Intelligent electronic assembly quality control system based on AOI and SPI integration

Abstract

The invention provides an intelligent electronic assembly quality control system based on AOI (automatic optical inspection) and SPI (solder paste printing inspection) integration, and belongs to the field of electronic information equipment flexible production lines. Comprising an AOI module, an SPI module, a data integration module, a feedback control module and a data storage and transmission module. All the modules carry out high-frequency data transmission through an industrial Ethernet; according to the system provided by the invention, the precision of quality detection is improved, and the accurate recognition of defects is realized through the data association of the AOI and the SPI module and the anomaly detection of the SVM and the CNN algorithm. Real-time process optimization is achieved, the feedback control module automatically adjusts printing and mounting process parameters according to the quality data model, and the product consistency is remarkably improved. The method improves the traceability and safety of the data, achieves the efficient storage and safety access of the detection data through distributed cloud storage and data encryption transmission, and supports the quality problem traceability.

Description

Technical Field

[0001] The present invention belongs to the field of flexible production lines for electronic information equipment, and particularly relates to an intelligent electronic assembly quality control system based on the integration of AOI (Automated Optical Inspection) and SPI (Solder Paste Inspection). Background Art

[0002] In the field of national defense science and technology, during the manufacturing process of electronic information equipment, the assembly accuracy and quality stability are crucial. The reliability of the equipment is directly related to the effectiveness of military operations. In the electronic assembly production line, solder paste printing and component mounting are two key links to ensure the final quality. In the solder paste printing process, the thickness and uniformity of the solder paste directly affect the welding quality. Too thick or too thin solder paste will cause poor welding, leading to potential problems such as solder joint voids and short circuits. In the component mounting process, parameters such as the position, angle, and polarity of the components determine the welding reliability and the stability of the finished product. Any deviation may cause product failures. Especially in the harsh military environment, any minor mistake may lead to serious consequences.

[0003] However, traditional quality control means, such as independent AOI (Automated Optical Inspection) and SPI (Solder Paste Inspection) inspections, although they can detect the solder paste and mounting quality respectively, their data are often scattered in different systems and it is difficult to form a unified analysis. In addition, traditional methods usually conduct inspections after the assembly process is completed, and the feedback process is slow, making it difficult to achieve real-time adjustment of production parameters. This delay is not conducive to the flexibility and efficiency required by modern national defense electronic equipment production lines, especially in dealing with diverse equipment types and complex production environments. Summary of the Invention

[0004] To solve the quality control problems in electronic assembly production, the present invention proposes a quality control system based on the integration of AOI and SPI, which forms a closed-loop management system through data fusion and real-time feedback, significantly improving the efficiency and accuracy of production. The system includes an AOI module and an SPI module, which are respectively responsible for detecting component mounting and solder paste printing quality. The AOI module uses a high-resolution imaging device and intelligent image processing technology to detect defects such as solder joint cracks and offsets; while the SPI module ensures the thickness and uniformity of the solder paste through a laser thickness sensor.

[0005] The data integration module fuses the data from AOI and SPI to form a multi-dimensional quality model, removes noise and outliers through standardization processing to ensure the accuracy of the data. The system can match and analyze the position and thickness information of components and solder paste, and identify potential production problems in advance. In terms of closed-loop control, the feedback control module analyzes the quality model and uses fuzzy logic algorithms to dynamically adjust process parameters, such as automatically adjusting the printing pressure or speed, to ensure the mounting quality, and has functions of status monitoring and alarm to ensure timely handling of serious defects or abnormalities.

[0006] Through the real-time feedback and data fusion system integrating AOI and SPI, the present invention realizes the closed-loop quality control of the flexible production line of defense electronic information equipment. This system can not only ensure that the production line continuously maintains high standards of quality during high-speed operation, but also quickly respond to abnormalities in production, reduce production losses, and improve the utilization efficiency of equipment and resources. Compared with traditional detection systems, this system uniformly manages the AOI and SPI detection data, making the production quality control more accurate and flexible, and at the same time greatly improving the automation and intelligence level of the production line. In this way, defense equipment manufacturing enterprises can more quickly respond to the needs of equipment upgrading and changes in battlefield requirements, enhancing the overall manufacturing capacity and adaptability of defense electronic information equipment.

[0007] The present invention proposes an integrated detection based on AOI and SPI modules. The AOI module integrating multi-angle platforms is used to detect indicators such as the position, offset, and polarity in the component mounting process, and the SPI module combined with an air flow cleaning system is used to detect the solder paste thickness, uniformity, etc. The combination of the two realizes multi-dimensional quality detection. A joint multi-dimensional quality model of AOI and SPI data based on feature association is proposed, and neural networks such as support vector machine (SVM) and convolutional neural network (CNN) are used to detect abnormalities. A decision-making engine based on fuzzy logic control is constructed to adjust the process according to real-time detection data and optimize production parameters.

[0008] The specific technical solution of an intelligent electronic assembly quality control system based on the integration of AOI and SPI of the present invention is as follows:

[0009] An intelligent electronic assembly quality control system based on the integration of AOI and SPI includes an AOI module, an SPI module, a data integration module, a feedback control module, and a data storage and transmission module; each module conducts high-frequency data transmission through an industrial Ethernet.

[0010] The AOI module is used for data processing after the component placement process in electronic assembly production; the AOI module includes a high-resolution imaging device, an intelligent image processing unit, and a data encoder; specifically, the high-resolution imaging device captures the image to be inspected, and then transmits the captured image to the intelligent image processing unit. The intelligent image processing unit analyzes and detects the captured image, uses a machine learning model to identify product defects, and then transmits the identification result to the data encoder. The data encoder further encodes this data and transmits it to the data integration module;

[0011] The SPI module is used to process the data after the solder paste printing process; the API module includes a laser thickness measurement sensor, a high-precision motion platform, and a data processing engine; specifically, the precision motion platform is driven by a servo motor and is equipped with a position feedback system. The laser thickness measurement sensor is set in the center of the motion platform. The thickness of the solder paste on each pad is measured by the laser thickness measurement sensor. The data processing engine converts the measurement data of the laser thickness measurement sensor into a thickness distribution map and further encodes it and transmits it to the data integration module;

[0012] The data integration module is responsible for integrating the detection data from the AOI module and the SPI module, including a data standardization unit, a feature association unit, and an anomaly detection unit; specifically, the data standardization unit automatically filters out abnormal data from the acquired data. For the solder paste thickness data obtained from the SPI module and the component position data obtained from the AOI module, the data standardization unit will perform mean-standard deviation standardization processing; the feature association unit uses spatial coordinate matching and process sequence association technologies to integrate the data into a multi-dimensional quality model; the anomaly detection unit uses support vector machines and convolutional neural networks in parallel and weighted to identify abnormal features in the quality model data;

[0013] The feedback control module implements automatic adjustment of the production line based on the data integration result. This module includes a decision engine, a parameter adjustment unit, a status monitoring unit, and an execution control interface; specifically, the decision engine uses a fuzzy logic control algorithm to analyze the quality model of the data integration module and determines the required process adjustment according to the type and severity of the defects; the parameter adjustment unit automatically sets the key process parameters according to the instructions of the decision engine; the status monitoring unit monitors the current production status, including the real-time comparison of the AOI and SPI detection data; the status monitoring unit also has an alarm system that automatically generates warning messages and alerts the operator when major abnormal deviations or faults are found. The execution control interface communicates with equipment such as the solder paste printer and the mounter;

[0014] The data storage and transmission module includes a data transmission module and an information storage module. The data transmission module synchronously transmits the data generated by each module to the information storage module through industrial Ethernet. The information storage module includes a database and a cloud storage system. The database stores all detection data, abnormal reports, and feedback adjustment records and has indexing and query functions. The cloud storage system adopts a distributed storage architecture and is configured with a redundant backup mechanism.

[0015] The feature association unit is specifically implemented as follows:

[0016] The spatial coordinate matching is specifically represented by the affine transformation matrix T:

[0017]

[0018] where R is the rotation matrix and t is the translation vector. The formula for converting the feature data in the SPI and AOI systems into standard coordinate data is as follows:

[0019] X″ feature,i = R·X′ feature,i + t

[0020] where X′ feature,i is the standardized feature data and X″ feature,i is the feature data in the quality system. Construct a multi-dimensional feature vector Q = [X″ thicknes,i …X″ position,i ; where X″ thickness,i is the thickness data in the quality system and X″ position,i is the position data in the quality system.

[0021] Then the feature association unit applies the cosine similarity to measure the similarity between different features and calculates the association weight w between features based on this:

[0022]

[0023] where σ is a constant for controlling the similarity and similarity(X i ,X j ) is the cosine similarity between different features. Then the anomaly detection unit identifies based on the association weights between different features to obtain the anomaly degree between features and transmits the result to the feedback control module.

[0024] The beneficial effects of the present invention are as follows:

[0025] 1. Improve the accuracy of quality inspection. Through the data association of the AOI and SPI modules and the anomaly detection of the SVM and CNN algorithms, accurate identification of defects is achieved.

[0026] 2. Realized real-time process optimization. The feedback control module automatically adjusts the printing and mounting process parameters according to the quality data model, significantly improving product consistency.

[0027] 3. Improved data traceability and security. Through distributed cloud storage and encrypted data transmission, efficient storage and secure access of detection data are achieved, supporting the traceability of quality problems. Brief Description of the Drawings

[0028] Figure 1 Schematic diagram of the AOI module structure and process of the present invention;

[0029] Figure 2 Schematic diagram of the SPI module structure and process of the present invention;

[0030] Figure 3 Schematic diagram of the working process of the quality control system based on the integration of AOI and SPI of the present invention. Detailed Embodiment

[0031] To better understand the purpose, structure and function of the present invention, the following further describes in detail an intelligent electronic assembly quality control system based on the integration of AOI and SPI of the present invention with reference to the accompanying drawings.

[0032] In an electronic assembly production line, AOI and SPI are key links in the electronic assembly process. SPI detects the thickness and uniformity of solder paste on the pads to ensure the basis of welding quality, while AOI checks the accuracy of welding and components after component mounting, forming a close process correlation; integrating these two modules can provide comprehensive quality control, real-time monitoring and analysis of the entire assembly process, and improve product quality. AOI and SPI provide complementary detection functions. By combining them, the monitoring ability of the production line can be improved more comprehensively. Since the data generated by these two has strong relevance, integration can ensure the accuracy of the data, enabling the production line to handle subtle quality problems.

[0033] This embodiment provides an intelligent electronic assembly quality control system based on the integration of AOI and SPI. The system specifically includes an AOI module, an SPI module, a data integration module, a feedback control module, and a data storage and transmission module; each module conducts high-frequency data transmission through industrial Ethernet (Ethernet) to ensure real-time synchronization of detection data; the AOI and SPI modules are respectively located at the back ends of the solder paste printing and component mounting processes. The data integration module correlates and aggregates the detection data from the two modules to generate a multi-dimensional quality model, and the feedback control module adjusts the process parameters in real time according to this model. The following elaborates on each module in detail.

[0034] The AOI module is used for data processing after the component placement process in electronic assembly production to detect whether the placed components meet the quality standards, such as Figure 1 shown. It mainly includes a high-resolution imaging device, an intelligent image processing unit, and a data encoder. Specifically, the high-resolution imaging device is a 50-million-pixel camera equipped with an optical filter and a 20- to 50-watt dynamically adjustable ring light source. The camera is installed on a multi-angle rotating platform and a vertical top frame and has an automatic angle adjustment and shooting preset program. The intelligent image processing unit uses multiple parallel image processors, and its core algorithms include edge detection, shape recognition, color matching, and brightness adjustment. The processor can also run a feature analysis program based on a machine learning model (convolutional neural network), which will compare the data features in the normal state and the abnormal state to accurately identify defects such as position offset, solder joint crack, and component polarity error. The data encoder converts the image analysis results into data in the JSON standard format, including position coordinates, angles, offsets, defect types, etc., and encodes and transmits this data. In specific implementation, the camera real-time collects the images to be inspected, specifically the product images after the component placement process, and then transmits the collected images to the intelligent image processing unit. The intelligent image processing unit analyzes and detects the collected images, uses the machine learning model in the existing technology to accurately identify the product defects, and then transmits the identification results to the data encoder. The data encoder further encodes these data and transmits them to the data integration module.

[0035] The SPI module is used to process the data after the solder paste printing process to ensure that the solder paste evenly covers the pads and meets the thickness standard; as Figure 2 shown, it mainly includes a laser thickness sensor, a high-precision motion platform, and a data processing engine, and the specific description is as follows:

[0036] The laser thickness sensor uses a sensor with a sensitivity within the range of 1 micron. Its core components include a high-precision laser emitter, a receiving sensor, and a processing unit. The laser emitter emits a high-precision laser beam, and the receiving sensor is installed opposite the emitter to receive the optical signal passing through the object to be measured. The processing unit is integrated at the rear of the sensor. The sensor is equipped with a dust cover made of acrylic or polycarbonate material and an air flow cleaning system. The air flow cleaning system consists of a micro-compressed air pump, an adjustable air flow nozzle, and an air filter assembly. The air filter assembly is installed at the air inlet of the air pump, and the adjustable air flow nozzle is installed around the laser thickness sensor. The micro-compressed air pump provides an air source for the air flow cleaning system. The high-precision motion platform is driven by a servo motor and is equipped with a position feedback system. The laser thickness sensor is set in the center of the motion platform. The data processing engine converts the data of the laser sensor into a thickness distribution map, encodes it into the JSON format, and transmits it to the data integration module.

[0037] In specific implementation, after the solder paste printing process is completed, the SPI module is activated, and the thickness of the solder paste on each pad is measured by a laser thickness sensor. The thickness data collected by the sensor is parsed by the data processing engine to generate a thickness distribution map, data such as coverage uniformity, and is encoded into a standardized JSON format and transmitted to the data integration module.

[0038] The data integration module is responsible for integrating the detection data from the AOI module and the SPI module to form a multi-dimensional quality monitoring data structure. The data integration module includes a data standardization unit, a feature association unit, and an anomaly detection unit;

[0039] Specifically, the data standardization unit automatically filters abnormal data from the acquired data, such as removing signal noise and extreme values. For the solder paste thickness data obtained from the SPI module and the component position data obtained from the AOI module, the data standardization unit will perform mean-standard deviation standardization processing and further filter the standardized data (X′ thickness,i ,X′ shape,i ,…); The feature association unit uses spatial coordinate matching and process sequence association technologies to integrate the detection data generated by the AOI and SPI modules into a multi-dimensional quality model. This unit uses a spatial transformation algorithm to accurately correspond the AOI and SPI data within the same coordinate system. The spatial transformation is represented by an affine transformation matrix T:

[0040]

[0041] where R is the rotation matrix and t is the translation vector; The formula for converting the feature data in the SPI and AOI systems into standard coordinate data is as follows:

[0042] X″ feature,i =R·X′ feature,i +t

[0043] where X′ feature,i is the standardized feature data, and X″ feature,i is the feature data in the quality system; Construct a multi-dimensional feature vector Q=[X″ thickness,i …X″ position,i .

[0044] In addition, the feature association unit applies cosine similarity to measure the similarity degree between different features, and calculates the association weight between features based on this. For example, the similarity formula between thickness and position is as follows:

[0045]

[0046] where, similarity(X″ thickness ,X″ position) is the similarity between the solder paste thickness and the component position, X″ thickness,i is the thickness data in the quality system, X″ position,i is the position data in the quality system.

[0047] The feature association unit applies an association analysis method based on the graph model G(V, E) to establish an association model between features. Where V represents the detection feature nodes, E represents the association edges between nodes, and the association weight w(V thickness , V position ) is determined by the following formula:

[0048]

[0049] Among them, σ is a constant controlling the similarity, similarity(X″ thickness , X″ position ) is the similarity between the solder paste thickness and the component position;

[0050] The anomaly detection unit uses support vector machine (SVM) and convolutional neural network (CNN) in parallel and weighted to identify anomaly features such as solder paste thickness deviation, component position offset, solder joint morphology anomaly, color brightness deviation, polarity or direction error, etc. in the quality model data. The overall identification result S total The formula is as follows:

[0051] S total = λ1·S thickness_SVM · w(V thickness , V position ) + λ2·S position_CNN · w(V position , V shape ) + λ3·S shape_CNN · w(V thickness , V shape ) + …

[0052] Among them, S feature_SVM is the recognition result of the corresponding feature by SVM, indicating the anomaly degree of the corresponding feature. S feature_CNN is the recognition result of the corresponding feature by CNN, indicating the anomaly degree of the corresponding feature. w(V i , V j ) is the association weight between different feature nodes in the feature association unit, and λ i is the weight of each feature recognition result.

[0053] The feedback control module implements automatic adjustment of the production line based on the data integration result. This module includes a decision engine, a parameter adjustment unit, a status monitoring unit, and an execution control interface;

[0054] Specifically, the decision-making engine uses a fuzzy logic control algorithm to analyze the quality model of the data integration module and determines the required process adjustments based on the types and severities of defects. During the fuzzy logic control analysis, the decision-making engine maps the input values of each feature (such as solder paste thickness, component position offset, and solder joint morphology) to fuzzy sets. For example, the solder paste thickness is classified as "too thin", "normal", or "too thick". Each input feature corresponds to multiple conditions in the fuzzy logic rules, forming a fuzzy rule base. These rules are constructed based on historical data and experience to meet different process parameter adjustment requirements. The construction of the fuzzy rule base includes data-driven methods and expert knowledge fusion methods. The data-driven method collects and analyzes historical production data and quality inspection results, statistically analyzes the relationship between the input values of features (such as solder paste thickness) and defect frequencies, and generates a set of rules. For example, the pass rates under different thickness conditions are used to determine the corresponding adjustment rules. The expert knowledge fusion method combines the judgments of experienced engineers and technicians, classifies the input values of features into different levels, and sets the condition priorities for each level to generate rules applicable to specific defect types and severities, thus forming the core framework of the rule base. The engine has a fault recording function and can perform production process analysis and quality traceability. The parameter adjustment unit automatically sets the key process parameters according to the instructions of the decision-making engine. The status monitoring unit monitors the current production status, including the real-time comparison of AOI and SPI detection data, the status of production equipment, etc. The status monitoring unit also has an alarm system that automatically generates warning messages and alerts the operator when major abnormal deviations or faults are detected. The feedback control module communicates with devices such as solder paste printers and pick-and-place machines through multiple interfaces and supports multiple communication protocols such as Modbus and CAN.

[0055] The data storage and transmission module includes a data transmission module and an information storage module. The data transmission module synchronously transmits the data generated by each module to the information storage module through industrial Ethernet. The information storage module includes a database and a cloud storage system;

[0056] The database module uses an SQL database to store all detection data, exception reports, and feedback adjustment records. The database has indexing and query functions and can be queried according to product batches, detection modules, and process adjustment situations. The cloud storage system adopts a distributed storage architecture and is configured with a redundant backup mechanism. The system supports multi-level access permission management, and different users can access the corresponding data according to their authorizations. When the data is uploaded to the cloud storage system, the TLS / SSL encryption transmission protocol is used, and the Brotli compression algorithm is used.

[0057] As Figure 3 shown, the working process of the intelligent electronic assembly quality control system based on the integration of AOI and SPI is specifically as follows:

[0058] Step 1: After the solder paste printing process is completed, the SPI module drives a laser thickness sensor through a high-precision motion platform to measure the thickness of the solder paste on each pad. The thickness data collected by the sensor is parsed by the data processing engine to generate a thickness distribution map, data such as coverage uniformity, and is encoded into a standardized JSON format and transmitted to the data integration module;

[0059] Step 2: After the component mounting process is completed, the AOI module is activated, and images of the soldered joints and components after mounting are collected through a high-resolution imaging device. The intelligent image processing unit segments and analyzes each image, and detects parameters such as the position of the component, the shape, color, and brightness of the soldered joint. At the same time, the image feature data is converted into a standardized JSON format. This data is then transmitted to the data integration module and added to the data stream of real-time quality inspection.

[0060] Step 3: The data integration module collects and synchronizes the inspection data transmitted by the AOI and SPI modules to form a multi-dimensional quality model. First, the data standardization unit denoises and filters outliers from the input data, and preprocesses information such as position and thickness. The feature correlation unit performs matching analysis on the coordinates of the pads and components to generate a multi-dimensional quality and correlation model integrating pad and component mounting information (including indicators related to the spatial coordinates of the pads and components, solder paste thickness and uniformity data, component mounting status, etc. and the correlation degree therein), and identifies potential quality problems, such as component offset or welding defects caused by insufficient solder paste thickness. Then, the anomaly detection unit uses support vector machine and convolutional neural network algorithms in parallel with weighted values to identify the types of defects and generate an anomaly report, and the report data is finally transmitted to the feedback control module.

[0061] Step 4: Based on the quality model provided by the data integration module, the feedback control module analyzes the type and severity of the defects using the decision engine, and instructs the parameter adjustment unit to make corresponding adjustments to the solder paste printing and mounting processes. Specific adjustment measures include modifying the squeegee pressure of the solder paste printer, adjusting the precision settings of the mounter, or adjusting the welding temperature, etc. The adjusted production parameters are fed back to the system in real time through the status monitoring unit to ensure the precise implementation of the control measures. The status monitoring unit continuously receives real-time data from the production line equipment, including production parameters and equipment status, and detects whether there are any anomalies by comparing the current parameters with the preset standards. The status monitoring unit is also responsible for equipment status monitoring during this process. If unexpected deviations or failures occur, it triggers an alarm system to alert the operator and records all information for fault analysis and quality traceability.

[0062] Step 5: All inspection data, defect records, and feedback adjustment data are synchronously transmitted to the database module and stored classified by product batch and inspection module, and key data is automatically backed up to the cloud storage system.

[0063] It is understood that the present invention is described by way of some embodiments, and those skilled in the art will know that various changes or equivalent substitutions can be made to these features and embodiments without departing from the spirit and scope of the present invention. Additionally, under the teaching of the present invention, these features and embodiments can be modified to adapt to specific circumstances and materials without departing from the spirit and scope of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application belong to the scope protected by the present invention.

Claims

1. An intelligent electronic assembly quality control system based on AOI and SPI integration, characterized in that: The system includes AOI module, SPI module, data integration module, feedback control module, data storage and transmission module; each module performs high-frequency data transmission via industrial Ethernet; The AOI module is used for data processing after the component placement process in electronic assembly production; the AOI module includes a high-resolution camera device, an intelligent image processing unit and a data encoder; specifically, the high-resolution camera device collects the image to be inspected, and then transmits the collected image to the intelligent image processing unit, which analyzes and detects the collected image, uses a machine learning model to identify product defects, and then transmits the identification results to the data encoder, which further encodes the data and transmits it to the data integration module; The SPI module is used to process data after the solder paste printing process; the API module includes a laser thickness sensor, a high-precision motion platform and a data processing engine; specifically, the precision motion platform is driven by a servo motor and is equipped with a position feedback system. The laser thickness sensor is set in the center of the motion platform. The laser thickness sensor is used to measure the solder paste thickness on each pad. The data processing engine converts the measurement data of the laser thickness sensor into a thickness distribution map, and further encodes and transmits it to the data integration module. The data integration module is responsible for integrating the inspection data from the AOI module and the SPI module, including a data standardization unit, a feature association unit, and an anomaly detection unit. Specifically, the data standardization unit automatically screens the acquired data for abnormal data. For the solder paste thickness data acquired in the SPI module and the component position data acquired in the AOI module, the data standardization unit will perform mean-standard deviation standardization processing. The feature association unit uses spatial coordinate matching and process sequence association technology to integrate data into a multi-dimensional quality model; the anomaly detection unit uses support vector machines and convolutional neural networks in parallel to weight the abnormal features in the quality model data; The feedback control module implements automatic adjustment of the production line based on the data integration results. The module includes a decision engine, a parameter adjustment unit, a status monitoring unit, and an execution control interface. Specifically, the decision engine uses a fuzzy logic control algorithm to analyze the quality model of the data integration module and determines the required process adjustment based on the type and severity of the defect. The parameter adjustment unit automatically sets key process parameters according to the instructions of the decision engine. The status monitoring unit monitors the current production status, including real-time comparison of AOI and SPI inspection data. The status monitoring unit also has an alarm system that automatically generates warning information and alerts operators when major abnormal deviations or failures are found. The execution control interface communicates with solder paste printers and placement machines. The data storage and transmission module includes a data transmission module and an information storage module. The data transmission module synchronously transmits the data generated by each module to the information storage module through industrial Ethernet. The information storage module includes a database and a cloud storage system. The database stores all detection data, abnormal reports and feedback adjustment records, and has indexing and query functions. The cloud storage system adopts a distributed storage architecture and is configured with a redundant backup mechanism.

2. The intelligent electronic assembly quality control system based on AOI and SPI integration according to claim 1 is characterized in that: The feature association unit is specifically implemented as follows: The spatial coordinate matching is specifically represented by the affine transformation matrix T: Where R is the rotation matrix and t is the translation vector; the formula for converting the feature data in the SPI and AOI systems into standard coordinate data is as follows: X″ feature,i =R·X′ feature,i +t where X′ feature,i is the standardized feature data, X″ feature,i Characteristic data in quality system; construct multidimensional characteristic vector Q = [X" thickness,i …X″ position,i ]; where X″ thickness,i is the thickness data in the quality system, X″ position,i It is the position data in the quality system; The feature association unit then applies cosine similarity to measure the similarity between different features and calculates the association weight w between the features based on this: Among them, σ is a constant that controls similarity, similarity(X i ,X j ) is the cosine similarity between different features; then the anomaly detection unit identifies based on the association weights between different features, obtains the degree of anomaly between the features, and transmits the result to the feedback control module.

3. The intelligent electronic assembly quality control system based on AOI and SPI integration according to claim 2 is characterized in that: The high-resolution camera device is a 50-megapixel camera equipped with an optical filter and a 20-50 watt dynamically adjustable ring lighting source; the camera is installed on a multi-angle rotating platform and a vertical top frame, and has automatic angle adjustment and shooting preset programs; the intelligent image processing unit uses multiple parallel image processors, and its core algorithms include edge detection, shape recognition, color matching and brightness adjustment; the processor can also run a feature analysis program based on a machine learning model to compare data features of normal and abnormal states; the data encoder converts the image analysis results into data in the JSON standard format, including position coordinates, angles, offsets, and defect types, and encodes and transmits these data.

4. The intelligent electronic assembly quality control system based on AOI and SPI integration according to claim 3 is characterized in that: The laser thickness sensor uses a sensor with a sensitivity within the range of 1 micron, including a high-precision laser transmitter, a receiving sensor and a processing unit; the laser transmitter emits a high-precision laser beam, the receiving sensor is installed opposite to the transmitter, and receives the light signal passing through the object to be measured, and the processing unit is integrated at the rear of the sensor; the sensor is equipped with a dust cover and an airflow cleaning system made of acrylic or polycarbonate material, and the airflow cleaning system consists of a micro-compressed air pump, an adjustable airflow nozzle and an air filter assembly; the air filter assembly is installed at the air inlet of the air pump, and the adjustable airflow nozzle is installed around the laser thickness sensor, and an air source is provided for the airflow cleaning system through a micro-compressed air pump.

5. The intelligent electronic assembly quality control system based on AOI and SPI integration according to claim 4 is characterized in that: The fuzzy logic control algorithm analyzes the quality model of the data integration module and determines the required process adjustments based on the defect type and severity as follows: The decision engine maps the input values ​​of each feature to a fuzzy set. Each input feature corresponds to multiple conditions in the fuzzy logic rule to form a fuzzy rule base. The construction of the fuzzy rule base includes data-driven methods and expert knowledge fusion methods. The data-driven method collects and analyzes historical production data and quality inspection results, collects statistics on the relationship between feature input values ​​and defect frequencies, and generates a rule set. The expert knowledge fusion method classifies the feature input values ​​into different levels and sets the condition priority of each level to generate rules applicable to specific defect types and severity.

6. The intelligent electronic assembly quality control system based on AOI and SPI integration according to claim 5 is characterized in that: The abnormal features specifically include solder paste thickness deviation, component position offset, solder joint morphology abnormality, color brightness deviation, polarity or direction error.

7. The intelligent electronic assembly quality control system based on AOI and SPI integration according to claim 6 is characterized in that: When data is uploaded to the cloud storage system, the TLS / SSL encryption transmission protocol is adopted and the Brotli compression algorithm is used.

Images