An electronic connector quality management system and method based on production data fusion

Abstract

The application provides an electronic connector quality management system and method based on production data fusion. Process technology characteristics, appearance morphology characteristics and internal micro stress characteristics of electronic connectors in a current production batch are extracted from machine timing parameters, machine vision images and acoustic emission signals of the electronic connectors. The process technology characteristics, the appearance morphology characteristics and the internal micro stress characteristics are dimensionally fused through environmental context information of the current production batch, and then a fusion feature vector of the electronic connector is obtained. The health degree of the electronic connector is coupled based on the fusion feature vector, and a health degree score of the electronic connector is obtained. The electronic connectors in the current production batch are graded and warned and process parameters are optimized based on the health degree score. The above scheme can realize multi-dimensional fusion evaluation of the quality of the electronic connector, thereby improving the accuracy of production quality process control.

Description

Technical Field

[0001] This application relates to the field of machine vision technology, and more specifically, to a quality control system and method for electronic connectors based on production data fusion. Background Technology

[0002] Electronic connectors are precision electromechanical components used to achieve circuit connections, signal transmission, and current conduction between electronic devices. Their core function is to establish reliable and separable electrical contact paths, and they are widely used in consumer electronics, communication equipment, automotive electronics, and industrial control.

[0003] Traditional quality control methods primarily rely on endpoint functional testing and single-point optical inspection. Functional testing, as the final criterion, can only screen out products that have completely failed, representing a post-event interception. While single-dimensional visual inspection can identify surface geometric deviations and obvious defects, its observational dimension is limited to the product's external macroscopic morphology. It cannot perceive the microscopic state of internal materials or the dynamic evolution of parameters during the manufacturing process. It cannot provide early warnings of performance degradation or potential consistency risks caused by process anomalies such as injection molding parameter drift and stamping die wear. Furthermore, it cannot address complete failure caused by microscopic defects such as microcracks and residual stress concentrations resulting from plastic deformation within the material. These defects may not be apparent in the early stages of product service but will severely impact long-term connection reliability. The lack of integrated perception and correlation analysis capabilities across the entire manufacturing process's multi-physics state leads to lagging and one-sided quality control. Therefore, how to achieve multi-dimensional integrated evaluation of electronic connector quality to improve the accuracy of production quality process control has become a challenge for the industry. Summary of the Invention

[0004] This application provides an electronic connector quality control system and method based on production data fusion, which can realize multi-dimensional fusion evaluation of electronic connector quality, thereby improving the accuracy of production quality process control.

[0005] In a first aspect, this application provides a method for quality control of electronic connectors based on production data fusion, comprising the following steps: Collect machine timing parameters, machine vision images and acoustic emission signals of the current production batch of electronic connectors during the production process to obtain multi-source heterogeneous data of the electronic connectors; The process characteristics, appearance characteristics, and internal micro-stress characteristics of the electronic connectors in the current production batch are extracted from the machine timing parameters, machine vision images, and acoustic emission signals of the multi-source heterogeneous data, respectively. By using the environmental context information of the current production batch, the process characteristics, appearance characteristics, and internal micro-stress characteristics are dimensionally fused to obtain the fused feature vector of the electronic connector. Based on the fused feature vector, the health of the electronic connector is coupled in a process to obtain the health score of the electronic connector. Based on the health score, the electronic connectors in the current production batch are classified for early warning and process parameter optimization is performed.

[0006] In some embodiments, extracting the process characteristics, appearance characteristics, and internal micro-stress characteristics of the electronic connectors in the current production batch from the machine timing parameters, machine vision images, and acoustic emission signals of the multi-source heterogeneous data specifically includes: The machine timing parameters are subjected to stationarity verification and noise reduction filtering, and the statistical and time-frequency domain features of each waveform segment are extracted to obtain the process characteristics. The machine vision image is preprocessed and feature regions are located to obtain the appearance morphology features of the connector pins and plastic shell in the electronic connector; Modal decomposition and spectral analysis are performed on the acoustic emission signal to extract frequency band energy characteristics related to microcracks and stress concentration inside the material, thus forming internal microstress characteristics.

[0007] In some embodiments, the process characteristics, appearance characteristics, and internal micro-stress characteristics are dimensionally fused using the environmental context information of the current production batch to obtain the fused feature vector of the electronic connector. Specifically, this includes: Dynamic attention weights are determined based on the environmental context information of the current production batch to identify process characteristics, appearance characteristics, and internal micro-stress characteristics. The process features, appearance features, and internal micro-stress features are weighted and fused using various dynamic attention weights to obtain the fused feature vector of the electronic connector.

[0008] In some embodiments, the process coupling of the health status of the electronic connector based on the fused feature vector to obtain the health score of the electronic connector specifically includes: The fused feature vector is used to perform regression prediction on the health of the electronic connector to obtain a basic quality score. The stability factor of the electronic connector is determined based on the real-time process capability index of the current production line process. The basic quality score is coupled with the stability factor to obtain the health score of the electronic connector.

[0009] In some embodiments, classifying and issuing early warnings and optimizing process parameters for the current production batch of electronic connectors based on the health score specifically includes: A scoring threshold range is set, and when the health score falls within different ranges of the scoring threshold range, different levels of warning signals are triggered, namely normal, observation, warning, and abnormal. For warning and abnormal warning signals, the root cause tracing process is triggered. By fusing feature vectors and performing reverse analysis, the process and parameter type that caused the score to drop are located. Based on the process and parameter type, compensation parameters are generated from a preset process parameter adjustment strategy library and sent to the corresponding production equipment for execution.

[0010] In some embodiments, the electronic connector is a board-to-board connector based on the PCIe protocol.

[0011] In some embodiments, the production batch is a collection of electronic connector products produced using the same set of process parameters.

[0012] Secondly, this application provides an electronic connector quality control system based on production data fusion, used to execute an electronic connector quality control method based on production data fusion, comprising a fusion evaluation unit, the fusion evaluation unit including: The acquisition module is used to acquire machine timing parameters, machine vision images and acoustic emission signals of the current production batch of electronic connectors during the production process, and obtain multi-source heterogeneous data of the electronic connectors. The processing module is used to extract the process characteristics, appearance characteristics, and internal micro-stress characteristics of the electronic connectors in the current production batch from the machine timing parameters, machine vision images, and acoustic emission signals of the multi-source heterogeneous data, respectively. The processing module is further configured to perform dimensional fusion of the process features, appearance features and internal micro-stress features through the environmental context information of the current production batch, thereby obtaining a fused feature vector of the electronic connector, and performing process coupling on the health of the electronic connector based on the fused feature vector to obtain a health score of the electronic connector. The execution module is used to perform graded early warning and process parameter optimization for the current production batch of electronic connectors based on the health score.

[0013] Thirdly, this application provides a computer device, the computer device including a memory and a processor, the memory storing code, and the processor being configured to acquire the code and execute the above-described method for quality control of electronic connectors based on production data fusion.

[0014] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method for quality control of electronic connectors based on production data fusion.

[0015] The technical solutions provided by the embodiments disclosed in this application have the following beneficial effects: This application provides a quality control system and method for electronic connectors based on production data fusion. The system collects machine timing parameters, machine vision images, and acoustic emission signals from the current production batch of electronic connectors during the production process to obtain multi-source heterogeneous data. Process characteristics, appearance features, and internal micro-stress features of the current production batch of electronic connectors are extracted from the machine timing parameters, machine vision images, and acoustic emission signals. These characteristics are then dimensionally fused using the environmental context information of the current production batch to obtain a fused feature vector for the electronic connector. Based on this fused feature vector, the health of the electronic connector is coupled to obtain a health score. Finally, based on the health score, the system performs graded early warning and process parameter optimization for the current production batch of electronic connectors.

[0016] Therefore, in this application, the health score is used to classify and warn about the electronic connectors in the current production batch and optimize the process parameters. In summary, firstly, by determining the process characteristics, appearance morphology characteristics, and internal micro-stress characteristics, production data from different physical sources and information modalities can be transformed into a unified quality representation vector, thus constructing an objective and comprehensive quantitative foundation for multi-dimensional quality assessment. Through feature engineering, the multi-source heterogeneous raw signals are analyzed into core indicators reflecting process stability, geometric compliance, and material integrity, respectively. This achieves the abstraction and alignment of quality information from dispersed signals to structured knowledge, overcoming the shortcomings of single-dimensional data bias in traditional quality inspection. This allows subsequent assessments to simultaneously obtain complementary information from three key dimensions: process control accuracy, appearance dimensional accuracy, and internal micro-state, providing data input for multi-source information fusion and coupling analysis. Then, by determining the health score, a comprehensive quantitative assessment based on multi-dimensional fused feature vectors can be achieved, thus fusing the aforementioned separated feature vectors into a single, interpretable quality status metric. This is achieved by introducing a process capability index. The system state factors and feature prediction scores are dynamically coupled, so that the scoring results not only characterize the intrinsic quality level of individual products, but also reflect the impact of the overall stability of the current production system on the confidence level of its quality consistency. This achieves an organic combination of static individual characteristics and dynamic system state, enabling the final score to not only identify isolated defects in products, but also reveal potential batch risks caused by process fluctuations or system degradation. This enhances the robustness and predictability of quality judgment, and allows control decisions to distinguish between random individual anomalies and deviations, thereby improving the accuracy and timeliness of process intervention. Based on the above scheme, a multi-dimensional integrated assessment of electronic connector quality can be achieved, thereby improving the accuracy of production quality process control. Attached Figure Description

[0017] Figure 1 This is an exemplary flowchart of an electronic connector quality control method based on production data fusion, according to some embodiments of this application; Figure 2 This is an exemplary flowchart illustrating the determination of a multi-scale digital surface model according to some embodiments of this application; Figure 3 This is a schematic diagram of the structure of the fusion evaluation unit shown in some embodiments of this application; Figure 4 This is a schematic diagram of the structure of a computer device that implements a method for quality control of electronic connectors based on production data fusion, according to some embodiments of this application. Detailed Implementation

[0018] To better understand the technical solution of this application, the technical solution of this application will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0019] refer to Figure 1 The figure is an exemplary flowchart of an electronic connector quality control method based on production data fusion, according to some embodiments of this application. The figure mainly includes the following steps: In step 101, machine timing parameters, machine vision images, and acoustic emission signals of the electronic connectors in the current production batch are collected during the production process to obtain multi-source heterogeneous data of the electronic connectors.

[0020] It should be noted that, in this application, machine timing parameters are time-series data used to record the changes of physical quantities such as injection pressure, holding pressure, injection speed, barrel temperature, and mold temperature of the injection molding machine over time; machine vision images are image data of the connector's plastic shell shape, pin spacing, and surface defects captured by an industrial camera; acoustic emission signals are stress wave signals generated by the plastic deformation of the material during the stamping process of the connector using a piezoelectric sensor; and multi-source heterogeneous data is a set of raw data from different physical dimensions with different data structures used to reflect the connector's production process status, appearance characteristics, and internal stress state.

[0021] In practice, data acquisition modules are deployed on the production line. Specifically, at the injection molding station, pressure and temperature sensors installed on the production equipment continuously collect machine timing parameters at a set sampling frequency. At the vision inspection station, a trigger-type industrial camera automatically captures high-resolution machine vision images of the connector's front and side when it reaches a designated position. At the stamping station, a piezoelectric acoustic emission sensor installed on the mold collects acoustic emission signals during the stamping process. After the above three types of data are synchronized through a unified clock source, they are associated and packaged according to the production batch identifier to form a raw data package based on the production batch. Thus, the combination of machine timing parameters, machine vision images, and acoustic emission signals serves as the multi-source heterogeneous data of the electronic connector.

[0022] In step 102, the process characteristics, appearance characteristics, and internal micro-stress characteristics of the electronic connectors of the current production batch are extracted from the machine timing parameters, machine vision images, and acoustic emission signals of the multi-source heterogeneous data, respectively.

[0023] In some embodiments, the extraction of process features, appearance features, and internal micro-stress features of the electronic connectors in the current production batch from the machine timing parameters, machine vision images, and acoustic emission signals of the multi-source heterogeneous data can be achieved by the following steps: The machine timing parameters are subjected to stationarity verification and noise reduction filtering, and the statistical and time-frequency domain features of each waveform segment are extracted to obtain the process characteristics. The machine vision image is preprocessed and feature regions are located to obtain the appearance morphology features of the connector pins and plastic shell in the electronic connector; Modal decomposition and spectral analysis are performed on the acoustic emission signal to extract frequency band energy characteristics related to microcracks and stress concentration inside the material, thus forming internal microstress characteristics.

[0024] It should be noted that, in this application, the process characteristics are a set of mathematical features that characterize the equipment operating status, process parameter control accuracy, process stability, and deviation from the standard process specification curve during injection molding and stamping of electronic connectors; the appearance characteristics are a set of multi-dimensional geometric and texture feature parameters that quantitatively describe the external geometry, dimensional accuracy, relative positional relationship of key components, and surface texture and defect status of electronic connector products; and the internal micro-stress characteristics are a set of frequency domain energy features that characterize the response of the metal components of electronic connectors to micro-mechanical events within the material during stamping.

[0025] In practical implementation, firstly, for the collected injection pressure timing parameters of the injection molding machine, unit root tests or sliding window statistical tests are used to verify the stability of the mean and variance of the pressure sequence in different production cycles. If non-stationary segments are found, they are marked as potential abnormal conditions. Then, a Butterworth low-pass filter is applied to the pressure sequence for noise reduction. The filter cutoff frequency is set according to the characteristic frequency of the injection molding process to eliminate sensor circuit noise and high-frequency interference. For each complete injection waveform segment after denoising and noise reduction, the average, maximum, standard deviation, and rising slope of the pressure data within that segment are calculated as statistical characteristics. Simultaneously... Discrete wavelet transform is performed on the waveform segment to extract the wavelet energy of the third and fourth layer detail coefficients as time-frequency domain features. The statistical features and time-frequency domain features are then concatenated to obtain process features. Next, the original connector image obtained by an industrial camera is first subjected to median filtering to eliminate salt-and-pepper noise, then histogram equalization is used to enhance image contrast, and finally distortion correction is performed on the image according to the camera calibration parameters to complete the preprocessing. On the preprocessed image, a target detection algorithm based on template matching or a lightweight convolutional neural network is applied to accurately locate the rectangular bounding boxes of all pins and the polygon coordinates of the outer contour of the plastic shell in the image. Within the located pin area, a sub-pixel edge detection algorithm is used to extract the edge contour of each pin, calculate the pixel distance between the center points of adjacent pins, and convert it into the actual physical spacing through calibration. Simultaneously, the standard deviation of the fitted plane on the top surface of all pins is calculated as the pin coplanarity feature. Within the located plastic shell area, its length, width, and thickness of key parts are measured, and the contrast and uniformity of the gray-level co-occurrence matrix of the plastic shell surface are extracted as surface texture features. The pin spacing, pin coplanarity, plastic shell size, and surface texture parameters together constitute the appearance morphology features. Finally, for a segment of acoustic emission signal acquired during the stamping process, the empirical mode decomposition method is first applied to decompose the signal into several intrinsic mode function components arranged from high frequency to low frequency, and a residual component. For example, a typical stamping signal may be decomposed into a first intrinsic mode function component containing high-frequency impact components and a third intrinsic mode function component containing low-frequency plastic deformation components. Fast Fourier transforms are then performed on the decomposed intrinsic mode function components (e.g., the first and third components) that are most relevant to the material's microscopic damage mechanism to obtain their spectra. Next, in the frequency spectrum, frequency bands closely related to the propagation of microcracks in the material (e.g., 100kHz to 200kHz) and frequency bands related to stress concentration and release (e.g., 300kHz to 400kHz) are predefined, and the signal energy integral values ​​of each intrinsic mode function component in these target frequency bands are calculated; these calculated frequency band energy values ​​are used as characteristic parameters characterizing the internal microstress state, constituting the internal microstress characteristics.

[0026] In step 103, the process features, appearance features and internal micro-stress features are dimensionally fused using the environmental context information of the current production batch to obtain the fused feature vector of the electronic connector. Based on the fused feature vector, the health of the electronic connector is coupled in a process to obtain the health score of the electronic connector.

[0027] In some embodiments, the process characteristics, appearance characteristics, and internal micro-stress characteristics are dimensionally fused using the environmental context information of the current production batch to obtain the fused feature vector of the electronic connector. This can be achieved through the following steps: Dynamic attention weights are determined based on the environmental context information of the current production batch to identify process characteristics, appearance characteristics, and internal micro-stress characteristics. The process features, appearance features, and internal micro-stress features are weighted and fused using various dynamic attention weights to obtain the fused feature vector of the electronic connector.

[0028] It should be noted that, in this application, the environmental context information is an auxiliary data set used to describe the specified production conditions of the current production batch, such as: raw material batch number, cumulative number of times the core mold has been used, production environment temperature, and current production line operation mode; the dynamic attention weight is a learnable parameter vector used to dynamically allocate according to the importance of the environmental context information, in order to adjust the relative contribution of the three types of features—process characteristics, appearance characteristics, and internal micro-stress characteristics—in the final quality assessment; the fused feature vector is the final feature representation of the overall quality status of a single electronic connector under the current production conditions, from process stability and appearance compliance to internal integrity.

[0029] In practice, the environmental context information of the current production batch, including but not limited to the raw material batch number, the cumulative number of stampings of the core mold, the temperature and humidity of the workshop environment, and equipment maintenance records, is first encoded into a fixed-dimensional context feature vector. This context feature vector is then input into a trained multilayer perceptron neural network. The output layer of this network has three nodes, each using a sigmoid activation function to output three values ​​between 0 and 1. These three values ​​are normalized so that their sum is 1, representing the dynamic attention weights for the process characteristics, appearance characteristics, and internal micro-stress characteristics of the current batch. For example, when the cumulative number of stampings of the mold exceeds a threshold, the network may automatically increase the weight corresponding to the internal micro-stress characteristics to pay more attention to the potential risks brought about by material fatigue. Then, the extracted process feature vector, appearance feature vector, and internal micro-stress feature vector are multiplied by their corresponding dynamic attention weights using scalar multiplication operations. Specifically, each element of the process feature vector is multiplied by the dynamic attention weight assigned to the process feature. The same operation is performed on the appearance morphology feature vector and the internal micro-stress feature vector, resulting in three weighted feature vectors. These three weighted feature vectors are then concatenated sequentially along their feature dimensions to form a longer, single-dimensional vector. This concatenated long vector is then subjected to global mean normalization to eliminate the influence of differences in feature scales, thus obtaining the fused feature vector of the electronic connector. This vector integrates information from all sources and adaptively adjusts the importance of information according to the specific production batch.

[0030] In some embodiments, the health status of the electronic connector is coupled process-wise based on the fused feature vector to obtain a health score for the electronic connector, which is then referenced. Figure 2 The figure is a flowchart illustrating the process of determining a multi-scale digital surface model in some embodiments of this application. In this embodiment, determining the multi-scale digital surface model can be achieved using the following steps: In step 1031, the fused feature vector is used to perform regression prediction on the health of the electronic connector to obtain a basic quality score; In step 1032, the stability factor of the electronic connector is determined based on the real-time process capability index of the current production line process. In step 1033, the basic quality score is coupled with the stability factor to obtain the health score of the electronic connector.

[0031] It should be noted that, in this application, the basic quality score is a continuous numerical prediction result used to conduct a preliminary quantitative assessment of the comprehensive quality status of a single product based purely on its multidimensional characteristic data under ideal conditions without considering the overall fluctuations of the production line; the stability factor is used to quantify the overall dynamic stability level of the production line as an adjustment coefficient between 0 and 1, which reflects the confidence level or amplification / attenuation effect of the system state on the reliability of the quality of a single product; the health score is used as the basis for the final quality judgment and decision, and it is a comprehensive quantitative indicator that includes the quality performance information of the individual electronic connector and the current stability information of the production system in which it is located.

[0032] In practice, the first step is to input the fused feature vector into a pre-trained support vector regression model or gradient boosting tree regression model. This model has been trained on historical data (containing fused feature vectors of a large number of products of different quality levels and their final expert quality scores), learning a non-linear mapping relationship from complex features to quality scores. After receiving the fused feature vector of the current product, the model calculates through its internal decision function and directly outputs a continuous value between 0 and 1, which is the product's basic quality score. For example, a product with a perfect appearance, stable process parameters, and no internal stress anomalies may have a basic quality score close to 0.95, while a product with potential risks may score below 0.6. This step integrates multi-dimensional features into a direct and easy-to-understand preliminary quality judgment. Then, for a selected key quality characteristic, such as the thickness of the stamped pin, measurements of this characteristic are collected in real time from the 500 most recently produced products. First, the mean and standard deviation of this set of measurements are calculated. Based on the upper and lower specification limits of the pin thickness, the difference between each value and the mean is calculated and divided by three times the standard deviation to obtain two values. The smaller of these two values ​​is taken as the real-time process capability index for this process on the current production line. This real-time process capability index is mapped to a stability factor using a predefined conversion function. Specifically, when the real-time process capability index is greater than or equal to 1.67, the stability factor is set to 1.0 (indicating high system stability and complete confidence in the base score); when the real-time process capability index is lower than 1.0, the stability factor linearly decreases to 0.6 (indicating system instability and a need to significantly reduce confidence in the score of individual products). Finally, the base quality score and the stability factor are multiplied to obtain the health score of the electronic connector. The calculation formula is: Health Score = Base Quality Score × Stability Factor. For example, if a product's base quality score is 0.9, indicating good inherent characteristics; if the current production line's stability factor is only 0.7 (indicating significant production line fluctuations), then its final health score is 0.9 * 0.7 = 0.63. This score reflects that in a highly volatile production environment, even if the product itself performs reasonably well, its overall quality reliability is lowered due to system instability, significantly increasing the probability that the product will be identified as a high-risk item requiring close monitoring.

[0033] In step 104, the electronic connectors of the current production batch are classified and warned based on the health score, and the process parameters are optimized.

[0034] In some embodiments, classifying and issuing early warnings and optimizing process parameters for the current production batch of electronic connectors based on the health score can be achieved through the following steps: A scoring threshold range is set, and when the health score falls within different ranges of the scoring threshold range, different levels of warning signals are triggered, namely normal, observation, warning, and abnormal. For warning and abnormal warning signals, the root cause tracing process is triggered. By fusing feature vectors and performing reverse analysis, the process and parameter type that caused the score to drop are located. Based on the process and parameter type, compensation parameters are generated from a preset process parameter adjustment strategy library and sent to the corresponding production equipment for execution.

[0035] It should be noted that, in this application, the scoring threshold range is used to divide the continuous health score values ​​of electronic connectors into several continuous numerical ranges, each range corresponding to a preset quality risk level classification standard; the warning signal is used to automatically generate different levels of quality status indications based on the scoring threshold range in which the health score is located, to trigger the corresponding subsequent processing procedures; the root cause tracing process is a systematic analysis process used to automatically analyze the potential sources of quality problems and accurately locate the specific production process and the type of key parameters within that process when a warning or abnormal state occurs; the process parameter adjustment strategy library is a database used to store standardized parameter compensation schemes that should be adopted for different processes and different parameter types when different quality deviation modes occur; the compensation parameter is the amount of temporary or permanent numerical adjustment made to the set parameters of the specified production equipment in order to correct the systematic deviation in the production process when a quality warning or abnormality occurs.

[0036] In practical implementation, firstly, a scoring threshold range configuration table containing four numerical ranges is pre-defined and maintained. For example, a health score greater than or equal to 0.85 is defined as the normal range, a score between 0.70 and 0.85 as the observation range, a score between 0.60 and 0.70 as the warning range, and a score below 0.60 as the abnormal range. During production, the system calculates the health score of each electronic connector in real time and immediately compares it with the above threshold ranges; based on the range in which the score falls, the system automatically generates the corresponding warning signal. For example, a product with a score of 0.92 will generate a "normal" level warning signal, indicating that no additional action is required; a product with a score of 0.65 will generate a "warning" level warning signal, accompanied by an audible and visual prompt, notifying the quality engineer to intervene; then, when a warning or abnormal signal is triggered, the system automatically initiates the root cause tracing process. First, the system calls a pre-trained gradient interpretation model (e.g., the SHAP model), which takes the fused feature vector of the current problematic product as input. This model can calculate the "contribution" of each original feature dimension (e.g., "energy of the third-scale wavelet of the injection pressure waveform" or "standard deviation of pin spacing") in the fused feature vector to the decrease in the current health score relative to the normal baseline. This value can be positive or negative; a negative contribution indicates that the feature caused the score decrease. Based on a predefined mapping table between feature dimensions and process and parameter types, the system attributes several features with the largest negative contributions to their source process and parameter type. For example, if the feature with the largest contribution is "energy of the stamping acoustic emission signal in the 150-200kHz frequency band," the system can locate the root cause as "abnormal internal stress state in the stamping process." Finally, based on the process (e.g., "injection molding process") and parameter type (e.g., "holding pressure") located by the root cause tracing process, the system queries a pre-defined process parameter adjustment strategy library. This strategy library, indexed by "process-parameter type-offset mode," stores the corresponding parameter adjustment rules. For example, for a pattern like "injection molding process - holding pressure - too low," the strategy library might store a rule: "Increase the holding pressure setpoint by 5% of the current value and maintain this for ten production cycles." The system scales the base adjustment amount in the strategy library proportionally based on the specific deviation (e.g., the difference between the calculated health score and the normal threshold), generating specific compensation parameters. The system then sends a work order containing the compensation parameters and activation instructions to the corresponding injection molding machine controller via the manufacturing execution equipment interface in the workshop. The equipment automatically executes the parameter adjustments, thus completing a closed-loop control from quality assessment to production correction.

[0037] Furthermore, in another aspect of this application, in some embodiments, this application provides an electronic connector quality control system based on production data fusion. This system includes a fusion evaluation unit, referencing... Figure 3The figure is a schematic diagram of the structure of a fusion evaluation unit according to some embodiments of this application. The fusion evaluation unit includes: an acquisition module 201, a processing module 202, and an execution module 203, which are described below: The acquisition module 201 in this application is mainly used to acquire machine timing parameters, machine vision images and acoustic emission signals of the electronic connectors in the current production batch during the production process, so as to obtain multi-source heterogeneous data of the electronic connectors. Processing module 202, in this application, is used to extract the process characteristics, appearance characteristics, and internal micro-stress characteristics of the electronic connectors of the current production batch from the machine timing parameters, machine vision images, and acoustic emission signals of the multi-source heterogeneous data, respectively. It should be noted that the processing module 202 is also used to perform dimensional fusion of the process features, appearance features and internal micro-stress features through the environmental context information of the current production batch, thereby obtaining the fused feature vector of the electronic connector, and performing process coupling on the health of the electronic connector based on the fused feature vector to obtain the health score of the electronic connector. The execution module 203 in this application is mainly used to perform graded early warning and process parameter optimization for the current production batch of electronic connectors based on the health score.

[0038] The foregoing has detailed examples of an electronic connector quality control system and method based on production data fusion provided in the embodiments of this application. It is understood that the corresponding apparatus, in order to achieve the above functions, includes hardware structures and / or software modules corresponding to the execution of each function. Those skilled in the art should readily recognize that, based on the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0039] In some embodiments, this application also provides a computer device, the computer device including a memory and a processor, the memory for storing a computer program, and the processor for calling and running the computer program from the memory, so that the computer device performs the above-described method for quality control of electronic connectors based on production data fusion.

[0040] In some embodiments, reference Figure 4The dashed lines in the figure indicate that the unit or module is optional. This figure is a schematic diagram of the structure of a computer device implementing a production data fusion-based electronic connector quality control method according to an embodiment of this application. The production data fusion-based electronic connector quality control method described in the above embodiments can be... Figure 4 The computer device shown is used to implement this, and the computer device includes at least one processor 301, a memory 302 and at least one communication unit 305. The computer device may be a terminal device, a server or a chip.

[0041] Processor 301 can be a general-purpose processor or a special-purpose processor. For example, processor 301 can be a central processing unit (CPU), which can be used to control computer devices, execute software programs, and process data from software programs. The computer device may also include a communication unit 305 for inputting (receiving) and outputting (transmitting) signals.

[0042] For example, the computer device may be a chip, and the communication unit 305 may be the input and / or output circuit of the chip, or the communication unit 305 may be the communication interface of the chip, which may be a component of a terminal device, network device or other device.

[0043] For example, the computer device may be a terminal device or a server, and the communication unit 305 may be a transceiver of the terminal device or the server, or the communication unit 305 may be a transceiver circuit of the terminal device or the server.

[0044] The computer device may include one or more memories 302 storing a program 304. The program 304 can be executed by a processor 301 to generate instructions 303, causing the processor 301 to execute the method described in the above method embodiments according to the instructions 303. Optionally, the memory 302 may also store data (such as a target audit model). Optionally, the processor 301 may also read data stored in the memory 302, which may be stored at the same storage address as the program 304, or it may be stored at a different storage address than the program 304.

[0045] The processor 301 and memory 302 can be configured separately or integrated together, for example, integrated on the system on chip (SOC) of the terminal device.

[0046] It should be understood that each step of the above method embodiment can be completed by hardware logic circuits or software instructions in the processor 301. The processor 301 can be a CPU, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, such as discrete gates, transistor logic devices, or discrete hardware components.

[0047] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0048] For example, in some embodiments, this application also provides a computer-readable storage medium storing instructions or code that, when executed on a computer, cause the computer to implement the above-described method for quality control of electronic connectors based on production data fusion.

[0049] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications within the scope of this application.

[0050] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for quality control of electronic connectors based on production data fusion, characterized in that, Includes the following steps: Collect machine timing parameters, machine vision images and acoustic emission signals of the current production batch of electronic connectors during the production process to obtain multi-source heterogeneous data of the electronic connectors; The process characteristics, appearance characteristics, and internal micro-stress characteristics of the electronic connectors in the current production batch are extracted from the machine timing parameters, machine vision images, and acoustic emission signals of the multi-source heterogeneous data, respectively. By using the environmental context information of the current production batch, the process characteristics, appearance characteristics, and internal micro-stress characteristics are dimensionally fused to obtain the fused feature vector of the electronic connector. Based on the fused feature vector, the health of the electronic connector is coupled in a process to obtain the health score of the electronic connector. Based on the health score, the electronic connectors in the current production batch are classified for early warning and process parameter optimization is performed.

2. The method as described in claim 1, characterized in that, Extracting the process characteristics, appearance characteristics, and internal micro-stress characteristics of the electronic connectors in the current production batch from the machine timing parameters, machine vision images, and acoustic emission signals of the multi-source heterogeneous data specifically includes: The machine timing parameters are subjected to stationarity verification and noise reduction filtering, and the statistical and time-frequency domain features of each waveform segment are extracted to obtain the process characteristics. The machine vision image is preprocessed and feature regions are located to obtain the appearance morphology features of the connector pins and plastic shell in the electronic connector; Modal decomposition and spectral analysis are performed on the acoustic emission signal to extract frequency band energy characteristics related to microcracks and stress concentration inside the material, thus forming internal microstress characteristics.

3. The method as described in claim 1, characterized in that, By fusing the process characteristics, appearance characteristics, and internal micro-stress characteristics using the environmental context information of the current production batch, the fused feature vector of the electronic connector is obtained, specifically including: Dynamic attention weights are determined based on the environmental context information of the current production batch to identify process characteristics, appearance characteristics, and internal micro-stress characteristics. The process features, appearance features, and internal micro-stress features are weighted and fused using various dynamic attention weights to obtain the fused feature vector of the electronic connector.

4. The method as described in claim 1, characterized in that, Based on the fused feature vector, the health score of the electronic connector is obtained by process coupling, specifically including: The fused feature vector is used to perform regression prediction on the health of the electronic connector to obtain a basic quality score. The stability factor of the electronic connector is determined based on the real-time process capability index of the current production line process. The basic quality score is coupled with the stability factor to obtain the health score of the electronic connector.

5. The method as described in claim 1, characterized in that, The classification, early warning, and process parameter optimization of the current production batch of electronic connectors based on the health score specifically include: A scoring threshold range is set, and when the health score falls within different ranges of the scoring threshold range, different levels of warning signals are triggered, namely normal, observation, warning, and abnormal. For warning signals at the early warning and abnormal levels, the root cause tracing process is triggered. By fusing feature vectors and performing reverse analysis, the process and parameter type that caused the score to drop are located. Based on the process and parameter type, compensation parameters are generated from a preset process parameter adjustment strategy library and sent to the corresponding production equipment for execution.

6. The method as described in claim 1, characterized in that, The electronic connector is a board-to-board connector based on the PCIe protocol.

7. The method as described in claim 1, characterized in that, The production batch refers to a collection of electronic connector products manufactured using the same set of process parameters.

8. A quality control system for electronic connectors based on production data fusion, used to execute the quality control method for electronic connectors based on production data fusion as described in any one of claims 1 to 7, wherein the quality control system for electronic connectors based on production data fusion includes a fusion evaluation unit, characterized in that, The fusion evaluation unit includes: The acquisition module is used to acquire machine timing parameters, machine vision images and acoustic emission signals of the current production batch of electronic connectors during the production process, and obtain multi-source heterogeneous data of the electronic connectors. The processing module is used to extract the process characteristics, appearance characteristics, and internal micro-stress characteristics of the electronic connectors in the current production batch from the machine timing parameters, machine vision images, and acoustic emission signals of the multi-source heterogeneous data, respectively. The processing module is further configured to perform dimensional fusion of the process features, appearance features and internal micro-stress features through the environmental context information of the current production batch, thereby obtaining a fused feature vector of the electronic connector, and performing process coupling on the health of the electronic connector based on the fused feature vector to obtain a health score of the electronic connector. The execution module is used to perform graded early warning and process parameter optimization for the current production batch of electronic connectors based on the health score.

9. A computer device, characterized in that, The computer device includes a memory and a processor, the memory storing code, and the processor being configured to retrieve the code and execute the electronic connector quality control method based on production data fusion as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the electronic connector quality control method based on production data fusion as described in any one of claims 1 to 7.

Images