitudes are calculated. The products of Dspectrum and the first preset weight, the relative offsets of the resonance peak frequencies and the second preset weight, and the relative offsets of the resonance peak amplitudes and the third preset weight are summed to obtain the vibration response anomaly degree V, where the first preset weight is 0.5, the second preset weight is 0.3, and the third preset weight is 0.2. The above calculation is repeated for all measurement points to obtain the anomaly degree matrix. The mean anomaly degree and the standard deviation (std) anomaly degree of all measurement points are statistically calculated. The anomaly degree threshold is set as the mean anomaly degree plus a preset multiple of 2 multiplied by the std anomaly degree. Measurement points with anomalies greater than this threshold are marked as anomalies.

[0009] As a further aspect of the present invention, the specific method for locating candidate defect regions through connected component analysis is as follows: First, an anomaly marker matrix with M rows and N columns is established. All measured points are marked according to their spatial positions, where i ranges from 1 to M, j ranges from 1 to N, and the element in the i-th row and j-th column corresponds to the (i-1)×N+j-th measured point. Initially, all elements of the matrix are 0, and the matrix elements corresponding to the identified anomalies are set to 1. For a measured point at position (i,j) in the marker matrix, its eight neighboring positions are: above (i-1,j), below (i+1,j), left (i,j-1), right (i,j+1), upper left (i-1,j-1), upper right (i-1,j+1), lower left (i+1,j-1), and lower right (i+1,j+1). Starting from the upper left corner of the marker matrix, the first unvisited anomaly with a value of 1 is scanned in row-major order as the seed. For each seed point, a new connected component is created and the seed point is added to that component. The eight neighboring locations of the seed point are checked. If the marker matrix element of a neighboring location is 1 and has not been visited, that neighboring location is added to the current connected component and marked as visited. At the same time, the eight neighboring locations of that neighboring location are recursively processed as new seed points. This process is repeated until the current connected component can no longer be expanded. The marker matrix is ​​then returned to continue scanning to find the next connected component. After all anomaly points have been visited, all connected components are output as defect candidate regions. For each connected component, the geometric center coordinates, area, equivalent circle diameter, and aspect ratio are calculated. The area of ​​the connected component is equal to the number of anomaly points contained in the connected component multiplied by the area represented by a single measurement point. The equivalent circle diameter is equal to the area of ​​the connected component divided by π (2 square roots of the area of ​​the connected component). The aspect ratio is the larger of the X-direction span and the Y-direction span of the connected component divided by the smaller value.

[0010] As a further aspect of the present invention, the specific method for determining defect types based on hierarchical decision rules is as follows: Vibration-related characteristic parameters include seven parameters: average vibration anomaly, peak vibration anomaly, average frequency offset, average amplitude offset, area, equivalent circle diameter, and aspect ratio. Potential-related characteristic parameters include four parameters: mean P, standard deviation P, contrast P, and gradient P. The mean P is the arithmetic mean of all potential values ​​within the defect area, the standard deviation P is the standard deviation of all potential values ​​within the defect area, the contrast P is the average potential value of the defect area minus the average potential value of the surrounding normal area, and the gradient P is the arithmetic mean of the potential gradient amplitudes at all measurement locations on the boundary of the defect area. Acoustic emission-related characteristic parameters include four parameters: acoustic emission event count N, total acoustic emission energy E, mean dominant frequency F, and standard deviation of dominant frequency. The acoustic emission event count N is the number of acoustic emission events selected and belonging to the current defect candidate area, the total acoustic emission energy E is the sum of the energies of all selected events, and the mean dominant frequency F is the arithmetic mean of the dominant frequencies of all selected events. The first-level judgment determines the defect as a dynamically active defect if the acoustic emission event count N is greater than a first preset threshold, and proceeds to the second-level judgment. Otherwise, it determines the defect as a statically stable defect, and proceeds to the third-level judgment. The first preset threshold is 15. The second-level judgment targets dynamically active defects. If the absolute value of the comparison P is greater than the second preset threshold, the average frequency offset is less than the third preset threshold, and the aspect ratio is less than the fourth preset threshold, it is determined to be a weld microcrack defect. If the absolute value of the comparison P is less than or equal to the second preset threshold, the total acoustic emission energy E is greater than the average acoustic emission energy of this batch of defects, and the aspect ratio is greater than the fifth preset threshold, it is determined to be a weld microcrack defect. The first layer is for copper foil breakage defects. The second preset threshold is 10, the third preset threshold is 0.05, the fourth preset threshold is 1.5, and the fifth preset threshold is 2. The third layer is for static stable defects. If the average amplitude offset is greater than the sixth preset threshold and the standard deviation P is greater than 1.5 times the standard deviation of the potential of the surrounding normal area, it is judged as a porosity defect. If the average frequency offset is less than the seventh preset threshold and the area is greater than the eighth preset threshold, it is judged as an intra-board delamination defect. The sixth preset threshold is 0.3, the seventh preset threshold is -0.05, and the eighth preset threshold is 100 square millimeters.

[0011] The specific method for determining the defect risk level based on the relationship between the comprehensive defect risk index and the preset risk threshold is as follows: The equivalent circle diameter is divided by the component thickness to obtain the relative defect size, where the equivalent circle diameter is equal to 2√2 times the defect area divided by pi, and the defect area is divided by the total area of ​​the component to obtain the defect area ratio. The relative defect size is divided by the theoretical value of 100 to obtain the normalized relative defect size. The product of the normalized relative defect size and the first geometric weight, the product of the defect area ratio multiplied by 10 and the second geometric weight are summed to obtain the geometric risk factor, where the first geometric weight is 0.6 and the second geometric weight is 0.4. The normalized vibration anomaly is obtained by dividing the average vibration anomaly by the average anomaly of all components. The average frequency offset is multiplied by 10 to obtain the frequency offset risk value, and the average amplitude offset is multiplied by 5 to obtain the amplitude offset risk value. The vibration response risk factor is obtained by summing the products of the normalized vibration anomaly and the third vibration weight, the frequency offset risk value and the fourth vibration weight, and the amplitude offset risk value and the fifth vibration weight, where the third vibration weight is 0.5, the fourth vibration weight is 0.3, and the fifth vibration weight is 0.2. The absolute value of the comparison P is used as the potential comparison risk value. The standard deviation P is divided by the potential standard deviation of the surrounding normal area as the potential standard deviation ratio. The gradient P is divided by the average potential gradient of all components as the normalized boundary potential gradient. The potential anomaly risk factor is obtained by summing the product of the potential comparison risk value and the first potential weight, the product of the potential standard deviation ratio and the second potential weight, and the product of the normalized boundary potential gradient and the third potential weight. The first potential weight is 0.5, the second potential weight is 0.3, and the third potential weight is 0.2. The acoustic emission event count N is divided by 10 to obtain the normalized acoustic emission count, the total acoustic emission energy E is divided by the average acoustic emission energy of this batch to obtain the normalized acoustic emission energy, and the average main frequency F is divided by 1000 kHz to obtain the normalized acoustic emission frequency. The products of the normalized acoustic emission count and the first acoustic emission weight, the normalized acoustic emission energy and the second acoustic emission weight, and the normalized acoustic emission frequency and the third acoustic emission weight are summed to obtain the acoustic emission activity risk factor, where the first acoustic emission weight is 0.5, the second acoustic emission weight is 0.3, and the third acoustic emission weight is 0.2. The comprehensive defect risk index is obtained by summing the products of geometric risk factor and weight w1, vibration response risk factor and weight w2, potential anomaly risk factor and weight w3, and acoustic emission active risk factor and weight w4, where w1 is 0.2, w2 is 0.3, w3 is 0.4, and w4 is 0.1. A defect is classified as high-risk when its comprehensive risk index is greater than or equal to the first risk threshold, as medium-risk when its comprehensive risk index is greater than or equal to the second risk threshold and less than the first risk threshold, and as low-risk when its comprehensive risk index is less than the second risk threshold. The first risk threshold is 7.0 and the second risk threshold is 4.0.

[0012] The specific method for overall quality classification based on the risk level distribution of all defects on components is as follows: If a component has no defects or only low-risk defects with a comprehensive risk index less than the third risk threshold, and the number of such defects does not exceed the first quantity threshold, and the total area of ​​such defects is less than the first area ratio threshold of the total area of ​​the component, then the component is classified as Grade A (superior). If there are no high-risk defects, the number of medium-risk defects does not exceed the second quantity threshold, the area of ​​a single medium-risk defect is less than the first area threshold, the total area of ​​medium-risk defects is less than the second area ratio threshold of the total area of ​​the component, and the comprehensive risk index of all medium-risk defects is less than the fourth risk threshold, then the component is classified as Grade B (qualified). If there are no high-risk defects but the number of medium-risk defects exceeds the third risk threshold, then the component is classified as Grade B (qualified). If the area of ​​a single medium-risk defect is greater than or equal to the first area threshold, or the total area of ​​medium-risk defects is greater than or equal to the second area ratio threshold, the component is classified as a Grade C downgraded product. If any high-risk defect or the total area of ​​medium-risk defects exceeds the third area ratio threshold, the component is classified as a Grade D non-conforming product. The third risk threshold is 2.0, the first quantity threshold is 3, the first area ratio threshold is 0.5%, the second quantity threshold is 2, the first area threshold is 100 square millimeters, the second area ratio threshold is 5%, the fourth risk threshold is 6.5, and the third area ratio threshold is 10%.

[0013] As a further aspect of the present invention: the specific method for performing corresponding processing procedures on components of different grades based on the quality grading results is as follows: For Grade A superior products and Grade B qualified products, generate labels and allow the status to be changed to qualified and ready for warehousing; for Grade C downgraded products, generate disposal orders and update the status to downgraded and ready for re-inspection; for Grade D unqualified products, generate isolation orders and update the status to unqualified and isolated, prohibiting warehousing confirmation or assembly binding.

[0014] As a further aspect of the present invention, the specific method for establishing full-process traceability management information based on serial numbers is as follows: Based on the production batch number of the faulty component, the database is searched for all component serial numbers in the same batch. The proportion of components with quality grade C or below in the batch is calculated as the batch anomaly ratio. When the batch anomaly ratio exceeds the preset warning threshold, a batch warning record is automatically generated. The record fields include batch number, anomaly ratio, statistical time, and trigger serial number. The record is pushed to the quality management account and marked as risk monitoring status in the batch file. The preset warning threshold is 8%.

[0015] Compared with the prior art, the beneficial effects of the present invention are: This invention employs a three-layer hierarchical decision-making rule for defect type determination. The first layer utilizes the dynamic response characteristics of acoustic emission signals to classify defects into two main categories: dynamically active and statically stable. The second and third layers further subdivide these two categories into four specific types: solder joint microcracks, copper foil fractures, porosity defects, and in-board delamination. The determination parameters for each layer are strongly correlated with the physical mechanism of the defect, improving the accuracy of defect type identification. By classifying component quality into Grade A (superior), Grade B (qualified), Grade C (downgraded), and Grade D (unqualified) based on risk indicators, refined quality grading and differentiated handling are achieved. By binding component serial numbers to the overall machine serial number, when the overall machine malfunctions, the associated component list can be queried in reverse, and the test data and risk assessment results of each component can be retrieved. The results are then sorted from high to low risk indicators to quickly locate the possible source of the problem. Simultaneously, when the abnormal proportion of components in the same batch exceeds a preset threshold, an early warning is automatically triggered, improving the efficiency of risk prevention and control. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the method framework structure of the present invention. Detailed Implementation

[0017] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Example 1: Please refer to Figure 1 This application provides an electronic component management system with serial number tracking, including: The serial number identification generation module generates a 32-bit serial number for the electronic components to be inspected. The serial number is physically imprinted onto the surface of the component to generate the serial number identification. At the same time, an initial file corresponding to the serial number is created in the database, and the basic information of the component is entered. Here, the permission to collect serial numbers is enabled by default when the component enters the production line, and the serial number information is only used for quality traceability analysis. The 32-bit serial number structure, from left to right, is as follows: bits 1 to 4 are the component type code, where PCB board is 1001, chip is 2001, and resistor is 3001; bits 5 to 10 are the production date code, using a 6-digit format of year, month, and day; bits 11 to 14 are the production line number, a 4-digit number; and bits 15 to 32 are the serial number, an 18-digit number that increases sequentially from 0000000000000000001. For example, serial number 1001241215L001000000000000012345 represents the 12345th PCB board produced on production line 1 on December 15, 2024. The uniqueness verification method for the serial number is: the system automatically queries the database, and if the serial number already exists, the serial number is automatically incremented by 1 until an unused serial number is obtained.

[0019] When generating a new serial number, the system automatically queries the central database. If the serial number already exists, the serial number is automatically incremented by 1 until an unused serial number is obtained. Otherwise, no processing is performed to ensure the uniqueness of the serial number. Create an initial file for this serial number, entering the component type, rated parameters, product model, physical dimensions, rated parameters, production batch number, supplier code, production line number, production time, operator ID, and current status. Component types include PCB boards, chips, resistors, connectors, etc., selected from a predefined type library; product model refers to the specific model number, such as "STM32F407VGT6" chip "FR-4 double-layer board"; physical dimensions include length, width, and thickness, such as PCB board dimensions of 100 mm x 80 mm x 1.6 mm; rated parameters include electrical parameters such as operating voltage, operating current, and power rating; production batch number refers to the batch identifier of products in the same batch, used for batch traceability; supplier code refers to the unique identifier of the supplier of raw materials or purchased components; production line number refers to the identifier of the production line that manufactured this component and its current status; production time refers to the production completion timestamp accurate to the second; operator ID refers to the unique identifier of the employee responsible for production; current status refers to the initial status as pending inspection.

[0020] Each electronic component's serial number is linked to its corresponding initial file, creating a unique digital identity for each component. This identity persists throughout the component's lifecycle, from production and testing to assembly and maintenance, ensuring complete traceability of quality information. Recording is done using physical QR codes or RFID tags, adapting to the size and material characteristics of different types of components. This ensures the reliability of the identification while avoiding any impact on the component's functionality.

[0021] The response data acquisition module places electronic components with serial numbers on the testing platform. It automatically identifies the serial numbers and queries the database to obtain component information through an industrial camera or RFID reader. Based on the component type code, it retrieves the corresponding testing parameter configuration scheme from the preset parameter knowledge base and controls the testing platform to synchronously collect response data of laser interference vibration signal, surface potential signal and acoustic emission signal. After the test is completed, the three types of data are bound to the serial number and stored in the database in real time. After the components are placed on the testing platform, the industrial camera or RFID reader automatically identifies the serial number. The system is set to an identification timeout threshold of 0.3 seconds. If the identification time exceeds this threshold, an alarm will be triggered to prompt the components to be replaced. The system sends a query request to the central database via gigabit Ethernet. The database query response timeout threshold is set to 0.1 seconds, and returns information such as the component's type code, physical dimensions, and material parameters.

[0022] The specific method for retrieving the detection parameter configuration scheme based on the type code is as follows: For PCB circuit boards (type code 1001), the testing platform automatically configures the ultrasonic excitation parameters: the ultrasonic excitation frequency range is 20 kHz to 150 kHz and divided into 51 frequency points; the laser vibration measurement parameters are: the laser vibration measurement point array is set to 11 points in the X direction multiplied by 9 points in the Y direction, equaling 99 measurement points; the surface potential scanning parameters are: the surface potential scanning step size is set to 0.2 mm; and the acoustic emission sampling parameters are: the acoustic emission sampling frequency is set to 20 MHz. For chips (type code 2001), the testing platform automatically configures the ultrasonic excitation frequency range to 50 kHz to 300 kHz, and the laser vibration measurement point array density is increased to one measurement point every 2 mm.

[0023] The testing platform has a surface size of 500 mm by 500 mm. 400 piezoelectric ceramic vibration units are embedded beneath the platform, arranged in 20 rows by 20 columns to form an ultrasonic vibration exciter array. A scanning laser Doppler vibrometer is installed directly above, with a laser wavelength of 632.8 nm and a velocity range of 0.01 mm / s to 10 m / s. A 16-channel Kelvin probe array is installed on the side, with probe diameters of 50 micrometers and a probe-to-component distance of 30 micrometers. Four acoustic emission sensors are installed on each of the four sides of the testing platform, totaling 16 sensors. The frequency response range is 50 kHz to 1 MHz, with a sampling frequency of 20 MHz. An atomic clock is used as the master clock source to achieve microsecond-level time synchronization.

[0024] Taking a 100mm x 80mm x 1.6mm double-layer PCB board as an example, the testing process is as follows: the tabletop vacuum pump starts to adsorb the PCB board and reaches the preset negative pressure value of -10 kPa. After that, the ultrasonic exciter starts to sweep the frequency to excite multiple frequency points. Each frequency point lasts for a preset time of 0.08 seconds. The total time is determined according to the number of frequency points. The laser vibrometer simultaneously collects data from multiple measurement points. The sampling frequency is a preset value of 1 MHz. The Kelvin probe scans along the X direction to collect surface potential data. The scanning step size is a preset value of 0.2 mm. The acoustic emission sensor continuously collects signals throughout the process.

[0025] After testing, the system binds and stores three types of data with serial numbers: the vibration response data table stores multiple records containing measurement point coordinates, frequency points, and vibration velocity amplitudes; the surface potential data table stores multiple records containing X-coordinates, Y-coordinates, and surface potential values; and it stores each triggered acoustic emission event, with each event record containing sound source coordinates, event energy, and dominant frequency, forming an acoustic emission event data table. The system automatically verifies the integrity of the records for that serial number in the three data tables and updates the component status to "Test Complete, Awaiting Analysis."

[0026] Through automatic serial number identification and adaptive configuration of detection parameters, the system achieves personalized detection for each individual component, automatically matching the optimal detection parameters for different types of components. Multi-physics collaborative detection technology overcomes the limitations of traditional single-physical-quantity detection: vibration signals reflect changes in the mechanical properties of components, surface potential signals reflect changes in electrical properties, and acoustic emission signals reflect the dynamic activity of defects. The fusion analysis of these three physical field signals significantly improves the detection rate of hidden defects. Real-time binding of detection data with serial numbers ensures the traceability of quality information, providing a complete data foundation for subsequent defect analysis and risk assessment.

[0027] The defect candidate region localization module extracts the vibration response detection data corresponding to the serial number from the database, selects multiple candidate reference regions on the surface of the component, calculates the coefficient of variation of the vibration data of the measurement points in each candidate reference region, selects the region with a coefficient of variation less than a preset threshold as the effective reference region, establishes the reference vibration response spectrum and standard deviation spectrum based on the vibration data of all measurement points in the effective reference region, calculates the degree of deviation between the vibration response spectrum of each measurement point and the reference spectrum as the vibration response anomaly degree, identifies and judges the anomaly points, and locates the defect candidate region through eight-neighbor connected domain analysis; The specific method for selecting multiple candidate reference areas on the surface of components is as follows: For PCB circuit boards, select one area with a preset size of 10 mm by 10 mm in each of the four corner areas and the center area as a candidate reference area; for chips, the same method is used, selecting one candidate reference area in the center area and one in each of the four edge areas. For each candidate reference region, the overall coefficient of variation is calculated to verify whether it is a defect-free uniform region. The overall coefficient of variation is calculated as follows: for each excitation frequency point, the average value of the vibration velocity amplitude of N measurement points in the region is calculated. The square root of the sum of the squares of the differences between the vibration velocity amplitude of N measurement points and the average value is divided by N to obtain the standard deviation of the frequency point. The standard deviation is divided by the average value to obtain the coefficient of variation of the frequency point. The mean of the coefficients of variation of all excitation frequency points is used as the overall coefficient of variation of the candidate reference region. Among the candidate reference regions, the candidate reference regions with an overall coefficient of variation less than the preset threshold of 0.1 are selected as valid reference regions.

[0028] For example, candidate reference region 1 contains 4 measurement points. The vibration velocity amplitudes at the 30th excitation frequency are 3.1, 3.3, 2.9, and 3.2 mm / s, respectively. The arithmetic mean is the sum of 3.1, 3.3, 2.9, and 3.2 divided by 4, which equals 3.125 mm / s. The standard deviation is the square root of the sum of the squares of the differences between the four amplitudes and the mean, divided by 4, which is approximately 0.15 mm / s. The coefficient of variation at this frequency point is 0.15 divided by 3.125, which equals 0.048. After calculating and averaging all 51 frequency points, assuming the overall coefficient of variation is 0.052, which is less than 0.1, it is confirmed as a valid reference region. It is assumed that 4 out of the 5 candidate reference regions pass the verification.

[0029] To establish a reference spectrum, the vibration data of all measurement points within four effective reference regions are statistically analyzed. Assuming that the four effective reference regions contain a total of 16 measurement points, for the k-th excitation frequency point, the arithmetic mean of the vibration velocity amplitude of the 16 measurement points at that frequency point is calculated as the reference vibration response amplitude SA reference k, and the standard deviation of the vibration velocity amplitude of the 16 measurement points at that frequency point is calculated as the reference standard deviation SB reference k. The reference spectrum and standard deviation spectrum of all frequency points are obtained by repeating the calculation for all frequency points.

[0030] For each measurement point, calculate the vibration response anomaly: extract the vibration velocity amplitude of that measurement point at all frequency points to form the measurement spectrum Smeasurement, and calculate the total spectral deviation Dspectrum. Dspectrum is calculated by taking the absolute value of the difference between the measurement spectrum Smeasurementk and the reference spectrum SAreferencek for each frequency point k, dividing it by the reference standard deviation σreferencek to obtain the normalized deviation value for that frequency point, and summing the normalized deviation values ​​for all frequency points to obtain Dspectrum. Identify the main resonance peaks of the measurement spectrum and the reference spectrum. Here, the criteria for identifying the main resonance peak are that the vibration velocity amplitude at that frequency point is both a local maximum (greater than the amplitudes of the three adjacent frequency points to the left and right) and satisfies the significance condition (greater than the average amplitude of the entire spectrum). The average amplitude is 1.5 times. The relative offset of the resonant peak frequency is calculated by dividing the absolute value of the difference between the main resonant peak frequency of the measured spectrum and the main resonant peak frequency of the reference spectrum by the main resonant peak frequency of the reference spectrum. The relative offset of the resonant peak amplitude is calculated by dividing the absolute value of the difference between the main resonant peak amplitude of the measured spectrum and the main resonant peak amplitude of the reference spectrum by the main resonant peak amplitude of the reference spectrum. The vibration response anomaly V is calculated by multiplying the D spectrum by the first preset weight, adding the relative offset of the resonant peak frequency by the second preset weight, and adding the relative offset of the resonant peak amplitude by the third preset weight. The sum of the three preset weights is 1, where the first preset weight is 0.5, the second preset weight is 0.3, and the third preset weight is 0.2.

[0031] Repeat the above calculation for all measurement points to obtain the anomaly matrix. Calculate the mean anomaly and the standard deviation (std) anomaly of all measurement points. Set the anomaly threshold to mean anomaly plus a preset multiple multiplied by std anomaly, where the preset multiple is 2. Mark measurement points with anomalies greater than this threshold as anomalies. Perform eight-neighbor connected component analysis to merge adjacent anomalies into the same defect candidate region.

[0032] Eight-neighbor connectivity analysis merges adjacent outliers into a single defect candidate region. The specific steps of eight-neighbor connectivity analysis are as follows: An anomaly marker matrix is ​​established, with M rows multiplied by N columns corresponding to all measurement points arranged according to their spatial positions. For example, 11 rows multiplied by 9 columns correspond to 99 measurement points. The element in the i-th row and j-th column corresponds to the (i-1)×N+j-th measurement point, where i ranges from 1 to M and j ranges from 1 to N. Initially, all elements of the matrix are 0, and the matrix elements corresponding to the previously identified anomalies are set to 1. For a measurement point at position (i,j) in the marker matrix, its eight neighboring positions are above (i-1,j), below (i+1,j), left (i,j-1), right (i,j+1), upper left (i-1,j-1), upper right (i-1,j+1), lower left (i+1,j-1), and lower right (i+1,j+1). If i-1 is less than 1, or i+1 is greater than M, or j-1 is less than 1, or j+1 is greater than M, then the anomaly marker matrix is ​​established. If 1 is greater than N, the neighborhood position is outside the matrix range and is marked as invalid and not included in the connectivity determination; scan from the top left corner of the marking matrix in row priority order, find the first unvisited outlier with an element value of 1 as the seed point, create a new connected component number and add the seed point to the connected component, and mark the seed point as visited; check the eight neighborhood positions of the seed point, if the marking matrix element of the neighbor position is 1 and has not been visited, add the neighbor position to the current connected component and mark it as visited, and recursively process its eight neighbors as a new seed point; repeat the neighborhood checking process until all eight neighborhood positions of the outliers added to the connected components have been visited or have an element value of 0; return to the marking matrix to continue scanning to find the next unvisited outlier with an element value of 1 and establish the next connected component; after all outliers have been visited, output all connected components.

[0033] For each connected component, calculate the following parameters: the X-coordinate of the geometric center equals the sum of the X-coordinates of all outliers in the connected component divided by the number of outliers; the Y-coordinate of the geometric center equals the sum of the Y-coordinates of all outliers in the connected component divided by the number of outliers; the area of ​​the connected component equals the number of outliers in the connected component multiplied by the area represented by a single measurement point. The area represented by a single measurement point is calculated as follows: the spacing between measurement points in the X direction equals the X-direction length of the PCB board (100 mm) divided by (the number of measurement points in the X direction minus 1); the spacing between measurement points in the Y direction equals the Y-direction width of the PCB board (80 mm) divided by (the number of measurement points in the Y direction minus 1). 1) The area represented by a single measurement point is equal to the distance between measurement points in the X direction multiplied by the distance between measurement points in the Y direction. For example, if the distance between measurement points in the X direction is 10 mm and the distance between measurement points in the Y direction is 10 mm, the area represented by a single measurement point is 100 square millimeters. The diameter of the equivalent circle of a connected region is equal to the square root of 2 multiplied by the area of ​​the connected region divided by pi. The aspect ratio of a connected region is the larger of the span in the X direction and the span in the Y direction divided by the smaller value. The span in the X direction is equal to the maximum value minus the minimum value of the X coordinates of all anomaly points in the connected region, and the span in the Y direction is equal to the maximum value minus the minimum value of the Y coordinates of all anomaly points in the connected region. The identified defect candidate regions and their geometric feature parameters are temporarily stored, including the defect candidate region number, the list of anomaly points contained therein, the coordinates of the geometric center, the area, the diameter of the equivalent circle, and the aspect ratio.

[0034] Multiple candidate reference regions are selected on the surface of the component. The coefficient of variation of vibration data at measurement points within each region is calculated. Defect-free, uniform regions with a coefficient of variation less than a preset threshold are selected as valid reference regions. Based on this, a reference vibration response spectrum and standard deviation spectrum are established. The vibration response anomaly degree of each measurement point is calculated, and anomaly points are marked with a threshold of the mean plus twice the standard deviation. Adjacent anomaly points are merged into defect candidate regions through eight-neighbor connected domain analysis, and parameters such as geometric center, area, equivalent circle diameter, and aspect ratio are output. The coefficient of variation verification mechanism adaptively selects defect-free reference regions, avoiding reference distortion; the adaptive threshold mechanism dynamically adapts to the data distribution, accurately distinguishing between normal fluctuations and abnormal deviations; and the eight-neighbor connected domain analysis accurately locates the defect contour and geometric features, achieving precise defect localization.

[0035] The defect type determination module extracts vibration-related feature parameters from vibration response data, potential-related feature parameters from surface potential data, and acoustic emission-related feature parameters from acoustic emission event data for defect candidate regions, constructing a multi-physics field coupled feature vector. Based on hierarchical decision rules, it determines the defect type. The first layer distinguishes between dynamically active and statically stable defects based on the relationship between acoustic emission event counts and a first preset threshold. The second layer further subdivides the defect type for dynamically active defects based on the relationship between parameters such as potential contrast and frequency offset and their respective preset thresholds. The third layer further subdivides the defect type for statically stable defects based on the relationship between parameters such as amplitude offset and potential standard deviation and their respective preset thresholds. For the defect candidate region, seven vibration feature parameters are extracted: average vibration anomaly degree is the arithmetic mean of the vibration anomalies of all abnormal points in the defect region; peak vibration anomaly degree is the maximum value of the vibration anomalies of all abnormal points in the defect region; average frequency offset is the arithmetic mean of the relative offsets of the resonant peak frequencies of all abnormal points in the defect region; average amplitude offset is the arithmetic mean of the relative offsets of the resonant peak amplitudes of all abnormal points in the defect region; area is the number of abnormal points contained in the defect region multiplied by the area represented by a single measurement point; equivalent circle diameter is the square root of 2 multiplied by the defect area divided by pi (3.1416); and aspect ratio is the larger of the X-direction span and the Y-direction span of the defect region divided by the smaller value.

[0036] The potential values ​​of the corresponding locations of the defect candidate regions are extracted from the surface potential data table. Specifically, for the geometric center coordinates and equivalent circle radius of the defect candidate regions, measurement locations in the surface potential data table that meet the following conditions are selected: the square root of the sum of the square of the difference between the X-coordinate of the measurement location and the X-coordinate of the geometric center plus the square of the difference between the Y-coordinate of the measurement location and the Y-coordinate of the geometric center is less than or equal to the equivalent circle radius. The potential values ​​of all measurement locations that meet this condition are extracted to form the defect region potential dataset. Simultaneously, the potential dataset of the surrounding normal region is extracted. The surrounding normal region is defined as an annular region whose distance from the geometric center is greater than the equivalent circle radius but less than the equivalent circle radius plus a preset width of 5 mm. The potential values ​​of all measurement locations in the surrounding normal region are selected to form the surrounding normal region potential dataset, and four potential-related feature parameters are extracted. The following are the values: mean P, standard deviation P, contrast P, and gradient P: mean P is the arithmetic mean of all potential values ​​within the defect area; standard deviation P is the standard deviation of all potential values ​​within the defect area; contrast P is the average potential value of the defect area minus the average potential value of the surrounding normal area, then divided by the standard deviation of the potential values ​​of the surrounding normal area; gradient P is calculated as follows: for each measurement location on the boundary of the defect area, find the potential values ​​of the adjacent locations to the right and left in the surface potential data table, calculate the potential gradient in the X direction, find the potential values ​​of the adjacent locations above and below, calculate the potential gradient in the Y direction, and the magnitude of the potential gradient at this location is equal to the square root of the sum of the squares of the potential gradients in the X and Y directions; the gradient P is obtained by taking the arithmetic mean of the potential gradient magnitudes at all measurement locations on the boundary of the defect area. Events whose distance from the sound source coordinates to the geometric center of the defect is less than or equal to the defect radius are selected from the acoustic emission event data table. The selection method is as follows: calculate the Euclidean distance from the sound source coordinates (X sound source, Y sound source) of each event in the acoustic emission event data table to the geometric center of the defect. If the Euclidean distance is less than or equal to the radius of the equivalent circle, the event belongs to the current defect candidate region. Four acoustic emission feature parameters are extracted, namely acoustic emission N, acoustic emission E, mean F, and standard deviation F. Among them, acoustic emission N is the number of acoustic emission events selected, acoustic emission E is the sum of the energy of all selected events, mean F is the arithmetic mean of the main frequencies of all selected events, and standard deviation F is the standard deviation of the main frequencies of all selected events.

[0037] The 15-dimensional multiphysics coupling feature vector is constructed as follows: (mean vibration anomaly, peak vibration anomaly, mean frequency offset, mean amplitude offset, area, equivalent circle diameter, aspect ratio, mean P, standard deviation P, contrast P, gradient P, acoustic emission N, acoustic emission E, mean F, standard deviation F).

[0038] Defect type determination is based on hierarchical decision rules, and the defect types include four categories: solder joint microcracks, copper foil fractures, in-board delamination, and porosity defects. The first level of judgment is dynamic activity classification. If the acoustic emission N is greater than the first preset threshold of 15, it is judged as a dynamic active defect and enters the second level of judgment. Otherwise, it is judged as a static stable defect and enters the third level of judgment. The judgment basis is that there is an interface or crack surface inside the dynamic active defect that can undergo micro-relative motion, and a detectable acoustic emission signal is generated under the action of alternating stress excited by ultrasonic vibration.

[0039] The second layer of judgment is a dynamic active type of defect subdivision. If the absolute value of the comparison P is greater than the second preset threshold of 10, the average frequency offset is less than the third preset threshold of 0.05, and the aspect ratio is less than the fourth preset threshold of 1.5, it is judged as a solder joint microcrack defect. The judgment basis is that the solder joint crack causes the contact resistance between the tin-lead alloy and the copper foil to increase, resulting in an abnormally significant surface potential, and the crack position is relatively concentrated and circular. If the absolute value of the comparison P is less than or equal to 10, the acoustic emission E is greater than the average acoustic emission energy of this batch of defects, and the aspect ratio is greater than the fifth preset threshold of 2, it is judged as a copper foil fracture defect. The judgment basis is that the copper foil fracture is linear with a large aspect ratio, and the fracture surface repeatedly contacts under stress to generate strong acoustic emission.

[0040] The third layer is a subdivision of statically stable defects. If the average amplitude offset is greater than the sixth preset threshold of 0.3 and the standard deviation P is greater than 1.5 times the standard deviation of the potential in the surrounding normal area, it is judged as a porosity defect. The judgment basis is that porosity causes a decrease in local mass density, which increases the vibration amplitude. At the same time, the uneven stress around the porosity increases the potential dispersion. If the average frequency offset is less than the seventh preset threshold of -0.05 and the area is greater than the eighth preset threshold of 100 square millimeters, it is judged as an intraplate delamination defect. The judgment basis is that delamination causes a decrease in local stiffness, which decreases the resonant frequency. Delamination defects usually have a large area.

[0041] The defect identification results are stored in the defect file table of the database, which includes information such as serial number, defect number, defect type, location coordinates, and 15-dimensional feature parameters. The component status is then updated to "defect identification completed, awaiting risk assessment".

[0042] The three-layer hierarchical decision rule is designed based on the differences in the response mechanisms of different physical field signals to different defect types. It avoids potential conflicts that might arise from parallel multi-condition decisions through a sequential, progressive decision-making process. The first layer uses the dynamic response characteristics of acoustic emission signals to classify defects into two main categories. The second and third layers further subdivide these two categories. The decision parameters for each layer are strongly correlated with the physical mechanism of that type of defect, ensuring accuracy. The fusion of multi-physical field features overcomes the information limitations of single physical quantity features. Vibration features reflect the impact of defects on structural mechanical properties, potential features reflect the impact of defects on electrical properties, and acoustic emission features reflect the dynamic activity and expansion trend of defects. These three features complement and verify each other, improving the accuracy of defect type identification.

[0043] The risk assessment module reads the defect type and 15-dimensional feature parameters corresponding to the serial number from the database, calculates the risk factors in four dimensions: geometric risk factor, vibration response risk factor, potential anomaly risk factor, and acoustic emission active risk factor, calculates the comprehensive defect risk index based on the four dimensions of risk factors, determines the defect risk level by comparing the risk index with the preset risk threshold, and performs overall quality grading based on the risk level distribution of all defects on the component. The equivalent circle diameter divided by the component thickness is used as the relative defect size, the defect area divided by the total component area is used as the defect area ratio, the relative defect size divided by the theoretical value of 100 is used as the normalized relative defect size, and the relative defect size divided by 100 multiplied by the first geometric weight of 0.6 plus the defect area ratio multiplied by 10 multiplied by the second geometric weight of 0.4 is used as the geometric risk factor. That is, geometric risk factor = normalized relative defect size × first geometric weight + defect area ratio × 10 × second geometric weight. The maximum value of geometric risk factor is limited to 10, where the first geometric weight is 0.6, the second geometric weight is 0.4, and the sum of the first geometric weight and the second geometric weight is 1. The normalized vibration anomaly is obtained by dividing the average vibration anomaly by the average anomaly of all components. The average frequency offset is multiplied by 10 to obtain the frequency offset risk value. The average amplitude offset is multiplied by 5 to obtain the amplitude offset risk value. The vibration response risk factor is obtained by multiplying the normalized vibration anomaly by the third vibration weight (0.5), adding the frequency offset risk value multiplied by the fourth vibration weight (0.3), and adding the amplitude offset risk value multiplied by the fifth vibration weight (0.2). That is, the vibration response risk factor = normalized vibration anomaly × third vibration weight + frequency offset risk value × fourth vibration weight + amplitude offset risk value × fifth vibration weight. The sum of the above three weights is 1, where the third vibration weight is 0.5, the fourth vibration weight is 0.3, and the fifth vibration weight is 0.2.

[0044] The absolute value of the comparison P is used as the potential comparison risk value. The standard deviation P divided by the standard deviation of the potential in the surrounding normal area is used as the potential standard deviation ratio. The gradient P divided by the average potential gradient of all components is used as the normalized boundary potential gradient. The potential comparison risk value multiplied by the first potential weight 0.5, plus the potential standard deviation ratio multiplied by the second potential weight 0.3, plus the normalized boundary potential gradient multiplied by the third potential weight 0.2, is used as the potential anomaly risk factor. The potential anomaly risk factor = potential comparison risk value × first potential weight + potential standard deviation ratio × second potential weight + normalized boundary potential gradient × third potential weight. The maximum value of the potential anomaly risk factor is limited to 10. The sum of the above three weights is 1, where the first potential weight is 0.5, the second potential weight is 0.3, and the third potential weight is 0.2.

[0045] The acoustic emission N is divided by 10 to obtain the normalized acoustic emission count. The acoustic emission E is divided by the average acoustic emission energy of this batch to obtain the normalized acoustic emission energy. The mean F is divided by 1000 kHz to obtain the normalized acoustic emission frequency. The normalized acoustic emission count is multiplied by the first acoustic emission weight 0.5, plus the normalized acoustic emission energy multiplied by the second acoustic emission weight 0.3, plus the normalized acoustic emission frequency multiplied by the third acoustic emission weight 0.2 to obtain the acoustic emission activity risk factor. That is, the acoustic emission activity risk factor = normalized acoustic emission count × first acoustic emission weight + normalized acoustic emission energy × second acoustic emission weight + normalized acoustic emission frequency × third acoustic emission weight. The sum of the above three weights is 1, where the first acoustic emission weight is 0.5, the second acoustic emission weight is 0.3, and the third acoustic emission weight is 0.2.

[0046] The comprehensive defect risk index is calculated by multiplying the geometric risk factor by its weight w1, adding the vibration response risk factor by its weight w2, the potential anomaly risk factor by its weight w3, and the acoustic emission active risk factor by its weight w4. That is, the comprehensive defect risk index = geometric risk factor × w1 + vibration response risk factor × w2 + potential anomaly risk factor × w3 + acoustic emission active risk factor × w4. Here, the weights w1, w2, w3, and w4 are 0.2, 0.3, 0.4, and 0.1, respectively.

[0047] A defect is classified as high-risk when the comprehensive defect risk index is greater than or equal to the first risk threshold of 7.0; a defect is classified as medium-risk when the comprehensive defect risk index is greater than or equal to the second risk threshold of 4.0 and less than the first risk threshold of 7.0; and a defect is classified as low-risk when the comprehensive defect risk index is less than 4.0.

[0048] The overall quality is graded based on the risk level distribution of all defects on the component: If there are no defects or only low-risk defects with a comprehensive risk index less than the third risk threshold of 2.0, and the number of such defects does not exceed the first quantity threshold of 3, and the total area of ​​such defects is less than the first area ratio threshold of 0.5% of the total area of ​​the component, the component is classified as Grade A (superior). If there are no high-risk defects, the number of medium-risk defects does not exceed the second quantity threshold of 2, the area of ​​a single medium-risk defect is less than the first area threshold of 100 square millimeters, the total area of ​​medium-risk defects is less than the second area ratio threshold of 5% of the total area of ​​the component, and the comprehensive risk index of all medium-risk defects is less than the fourth risk threshold of 6.5, the component is classified as Grade B (qualified). If there are no high-risk defects but the number of medium-risk defects exceeds the second quantity threshold of 2, or the area of ​​a single medium-risk defect is greater than or equal to the first area threshold of 100 square millimeters, or the total area of ​​medium-risk defects is greater than or equal to the second area ratio threshold of 5%, the component is classified as Grade C (downgraded). If there are any high-risk defects or the total area of ​​medium-risk defects exceeds the third area ratio threshold of 10%, the component is classified as Grade D (unqualified).

[0049] The risk assessment results are stored in the database risk assessment table, which includes information such as serial number, quality level, comprehensive risk index, values ​​of each risk factor, and handling suggestions. The component status is then updated to indicate that the risk assessment is complete.

[0050] The comprehensive defect risk index is calculated by multiplying the geometric risk factor by its weight w1, adding the vibration response risk factor by its weight w2, the potential anomaly risk factor by its weight w3, and the acoustic emission active risk factor by its weight w4. Here, the weights w1, w2, w3, and w4 are 0.2, 0.3, 0.4, and 0.1, respectively.

[0051] A defect is classified as high-risk when the overall defect risk index is greater than or equal to 7.0, medium-risk when the overall defect risk index is greater than or equal to 4.0 and less than 7.0, and low-risk when the overall defect risk index is less than 4.0.

[0052] The overall quality is graded based on the risk level distribution of all defects on the component: If there are no defects or only low-risk defects with a comprehensive risk index of less than 2.0, and the number of such defects does not exceed 3, and the total area of ​​such defects is less than 0.5% of the total area of ​​the component, the component is classified as Grade A (superior). If there are no high-risk defects, the number of medium-risk defects does not exceed 2, the area of ​​a single medium-risk defect is less than 100 square millimeters, the total area of ​​medium-risk defects is less than 5% of the total area of ​​the component, and the comprehensive risk index of all medium-risk defects is less than 6.5, the component is classified as Grade B (qualified). If there are no high-risk defects but the number of medium-risk defects exceeds 2, or the area of ​​a single medium-risk defect is greater than or equal to 100 square millimeters, or the total area of ​​medium-risk defects is greater than or equal to 5%, the component is classified as Grade C (downgraded). If there are any high-risk defects or the total area of ​​medium-risk defects exceeds 10%, the component is classified as Grade D (unqualified).

[0053] The risk assessment results are stored in a database risk assessment table, which includes information such as serial number, quality level, comprehensive risk index, values ​​of each risk factor, and disposal recommendations. The component status is updated to indicate that the risk assessment is complete. The impact of defects on component performance is comprehensively quantified through four-dimensional risk factors, and the comprehensive index achieves scientific rating of single defects. The overall quality grading rules are in line with actual production, providing a clear and quantitative basis for differentiated disposal decisions.

[0054] The traceability management module implements differentiated handling procedures for components of different grades based on the quality grading results, and establishes full-process traceability management based on serial numbers; For Grade A superior products and Grade B qualified products, the system automatically calls the label generation module after the risk assessment is completed. It generates a data file containing the serial number, quality grade, inspection completion time, and QR code image according to a preset label template. The QR code content is the string encoding of the corresponding serial number. After the label is printed, the operator scans the component's serial number for confirmation. When the system verifies that the current status is risk assessment completed and the quality grade is Grade A or B, it allows the status to be changed to qualified and ready for warehousing. The serial number, quality grade, disposal time, and operator's employee number are written into the disposal record table. The disposal record table fields include: serial number, original status, new status, disposal type, disposal time, and operator.

[0055] For Class C downgraded products, the system automatically generates a downgrade disposal order based on the risk assessment results. The disposal order data fields include: serial number, defect number, defect type, defect location coordinates, comprehensive risk index, generation time, and suggested disposal method. The system updates the component status to downgraded and awaiting re-inspection. After the re-inspection is completed, authorized auditors log in to the system with their accounts and scan the serial number to confirm the operation, selecting the final disposal method as rework, downgraded use, or scrap. The system writes the auditor's number, confirmation time, and final disposal result into the disposal record table and updates the current status of the component.

[0056] For Class D non-conforming products, the system automatically generates an isolation order and updates the component status to "non-conforming and isolated". The system sets status verification rules in the inventory management module: when the component status is "non-conforming and isolated", it is not allowed to be confirmed for warehousing or assembled and bound. If rework or scrapping is required, the status change process must be triggered by scanning the serial number. The system records the processing type, processing time, processing personnel and processing results in the disposal record table, and retains the original risk assessment record without deletion, so as to achieve full traceability.

[0057] When components are assembled into the finished product, the component serial number and the finished product serial number are scanned at the assembly station using a barcode scanner. The system then executes assembly verification logic: assembly associations are only allowed when the component's current status is either qualified and ready for warehousing or downgraded for use; the system writes the association record to the assembly relationship table, with fields including: finished product serial number, component serial number, assembly time, assembly station number, and operator number; only one valid assembly record is allowed for the same component serial number to prevent duplicate bindings; When a fault occurs during the use of the complete product, by entering or scanning the serial number of the complete product, the system can query the list of all related component serial numbers from the assembly relationship table, and then retrieve and display the initial files, test data, defect files and risk assessment results of each component in sequence; the system outputs the results in order of high to low according to the comprehensive defect risk index to help locate the possible source of the fault.

[0058] If the cause of the fault is confirmed to be related to a defect in a certain component, the database is searched for all component serial numbers in the same batch based on the production batch number of that component. The number of components with fault feedback or quality level C or below is counted, and the proportion of these components to the total number of components in the batch is taken as the batch anomaly ratio. When the batch anomaly ratio exceeds a preset threshold (8%), a batch warning record is automatically generated. The record fields include batch number, anomaly ratio, statistical time, and trigger serial number. The record is then pushed to the quality management account, and the batch is marked as being under risk monitoring in the batch file.

[0059] This allows the serial number to be used throughout the entire process of components, from production filing, testing and analysis, risk assessment, disposal and warehousing to complete assembly and after-sales fault tracing, achieving closed-loop management of the entire life cycle based on the serial number.

[0060] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. An electronic component management system with serial number tracking, characterized in that, include: The serial number identification generation module generates serial numbers for the electronic components to be tested, solidifies the serial numbers onto the surface of the components to generate serial number identification, and simultaneously enters the basic information of the components into the database to establish the corresponding initial files. The response data acquisition module identifies the serial number and obtains the component information. Based on the component type code, it loads the corresponding detection parameter configuration scheme. Based on the configuration scheme, it controls the detection platform to synchronously collect response data of laser interference vibration signal, surface potential signal and acoustic emission signal. After the detection is completed, the data is bound and stored with the serial number. The defect candidate region localization module extracts the vibration response detection data corresponding to the serial number from the database, selects multiple candidate reference regions, establishes a reference vibration response spectrum, calculates the degree of deviation between the vibration response spectrum of each measurement point and the reference spectrum as the vibration response anomaly degree, identifies and judges the anomaly points, and locates the defect candidate region through connected component analysis. The defect type determination module extracts feature parameters from vibration response data, surface potential data, and acoustic emission event data for defect candidate regions, and determines the defect type based on hierarchical decision rules. The risk assessment module calculates geometric risk factors, vibration response risk factors, potential anomaly risk factors, and acoustic emission activity risk factors, and obtains a comprehensive defect risk index by weighting and summing them according to preset weights. Based on the relationship between the comprehensive defect risk index and the preset risk threshold, the defect risk level is determined, and the overall quality is graded according to the risk level distribution. The traceability management module executes corresponding processing procedures for components of different grades based on the quality grading results, and establishes full-process traceability management information based on serial numbers.

2. The electronic component management system with serial number tracking according to claim 1, characterized in that, The specific method for generating serial numbers for the electronic components to be tested is as follows: The serial number consists of 32 digits, and its structure from left to right is as follows: digits 1 to 4 are the component type code, digits 5 to 10 are the production date code, digits 11 to 14 are the production line number, and digits 15 to 32 are the serial number. After the serial number is generated, the database is automatically queried. If the serial number already exists, the serial number is automatically incremented by 1 until an unused serial number is obtained.

3. The electronic component management system with serial number tracking according to claim 1, characterized in that, After placing the electronic components with serial numbers on the detection platform, a recognition timeout threshold is set. If the recognition time exceeds the threshold, an alarm is triggered to prompt the components to be repositioned. The detection parameter configuration scheme includes ultrasonic excitation parameters, laser vibration measurement parameters, surface potential scanning parameters, and acoustic emission sampling parameters. The recognition timeout threshold is 0.3 seconds.

4. The electronic component management system with serial number tracking according to claim 1, characterized in that, The specific method for identifying and determining anomalies is as follows: The total spectral deviation value Dspectrum between the measured spectrum and the reference spectrum is calculated. The Dspectrum is calculated as follows: for each frequency point, the absolute value of the difference between the measured spectrum and the reference spectrum is calculated, and then divided by the reference standard deviation to obtain the normalized deviation value for that frequency point. The normalized deviation values ​​of all frequency points are summed to obtain the Dspectrum. The main resonance peaks of the measured spectrum and the reference spectrum are identified, and the relative offsets of the resonance peak frequencies and amplitudes are calculated. The products of Dspectrum and the first preset weight, the relative offsets of the resonance peak frequencies and the second preset weight, and the relative offsets of the resonance peak amplitudes and the third preset weight are summed to obtain the vibration response anomaly degree V, where the first preset weight is 0.5, the second preset weight is 0.3, and the third preset weight is 0.

2. The above calculation is repeated for all measurement points to obtain the anomaly degree matrix. The mean anomaly degree and the standard deviation (std) anomaly degree of all measurement points are statistically calculated. The anomaly degree threshold is set as the mean anomaly degree plus a preset multiple of 2 multiplied by the std anomaly degree. Measurement points with anomalies greater than this threshold are marked as anomalies.

5. An electronic component management system with serial number tracking according to claim 1, characterized in that, The specific method for locating candidate defect regions using connected component analysis is as follows: First, an anomaly marker matrix with M rows and N columns is established. All measured points are arranged in the anomaly marker matrix according to their spatial positions, where i ranges from 1 to M and j ranges from 1 to N. The element in the i-th row and j-th column corresponds to the (i-1)×N+j-th measured point. Initially, all elements of the matrix are 0, and the matrix elements corresponding to the identified anomalies are set to 1. For a measured point at position (i,j) in the marker matrix, its eight neighboring positions are (i-1,j) above, (i+1,j) below, (i,j-1) to the left, (i,j+1) to the right, (i-1,j-1) to the upper left, (i-1,j-1) to the upper right, (i-1,j+1) to the lower left, and (i+1,j-1) to the lower right. Starting from the upper left corner of the marker matrix, the first unvisited anomaly with an element value of 1 is scanned in row-major order as the seed. For each seed point, a new connected component is created and the seed point is added to that component. The eight neighboring locations of the seed point are checked. If the marker matrix element of a neighboring location is 1 and has not been visited, that neighboring location is added to the current connected component and marked as visited. At the same time, the eight neighboring locations of that neighboring location are recursively processed as new seed points. This process is repeated until the current connected component can no longer be expanded. The marker matrix is ​​then returned to continue scanning to find the next connected component. After all anomaly points have been visited, all connected components are output as defect candidate regions. For each connected component, the geometric center coordinates, area, equivalent circle diameter, and aspect ratio are calculated. The area of ​​the connected component is equal to the number of anomaly points contained in the connected component multiplied by the area represented by a single measurement point. The equivalent circle diameter is equal to the area of ​​the connected component divided by π (2 square roots of the area of ​​the connected component). The aspect ratio is the larger of the X-direction span and the Y-direction span of the connected component divided by the smaller value.

6. The electronic component management system with serial number tracking according to claim 5, characterized in that, The specific method for determining defect types based on hierarchical decision rules is as follows: Vibration-related characteristic parameters include seven parameters: average vibration anomaly, peak vibration anomaly, average frequency offset, average amplitude offset, area, equivalent circle diameter, and aspect ratio. Potential-related characteristic parameters include four parameters: mean P, standard deviation P, contrast P, and gradient P. The mean P is the arithmetic mean of all potential values ​​within the defect area, the standard deviation P is the standard deviation of all potential values ​​within the defect area, the contrast P is the average potential value of the defect area minus the average potential value of the surrounding normal area, and the gradient P is the arithmetic mean of the potential gradient amplitudes at all measurement locations on the boundary of the defect area. Acoustic emission-related characteristic parameters include four parameters: acoustic emission event count N, total acoustic emission energy E, mean dominant frequency F, and standard deviation of dominant frequency. The acoustic emission event count N is the number of acoustic emission events selected and belonging to the current defect candidate area, the total acoustic emission energy E is the sum of the energies of all selected events, and the mean dominant frequency F is the arithmetic mean of the dominant frequencies of all selected events. If the acoustic emission event count N is greater than the first preset threshold, the first layer of judgment determines it as a dynamic active defect and proceeds to the second layer of judgment; otherwise, it determines it as a static stable defect and proceeds to the third layer of judgment. The first preset threshold is 15. The second layer of judgment targets dynamic and active defects. If the absolute value of the comparison P is greater than the second preset threshold, the average frequency offset is less than the third preset threshold, and the aspect ratio is less than the fourth preset threshold, it is judged as a solder joint microcrack defect. If the absolute value of the comparison P is less than or equal to the second preset threshold, the total acoustic emission energy E is greater than the average acoustic emission energy of this batch of defects, and the aspect ratio is greater than the fifth preset threshold, it is judged as a copper foil fracture defect. The second preset threshold is 10, the third preset threshold is 0.05, the fourth preset threshold is 1.5, and the fifth preset threshold is 2. The third layer of judgment targets statically stable defects. If the average amplitude offset is greater than the sixth preset threshold and the standard deviation P is greater than 1.5 times the standard deviation of the potential in the surrounding normal area, it is judged as a porosity defect. If the average frequency offset is less than the seventh preset threshold and the area is greater than the eighth preset threshold, it is judged as an intra-plate delamination defect. The sixth preset threshold is 0.3, the seventh preset threshold is -0.05, and the eighth preset threshold is 100 square millimeters.

7. An electronic component management system with serial number tracking according to claim 6, characterized in that, The specific method for determining the defect risk level based on the relationship between the comprehensive defect risk index and the preset risk threshold is as follows: The equivalent circle diameter is divided by the component thickness to obtain the relative defect size, where the equivalent circle diameter is equal to 2√2 times the defect area divided by pi, and the defect area is divided by the total area of ​​the component to obtain the defect area ratio. The relative defect size is divided by the theoretical value of 100 to obtain the normalized relative defect size. The product of the normalized relative defect size and the first geometric weight, the product of the defect area ratio multiplied by 10 and the second geometric weight are summed to obtain the geometric risk factor, where the first geometric weight is 0.6 and the second geometric weight is 0.

4. The normalized vibration anomaly is obtained by dividing the average vibration anomaly by the average anomaly of all components. The average frequency offset is multiplied by 10 to obtain the frequency offset risk value, and the average amplitude offset is multiplied by 5 to obtain the amplitude offset risk value. The vibration response risk factor is obtained by summing the products of the normalized vibration anomaly and the third vibration weight, the frequency offset risk value and the fourth vibration weight, and the amplitude offset risk value and the fifth vibration weight, where the third vibration weight is 0.5, the fourth vibration weight is 0.3, and the fifth vibration weight is 0.

2. The absolute value of the comparison P is used as the potential comparison risk value. The standard deviation P is divided by the potential standard deviation of the surrounding normal area as the potential standard deviation ratio. The gradient P is divided by the average potential gradient of all components as the normalized boundary potential gradient. The potential anomaly risk factor is obtained by summing the product of the potential comparison risk value and the first potential weight, the product of the potential standard deviation ratio and the second potential weight, and the product of the normalized boundary potential gradient and the third potential weight. The first potential weight is 0.5, the second potential weight is 0.3, and the third potential weight is 0.

2. The acoustic emission event count N is divided by 10 to obtain the normalized acoustic emission count, the total acoustic emission energy E is divided by the average acoustic emission energy of this batch to obtain the normalized acoustic emission energy, and the average main frequency F is divided by 1000 kHz to obtain the normalized acoustic emission frequency. The products of the normalized acoustic emission count and the first acoustic emission weight, the normalized acoustic emission energy and the second acoustic emission weight, and the normalized acoustic emission frequency and the third acoustic emission weight are summed to obtain the acoustic emission activity risk factor, where the first acoustic emission weight is 0.5, the second acoustic emission weight is 0.3, and the third acoustic emission weight is 0.

2. The comprehensive defect risk index is obtained by summing the products of geometric risk factor and weight w1, vibration response risk factor and weight w2, potential anomaly risk factor and weight w3, and acoustic emission active risk factor and weight w4, where w1 is 0.2, w2 is 0.3, w3 is 0.4, and w4 is 0.

1. A defect is classified as high-risk when its comprehensive risk index is greater than or equal to the first risk threshold, as medium-risk when its comprehensive risk index is greater than or equal to the second risk threshold and less than the first risk threshold, and as low-risk when its comprehensive risk index is less than the second risk threshold. The first risk threshold is 7.0 and the second risk threshold is 4.

0.

8. An electronic component management system with serial number tracking according to claim 7, characterized in that, The specific method for classifying overall quality based on risk level distribution is as follows: If a component has no defects or only low-risk defects with a comprehensive risk index less than the third risk threshold and the number of defects does not exceed the first quantity threshold and the total area is less than the first area ratio threshold of the total area of ​​the component, then the component is judged to be a Grade A superior product. If there are no high-risk defects, the number of medium-risk defects does not exceed the second quantity threshold, the area of ​​a single medium-risk defect is smaller than the first area threshold, the total area of ​​medium-risk defects is less than the second area ratio threshold of the total area of ​​the component, and the comprehensive risk index of all medium-risk defects is less than the fourth risk threshold, the component is judged to be a Grade B qualified product. If there are no high-risk defects but the number of medium-risk defects exceeds the second quantity threshold, or the area of ​​a single medium-risk defect is greater than or equal to the first area threshold, or the total area of ​​medium-risk defects is greater than or equal to the second area ratio threshold, the component is determined to be a grade C downgraded product; if there are any high-risk defects or the total area of ​​medium-risk defects exceeds the third area ratio threshold, the component is determined to be a grade D non-conforming product. The third risk threshold is 2.0, the first quantity threshold is 3, the first area ratio threshold is 0.5%, the second quantity threshold is 2, the first area threshold is 100 square millimeters, the second area ratio threshold is 5%, the fourth risk threshold is 6.5, and the third area ratio threshold is 10%.

9. An electronic component management system with serial number tracking according to claim 8, characterized in that, Based on the quality grading results, the specific procedures for implementing corresponding processing flows for components of different grades are as follows: For Grade A superior products and Grade B qualified products, generate labels and allow the status to be changed to qualified and ready for warehousing; for Grade C downgraded products, generate disposal orders and update the status to downgraded and ready for re-inspection; for Grade D unqualified products, generate isolation orders and update the status to unqualified and isolated, prohibiting warehousing confirmation or assembly binding.

10. An electronic component management system with serial number tracking according to claim 9, characterized in that, The specific method for establishing end-to-end traceability management information based on serial numbers is as follows: Based on the production batch number of the faulty component, the database is searched for all component serial numbers in the same batch. The proportion of components with quality grade C or below in the batch is calculated as the batch anomaly ratio. When the batch anomaly ratio exceeds the preset warning threshold, a batch warning record is automatically generated. The record fields include batch number, anomaly ratio, statistical time, and trigger serial number. The record is pushed to the quality management account and marked as risk monitoring status in the batch file. The preset warning threshold is 8%.

Images