Data real-time transmission and feedback method and system for IC carrier board repair process

Abstract

The application belongs to the technical field of IC carrier plate manufacturing and data processing, and relates to a data real-time transmission and feedback method and system for an IC carrier plate repair process. The method realizes real-time collection of multi-source repair data by deploying an edge collection unit, realizes data connection between front and rear processes by a double-link low-delay transmission architecture and a cross-process data association library, constructs a scene-specific evaluation system by machine vision and the analytic hierarchy process, dynamically generates parameter adjustment instructions in combination with an intelligent feedback model, forms a closed-loop control of "collection-analysis-feedback-adjustment", and further realizes continuous optimization and cooperation of the closed-loop control of the repair process by data sedimentation and model iteration, improves the accuracy and stability of the repair process, and solves the defects of fragmented data collection, delayed data transmission and reliance on manual experience feedback.

Description

Technical Field

[0001] This invention belongs to the field of IC substrate manufacturing and data processing technology, specifically relating to a method and system for real-time data transmission and feedback during the IC substrate repair process. Background Technology

[0002] IC substrates are prone to various defects in the front-end processes, requiring subsequent repairs to improve yield. Repair involves precise multi-parameter operations, and its effectiveness is strongly correlated with defect characteristics and front-end parameters, demanding extremely high data real-time performance, accuracy, and feedback loop stability.

[0003] Existing repair data processing methods have significant shortcomings: First, data acquisition is fragmented, with heterogeneous equipment protocols and inconsistent formats, making real-time aggregation and trajectory tracking across the entire process impossible. Second, under traditional architectures, multi-source data transmission latency reaches hundreds of milliseconds, failing to meet the real-time adjustment requirements of micron-level repairs and easily leading to unstable repair quality. Furthermore, feedback relies on human experience, lacking intelligent models based on real-time data, making dynamic adaptation of parameters and defects difficult and disconnected from previous processes. Finally, the value of data is not fully exploited; it is merely archived without being used for parameter optimization, resulting in poor process stability and high costs.

[0004] In view of this, it is very necessary to provide a method and system for real-time data transmission and feedback in the IC carrier board repair process to solve the above-mentioned defects in the prior art. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies, such as fragmented data acquisition, high data transmission latency, and reliance on manual experience feedback, by providing a real-time data transmission and feedback method and system for IC carrier board repair processes, thereby solving the aforementioned technical problems.

[0006] To achieve the above objectives, the present invention provides the following technical solution: A method for real-time data transmission and feedback during IC carrier board repair includes the following steps: Step S1: Deploy edge acquisition units to collect repair process parameters, equipment status data, defect and repair effect data, and environmental data in real time; Assign a unique traceability code to each carrier board to be repaired, linking it to information from previous processes; Step S2: Transmit data through an edge-cloud dual-link architecture, transmit real-time data through a 5G-U network, and transmit non-real-time data through an industrial Ethernet network; Establish a cross-process data association library to link the repair data with the previous process parameters; Step S3: Establish a scene-specific repair effect evaluation system and generate repair effect evaluation results; Establish an intelligent feedback model to generate feedback adjustment schemes for single defect repair scenarios as well as complex defect or secondary repair scenarios; Step S4: Convert the feedback adjustment scheme into standardized control commands, send them to the edge acquisition unit of the corresponding repair node and execute them by the repair equipment, collect the repair data after the feedback adjustment scheme is executed, and form a closed-loop control. The repair trajectory map of each carrier board is updated based on the real-time collected repair data and synchronized to the cross-process data association library in real time. Step S5: Based on historical repair data, iteratively update the intelligent feedback model and the repair effect evaluation system; Based on cross-process data association database, parameter adjustment suggestions are fed back to the preceding process to achieve collaborative optimization between the preceding process and the repair process.

[0007] Preferably, step S1 specifically includes: Step S11, Repair node and defect type adaptation: The repair process for IC substrates can be summarized as follows: laser repair, electroplating soldering, mechanical polishing, and plasma repair. The laser repair includes: repairing pinholes on the carrier board and repairing open circuits on the carrier board. The electroplating repair welding includes: repairing defects in the coating of the carrier plate; The mechanical polishing includes: repairing scratches on the surface of the carrier plate; The plasma repair includes: removing residual impurities; Each repair process is assigned to a repair node, and each repair node is configured with a unique identifier, which is associated with the corresponding repair equipment model, repair process, and applicable defect range. Establish a defect classification and coding system, and standardize coding according to defect cause, defect form and defect level to achieve unified representation of defect information; The causes of the defects include: equipment-related issues, environmental factors, and substrate-related issues; The defect types mentioned include: point defects, line defects, and surface defects; The defect levels are: minor, moderate, and severe. Step S12, Edge Acquisition Unit Deployment and Data Traceability: Lightweight edge acquisition units are deployed at each repair node, and multiple communication protocols such as SECS / GEM, ModbusTCP, and EtherNet / IP are adapted through customized interfaces to establish real-time connections with repair equipment, online detection equipment, and environmental sensors. Each carrier board to be repaired is assigned a unique traceability code, which is linked to its previous process parameters, defect detection reports and repair work order information. The edge acquisition unit binds the repair data to the carrier board based on the traceability code to ensure the uniqueness of data traceability. Step S13, Real-time Acquisition and Preprocessing of Multi-Dimensional Data: The edge acquisition unit collects multi-source data during the repair process at a sampling frequency of 100Hz. The collected data includes: repair process parameters, equipment status data, defect and repair effect data, and environmental data. The repair process parameters include: laser repair power, welding temperature, polishing pressure, and repair time; The equipment status data includes: laser head wear, welding torch verticality, grinding head rotation speed, and equipment vibration value; The defect and repair effect data include: defect size, repair area, circuit impedance after repair, and surface flatness; The environmental data includes: temperature, humidity, cleanliness, and air pressure of the remediation area; Local preprocessing of the collected data includes: outlier removal, unit standardization, data compression, and feature extraction; The preprocessed data is encapsulated into standardized repair data frames containing carrier traceability codes, repair node identifiers, timestamps, data types, and structured data bodies, with a compression ratio controlled within 10:1 to ensure data transmission efficiency.

[0008] Preferably, step S2 specifically includes: Step S21, Establishing a dual-link data transmission architecture: Build an edge-cloud dual-link transmission architecture; The standardized repair data frames are divided into two categories for transmission by the edge acquisition unit: data with high real-time requirements (such as repair process parameters and equipment status data) are transmitted through the 5G-U low-latency network to ensure that the transmission latency is ≤5ms; non-real-time data (such as defect images before and after repair, complete data logs, etc.) are transmitted through industrial Ethernet to achieve a balance between the reliability and efficiency of data transmission. Establish a data verification mechanism and verify data integrity using the CRC-32 check algorithm to prevent data loss or tampering during data transmission. Step S22, Repair the data bus and cross-process related library construction: A unified repair data bus is built to receive data uploaded by each edge acquisition unit, and the data is transmitted and classified for storage based on the traceability code of the carrier board. A cross-process data association library is constructed, integrating the process parameters of the preceding process of the carrier board, defect detection data and repair process data. A full life cycle data association model with the carrier board traceability code as the core is established, and the influence weight of the preceding process parameters on the repair difficulty and repair effect is marked, so as to provide data support for subsequent intelligent feedback.

[0009] Preferably, step S3 specifically includes: Step S31, Construction of a multi-dimensional repair effect evaluation system: Based on defect classification coding, repair process specifications and historical repair data, a scenario-based repair effect evaluation system is established to generate repair effect evaluation results. The restoration effect evaluation results include: image segmentation and feature extraction of the restored area based on the U-Net model, obtaining feature parameters such as edge integrity, line continuity, and surface roughness. Edge integrity The normalized value of the Hausdorff distance between the repaired area and the original design outline is obtained by calculating:

[0010] in, To repair the edge point set, To design the contour point set, To prevent the constant term from being divided by zero.

[0011] Line continuity By extracting the skeleton of the repair area and performing connectivity analysis, the proportion of the longest connected path to the total design path is calculated as follows:

[0012] in, This represents the actual connected length of the line after repair. This represents the theoretical length of the original design route; Surface roughness The contrast ratio, calculated from the gray-level co-occurrence matrix, is used as a surface texture evaluation metric.

[0013] in, The gray-level co-occurrence matrix represents the gray-level co-occurrence matrix in The value at that location, Represents the gray level of an image. The square of the grayscale difference; The repair effectiveness evaluation results also include: a comprehensive repair effectiveness score calculated using the analytic hierarchy process (AHP) based on electrical performance testing data.

[0014] in, For visual weight, For electrical weights, Weights for edge integrity feature parameters, The weights are the characteristic parameters of line continuity. For the surface roughness characteristic parameter weights, Weights for line impedance characteristic parameters. Weights for insulation resistance characteristic parameters This is the standard value of the line impedance. This is the standard value for insulation resistance; The comprehensive score for the repair effect includes three levels: excellent, qualified, and unqualified. A score of 90 or above is excellent, a score between 70 and 89 is qualified, and a score less than 70 is unqualified. The traceability code for carrier boards requiring secondary repair is automatically marked based on the comprehensive score of the repair effect. Step S32, Dynamic Feedback Parameter Generation Mechanism: Based on the cross-process data association library and the repair effect evaluation results, an intelligent feedback model is established based on deep reinforcement learning methods. For single defect repair scenarios, an association rule mining algorithm is used to discover the matching rules between repair parameters and defect type and defect size, and generate a feedback adjustment scheme for parameter fine-tuning (e.g., for every 0.2μm increase in the diameter of a pinhole defect, the laser power is increased by 5-8W and the repair time is extended by 10ms). For complex defects or scenarios requiring secondary repair, the root cause of poor repair results is located by combining the parameter deviation data of the previous process and using a causal inference model. A multi-parameter collaborative feedback adjustment plan is generated, which clarifies the parameter adjustment range, timing and priority to avoid repair conflicts.

[0015] Preferably, step S4 specifically includes: Step S41, hierarchical issuance and execution of instructions: By repairing the data bus, the parameters in the feedback adjustment scheme are converted into standardized control commands, which are then sent to the edge acquisition unit corresponding to the repair node according to the node's identity. After receiving and verifying the legality of the control command, the edge acquisition unit sends it to the repair equipment in real time to perform the adjustment according to the feedback adjustment plan, and collects the adjusted repair data to form a closed-loop control of "acquisition-analysis-feedback-adjustment-reacquisition". In case of fatal parameter deviations (such as laser power exceeding the safety threshold), the repair equipment will automatically be shut down and an alarm will be issued to ensure the safety of the repair process. Step S42, Repair the dynamic update and tracing of the trajectory: Based on real-time collected repair data, the repair trajectory map of each carrier board is dynamically updated; The repair trajectory map uses a time axis as a guide to fully record the defect location, changes in repair parameters, repair effect evaluation results, and operator information; The repair trajectory map is synchronized to the cross-process data association library in real time, supporting full-process data traceability and review analysis.

[0016] Preferably, step S5 specifically includes: The entire repair process data, feedback adjustment plans, and repair effect evaluation results are tagged by carrier board batch, defect type, and repair node, and stored in the historical database. Establish a data cleaning mechanism to regularly remove invalid and duplicate data, and retain high-quality data as training samples for process optimization; By mining the optimization patterns of various data parameters in the historical database through a multilayer perceptron network model, the intelligent feedback model and the repair effect evaluation system are iteratively updated to optimize the initial repair parameter range for different defect types. Based on data from the cross-process data association library, parameter adjustment suggestions are fed back to the upstream process (e.g., for the frequently repaired short circuit problem, it is recommended to optimize the upstream electroplating current density by ±5A / dm²), so as to achieve synergistic optimization between the upstream process and the repair process and gradually reduce the defect incidence and repair cost.

[0017] Furthermore, the present invention also provides a real-time data transmission and feedback system for the IC carrier board repair process, comprising: The multi-source data acquisition system construction module includes: Deploy edge acquisition units to collect repair process parameters, equipment status data, defect and repair effect data, and environmental data in real time; Assign a unique traceability code to each carrier board to be repaired, linking it to information from previous processes; The cross-process data association module contains: Data is transmitted via an edge-cloud dual-link architecture, real-time data is transmitted via 5G-U network, and non-real-time data is transmitted via industrial Ethernet; Establish a cross-process data association library to link the repair data with the previous process parameters; Repair assessment and feedback module, in which: Establish a scenario-based repair effect evaluation system and generate repair effect evaluation results; Establish an intelligent feedback model to generate feedback adjustment schemes for single defect repair scenarios as well as complex defect or secondary repair scenarios; The command issuance and closed-loop control module contains: The feedback adjustment scheme is transformed into standardized control commands, which are then sent to the edge acquisition units of the corresponding repair nodes and executed by the repair equipment. The repair data after the feedback adjustment scheme is executed is collected to form a closed-loop control. The repair trajectory map of each carrier board is updated based on the real-time collected repair data and synchronized to the cross-process data association library in real time. The data accumulation and process optimization module includes: Based on historical restoration data, the intelligent feedback model and restoration effect evaluation system are iteratively updated; Based on cross-process data association database, parameter adjustment suggestions are fed back to the preceding process to achieve collaborative optimization between the preceding process and the repair process.

[0018] As a preferred embodiment, the multi-source data acquisition system construction module specifically includes: a repair node and defect type adaptation sub-module, an edge acquisition unit deployment and data tracing sub-module, and a multi-dimensional data real-time acquisition and preprocessing sub-module. The aforementioned repair node and defect type adaptation submodule includes: The repair process for IC substrates can be summarized as follows: laser repair, electroplating soldering, mechanical polishing, and plasma repair. The laser repair includes: repairing pinholes on the carrier board and repairing open circuits on the carrier board. The electroplating repair welding includes: repairing defects in the coating of the carrier plate; The mechanical polishing includes: repairing scratches on the surface of the carrier plate; The plasma repair includes: removing residual impurities; Each repair process is assigned to a repair node, and each repair node is configured with a unique identifier, which is associated with the corresponding repair equipment model, repair process, and applicable defect range. Establish a defect classification and coding system, and standardize coding according to defect cause, defect form and defect level to achieve unified representation of defect information; The causes of the defects include: equipment-related issues, environmental factors, and substrate-related issues; The defect types mentioned include: point defects, line defects, and surface defects; The defect levels are: minor, moderate, and severe. The edge acquisition unit deployment and data traceability submodule includes: Lightweight edge acquisition units are deployed at each repair node, and multiple communication protocols such as SECS / GEM, ModbusTCP, and EtherNet / IP are adapted through customized interfaces to establish real-time connections with repair equipment, online detection equipment, and environmental sensors. Each carrier board to be repaired is assigned a unique traceability code, which is linked to its previous process parameters, defect detection reports and repair work order information. The edge acquisition unit binds the repair data to the carrier board based on the traceability code to ensure the uniqueness of data traceability. The aforementioned multi-dimensional data real-time acquisition and preprocessing submodule includes: The edge acquisition unit collects multi-source data during the repair process at a sampling frequency of 100Hz. The collected data includes: repair process parameters, equipment status data, defect and repair effect data, and environmental data. The repair process parameters include: laser repair power, welding temperature, grinding pressure, and repair time; The equipment status data includes: laser head wear, welding torch verticality, grinding head rotation speed, and equipment vibration value; The defect and repair effect data include: defect size, repair area, circuit impedance after repair, and surface smoothness; The environmental data includes: temperature, humidity, cleanliness, and air pressure of the remediation area; Local preprocessing of the collected data includes: outlier removal, unit standardization, data compression, and feature extraction; The preprocessed data is encapsulated into standardized repair data frames containing carrier traceability codes, repair node identifiers, timestamps, data types, and structured data bodies, with a compression ratio controlled within 10:1 to ensure data transmission efficiency.

[0019] Preferably, the cross-process data association module specifically includes: a dual-link data transmission architecture construction submodule and a data bus repair and cross-process association library construction submodule; The aforementioned dual-link data transmission architecture construction submodule includes: Build an edge-cloud dual-link transmission architecture; The standardized repair data frames are divided into two categories for transmission by the edge acquisition unit: data with high real-time requirements (such as repair process parameters and equipment status data) are transmitted through the 5G-U low-latency network to ensure that the transmission latency is ≤5ms; non-real-time data (such as defect images before and after repair, complete data logs, etc.) are transmitted through industrial Ethernet to achieve a balance between the reliability and efficiency of data transmission. Establish a data verification mechanism and verify data integrity using the CRC-32 check algorithm to prevent data loss or tampering during data transmission. The aforementioned repair data bus and cross-process association library construction submodule includes: A unified repair data bus is built to receive data uploaded by each edge acquisition unit, and the data is transmitted and classified for storage based on the traceability code of the carrier board. A cross-process data association library is constructed, integrating the process parameters of the preceding process of the carrier board, defect detection data and repair process data. A full life cycle data association model with the carrier board traceability code as the core is established, and the influence weight of the preceding process parameters on the repair difficulty and repair effect is marked, so as to provide data support for subsequent intelligent feedback.

[0020] Preferably, the repair assessment and feedback module specifically includes: a multi-dimensional repair effect assessment system construction sub-module and a dynamic feedback parameter generation mechanism sub-module; The aforementioned sub-module for constructing a multi-dimensional repair effect evaluation system includes: Based on defect classification coding, repair process specifications and historical repair data, a scenario-based repair effect evaluation system is established to generate repair effect evaluation results. The evaluation results of the repair effect include: image segmentation and feature extraction of the image of the repaired area based on the U-Net model, obtaining feature parameters such as edge integrity, line continuity, and surface roughness. Edge integrity The normalized value of the Hausdorff distance between the repaired area and the original design outline is obtained by calculating:

[0021] in, To repair the edge point set, To design the contour point set, To prevent the constant term from being divided by zero.

[0022] Line continuity By extracting the skeleton of the repair area and performing connectivity analysis, the proportion of the longest connected path to the total design path is calculated as follows:

[0023] in, This represents the actual connected length of the line after repair. This represents the theoretical length of the original design route; Surface roughness The contrast ratio, calculated from the gray-level co-occurrence matrix, is used as a surface texture evaluation metric.

[0024] in, The gray-level co-occurrence matrix represents the gray-level co-occurrence matrix in The value at that location, Represents the gray level of an image. The square of the grayscale difference; The repair effectiveness evaluation results also include: a comprehensive repair effectiveness score calculated using the analytic hierarchy process (AHP) based on electrical performance testing data.

[0025] in, For visual weight, For electrical weights, Weights for edge integrity feature parameters, The weights are the characteristic parameters of line continuity. For the surface roughness characteristic parameter weights, Weights for line impedance characteristic parameters. Weights for insulation resistance characteristic parameters This is the standard value of the line impedance. This is the standard value for insulation resistance; The comprehensive score for the repair effect includes three levels: excellent, qualified, and unqualified. A score of 90 or above is excellent, a score between 70 and 89 is qualified, and a score less than 70 is unqualified. The traceability code for carrier boards requiring secondary repair is automatically marked based on the comprehensive score of the repair effect. The dynamic feedback parameter generation mechanism submodule includes: Based on the cross-process data association library and the repair effect evaluation results, an intelligent feedback model is established based on deep reinforcement learning methods. For single defect repair scenarios, an association rule mining algorithm is used to discover the matching rules between repair parameters and defect type and defect size, and generate a feedback adjustment scheme for parameter fine-tuning (e.g., for every 0.2μm increase in the diameter of a pinhole defect, the laser power is increased by 5-8W and the repair time is extended by 10ms). For complex defects or scenarios requiring secondary repair, the root cause of poor repair results is located by combining the parameter deviation data of the previous process and using a causal inference model. A multi-parameter collaborative feedback adjustment plan is generated, which clarifies the parameter adjustment range, timing and priority to avoid repair conflicts.

[0026] Preferably, the instruction issuance and closed-loop control module specifically includes: an instruction hierarchical issuance and execution submodule and a repair trajectory dynamic update and tracing submodule; The aforementioned instruction hierarchical distribution and execution submodule includes: By repairing the data bus, the parameters in the feedback adjustment scheme are converted into standardized control commands, which are then sent to the edge acquisition unit corresponding to the repair node according to the node's identity. After receiving and verifying the legality of the control command, the edge acquisition unit sends it to the repair equipment in real time to perform the adjustment according to the feedback adjustment plan, and collects the adjusted repair data to form a closed-loop control of "acquisition-analysis-feedback-adjustment-reacquisition". In case of fatal parameter deviations (such as laser power exceeding the safety threshold), the repair equipment will automatically be shut down and an alarm will be issued to ensure the safety of the repair process. The aforementioned repair trajectory dynamic update and tracing submodule includes: Based on real-time collected repair data, the repair trajectory map of each carrier board is dynamically updated; The repair trajectory map uses a time axis as a guide to fully record the defect location, changes in repair parameters, repair effect evaluation results, and operator information; The repair trajectory map is synchronized to the cross-process data association library in real time, supporting full-process data traceability and review analysis.

[0027] Preferably, the data accumulation and process optimization module specifically includes: The entire repair process data, feedback adjustment plans, and repair effect evaluation results are tagged by carrier board batch, defect type, and repair node, and stored in the historical database. Establish a data cleaning mechanism to regularly remove invalid and duplicate data, and retain high-quality data as training samples for process optimization; By mining the optimization patterns of various data parameters in the historical database through a multilayer perceptron network model, the intelligent feedback model and the repair effect evaluation system are iteratively updated to optimize the initial repair parameter range for different defect types. Based on data from the cross-process data association library, parameter adjustment suggestions are fed back to the upstream process (e.g., for the frequently repaired short circuit problem, it is recommended to optimize the upstream electroplating current density by ±5A / dm²), so as to achieve synergistic optimization between the upstream process and the repair process and gradually reduce the defect incidence and repair cost.

[0028] The beneficial effects of this invention are as follows: First, by using edge acquisition units and a unique traceability code mechanism, the multi-source data format is unified, and standardized acquisition and precise binding of multi-source repair data are achieved. Second, by combining a dual-link transmission architecture and utilizing a 5G-U low-latency network, the latency of key data transmission is ensured to be ≤5ms, enabling real-time feedback and adjustment of repair parameters and avoiding instability in repair quality due to latency. Furthermore, based on an intelligent feedback model and closed-loop control mechanism, dynamic adaptation of repair parameters to defect types and upstream process parameters is achieved, replacing traditional manual experience-based judgment, reducing the false repair rate, increasing repair yield by 8%, and improving the accuracy and stability of the repair process. Moreover, through a cross-process data association library and model iteration mechanism, repair data can be transformed into a basis for upstream process optimization, forming a closed loop of "pre-process prevention - downstream repair - data feedback - global optimization," reducing the overall defect incidence and manufacturing cost of IC substrates. Finally, through tag-based management and machine learning mining, the value of historical repair data is fully activated, continuously optimizing the repair process model to adapt to the repair needs of different substrates and different defect types, reducing reliance on operator experience and improving the flexibility of the production line.

[0029] Therefore, it is evident that the present invention has outstanding substantive features and significant progress compared with the prior art, and the beneficial effects of its implementation are also obvious. Attached Figure Description

[0030] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0031] Figure 1 This is a flowchart of a real-time data transmission and feedback method for the IC carrier board repair process provided by the present invention.

[0032] Figure 2 This is a timing diagram of a real-time data transmission and feedback method for the IC carrier board repair process provided by the present invention.

[0033] Figure 3 This is a schematic diagram of a real-time data transmission and feedback system for the IC carrier board repair process provided by the present invention.

[0034] Among them, 1-Multi-source data acquisition system construction module, 2-Cross-process data association module, 3-Repair evaluation and feedback module, 4-Command issuance and closed-loop control module, and 5-Data accumulation and process optimization module. Detailed Implementation

[0035] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. The following embodiments are explanations of the present invention, but the present invention is not limited to the following implementation methods.

[0036] Example 1: like Figure 1 and Figure 2 As shown in this embodiment, a real-time data transmission and feedback method for IC carrier board repair process is provided to repair IC carrier boards with open circuit defects. The method includes the following steps: Step S1: Deploy edge acquisition units to collect repair process parameters, equipment status data, defect and repair effect data, and environmental data in real time; Each carrier board to be repaired is assigned a unique traceability code, which is linked to the information of the previous process. In this embodiment, the carrier board to be repaired with the traceability code "CB-Repair-20260122-001" is used as an example. Its previous process data (including drilling diameter and electroplating thickness) and defect information (open circuit location: X=12.3mm, Y=8.5mm, open circuit length: 0.8μm, defect level: moderate) Step S1 specifically includes: Step S11, Repair node and defect type adaptation: The repair process of the IC substrate is reviewed, and the laser repair node (equipment model: LaserFix-3000) is selected as the analysis node in this embodiment. The identity identifier "Laser-Repair-001" is configured and associated with the corresponding repair equipment model, repair process and applicable defect range. Establish a defect classification and coding system, and standardize coding according to defect cause, defect form and defect level to achieve unified representation of defect information; The causes of the defects include: equipment-related issues, environmental factors, and substrate-related issues; The defect types mentioned include: point defects, line defects, and surface defects; The defect levels are: minor, moderate, and severe. Step S12, Edge Acquisition Unit Deployment and Data Traceability: Lightweight edge acquisition units are deployed at each repair node, and multiple communication protocols such as SECS / GEM, ModbusTCP, and EtherNet / IP are adapted through customized interfaces to establish real-time connections with repair equipment, online detection equipment, and environmental sensors. Each carrier board to be repaired is assigned a unique traceability code, which is linked to its previous process parameters, defect detection reports and repair work order information. The edge acquisition unit binds the repair data to the carrier board based on the traceability code to ensure the uniqueness of data traceability. Step S13, Real-time Acquisition and Preprocessing of Multi-Dimensional Data: The edge acquisition unit collects multi-source data during the repair process at a sampling frequency of 100Hz. The collected data includes: laser power parameters, repair time, laser head position, repair area image, equipment vibration value, and environmental data. Local preprocessing of the collected data includes: outlier removal, unit standardization, data compression, and feature extraction; The preprocessed data is encapsulated into standardized repair data frames containing carrier traceability codes, repair node identifiers, timestamps, data types, and structured data bodies, with a compression ratio controlled within 10:1 to ensure data transmission efficiency.

[0037] Step S2: Transmit data through an edge-cloud dual-link architecture, transmit real-time data through a 5G-U network, and transmit non-real-time data through an industrial Ethernet network; Establish a cross-process data association library to link the repair data with the previous process parameters; Step S2 specifically includes: Step S21, Establishing a dual-link data transmission architecture: Build an edge-cloud dual-link transmission architecture; The standardized repair data frames are divided into two types of data for transmission through the edge acquisition unit: real-time parameters (laser power, repair time) are transmitted through the 5G-U network with a transmission delay of 3.2ms; and defect images before and after repair are transmitted through industrial Ethernet with a compressed transmission rate of 100Mbps, achieving a balance between reliability and efficiency in data transmission. Establish a data verification mechanism and verify data integrity using the CRC-32 check algorithm to prevent data loss or tampering during data transmission. Step S22, Repair the data bus and cross-process related library construction: A unified repair data bus is built to receive data uploaded by each edge acquisition unit, and the data is transmitted and classified for storage based on the traceability code of the carrier board. A cross-process data association library is constructed, integrating the process parameters of the preceding process of the carrier board, defect detection data and repair process data. A full life cycle data association model with the carrier board traceability code as the core is established, and the influence weight of the preceding process parameters on the repair difficulty and repair effect is marked, so as to provide data support for subsequent intelligent feedback.

[0038] Step S3: Establish a scene-specific repair effect evaluation system and generate repair effect evaluation results; Establish an intelligent feedback model to generate feedback adjustment schemes for single defect repair scenarios as well as complex defect or secondary repair scenarios; Step S3 specifically includes: Step S31, Construction of a multi-dimensional repair effect evaluation system: The intelligent feedback model, combined with historical data, initially set the laser power to 45W and the repair time to 80ms for this moderate open circuit defect. Based on defect classification coding, repair process specifications and historical repair data, a scenario-based repair effect evaluation system is established to generate repair effect evaluation results. The U-Net model was used to identify anomalies in the repair area image, and it was inferred and verified that the line impedance was at risk of being too high (the line impedance exceeded the standard value by 5%). The overall score calculated using the Analytic Hierarchy Process (AHP) is 78 points, which is considered a passing grade. The traceability code for carrier boards requiring secondary repair is automatically marked based on the comprehensive score of the repair effect. Step S32, Dynamic Feedback Parameter Generation Mechanism: Based on the cross-process data association library and the repair effect evaluation results, an intelligent feedback model is established based on a deep reinforcement learning model. For the current defect repair scenario, the association rule mining algorithm is used to explore the matching rules between repair parameters and defect type and defect size, and generate a feedback adjustment scheme for parameter fine-tuning: the intelligent feedback model analyzes that the root cause is slightly low laser power, and generates a feedback instruction: increase the laser power to 48W, and keep the repair time unchanged.

[0039] Step S4: Convert the feedback adjustment scheme into standardized control commands, send them to the edge acquisition unit of the corresponding repair node and execute them by the repair equipment, collect the repair data after the feedback adjustment scheme is executed, and form a closed-loop control. The repair trajectory map of each carrier board is updated based on the real-time collected repair data and synchronized to the cross-process data association library in real time. Step S4 specifically includes: Step S41, hierarchical issuance and execution of instructions: By repairing the data bus, the parameters in the feedback adjustment scheme are converted into standardized control commands, which are then sent to the edge acquisition unit corresponding to the repair node according to the node's identity. After receiving and verifying the legality of the control command, the edge acquisition unit sends it to the repair equipment in real time to perform the adjustment according to the feedback adjustment plan, and collects the adjusted repair data to form a closed-loop control of "acquisition-analysis-feedback-adjustment-reacquisition". After secondary repair, the line impedance was found to be up to standard, with a comprehensive score of 92 points. In case of fatal parameter deviations (such as laser power exceeding the safety threshold), the repair equipment will automatically be shut down and an alarm will be issued to ensure the safety of the repair process. Step S42, Repair the dynamic update and tracing of the trajectory: Based on real-time collected repair data, the repair trajectory map of each carrier board is dynamically updated; The repair trajectory map records the parameter changes and effect evaluation results of the two repairs, and links them to the carrier traceability code to achieve full-process traceability; The repair trajectory map is synchronized to the cross-process data association library in real time, supporting full-process data traceability and review analysis.

[0040] Step S5: Based on historical repair data, iteratively update the intelligent feedback model and the repair effect evaluation system; Based on cross-process data association database, parameter adjustment suggestions are fed back to the preceding process to achieve collaborative optimization between the preceding process and the repair process.

[0041] Step S5 specifically includes: The data from this repair will be tagged and stored (tags: line break, moderate defect, laser repair) and stored in the historical database as training samples to iterate the intelligent feedback model; By analyzing the optimization patterns of various data parameters in the historical database using a multilayer perceptron network model, and iteratively updating the intelligent feedback model and the repair effect evaluation system, it was found that the initial power parameters for optimizing this type of defect should be in the range of 46W-49W. Based on data from the cross-process data association library, parameter adjustment suggestions are provided to the upstream processes: For the line open circuit problem that is frequently repaired, it is recommended to optimize the initial power parameters to be in the range of 46W-49W, so as to achieve collaborative optimization between the upstream and repair processes and gradually reduce the defect rate and repair costs.

[0042] Example 2: In this embodiment, 15% of a certain batch of IC substrates showed a high frequency of electroplating repair and soldering needs. Through cross-process data correlation analysis, the root cause of the defect was located as uneven coating caused by the fluctuation of current density (deviation +8A / dm²) in the previous electroplating process. Based on the feedback from the repair data, a parameter adjustment instruction was issued to the upstream process: the electroplating current density was stabilized within the range of ±3A / dm², and the initial temperature of the repair welding was optimized to 220℃ (the original initial temperature was 230℃). After the adjustment, the frequency of subsequent IC substrate electroplating and soldering repairs in this batch was reduced to below 3%, the repair efficiency was improved by 20%, and the overall production cost was reduced by 12%.

[0043] Example 3: like Figure 3 As shown in the figure, this embodiment provides a real-time data transmission and feedback system for the IC carrier board repair process, including: Multi-source data acquisition system construction module 1, in which: Deploy edge acquisition units to collect repair process parameters, equipment status data, defect and repair effect data, and environmental data in real time; Assign a unique traceability code to each carrier board to be repaired, linking it to information from previous processes; The multi-source data acquisition system construction module 1 specifically includes: a repair node and defect type adaptation sub-module, an edge acquisition unit deployment and data tracing sub-module, and a multi-dimensional data real-time acquisition and preprocessing sub-module; The aforementioned repair node and defect type adaptation submodule includes: The repair process for IC substrates can be summarized as follows: laser repair, electroplating soldering, mechanical polishing, and plasma repair. The laser repair includes: repairing pinholes on the carrier board and repairing open circuits on the carrier board. The electroplating repair welding includes: repairing defects in the coating of the carrier plate; The mechanical polishing includes: repairing scratches on the surface of the carrier plate; The plasma repair includes: removing residual impurities; Each repair process is assigned to a repair node, and each repair node is configured with a unique identifier, which is associated with the corresponding repair equipment model, repair process, and applicable defect range. Establish a defect classification and coding system, and standardize coding according to defect cause, defect form and defect level to achieve unified representation of defect information; The causes of the defects include: equipment-related issues, environmental factors, and substrate-related issues; The defect types mentioned include: point defects, line defects, and surface defects; The defect levels are: minor, moderate, and severe. The edge acquisition unit deployment and data traceability submodule includes: Lightweight edge acquisition units are deployed at each repair node, and multiple communication protocols such as SECS / GEM, ModbusTCP, and EtherNet / IP are adapted through customized interfaces to establish real-time connections with repair equipment, online detection equipment, and environmental sensors. Each carrier board to be repaired is assigned a unique traceability code, which is linked to its previous process parameters, defect detection reports and repair work order information. The edge acquisition unit binds the repair data to the carrier board based on the traceability code to ensure the uniqueness of data traceability. The aforementioned multi-dimensional data real-time acquisition and preprocessing submodule includes: The edge acquisition unit collects multi-source data during the repair process at a sampling frequency of 100Hz. The collected data includes: repair process parameters, equipment status data, defect and repair effect data, and environmental data. The repair process parameters include: laser repair power, welding temperature, grinding pressure, and repair time; The equipment status data includes: laser head wear, welding torch verticality, grinding head rotation speed, and equipment vibration value; The defect and repair effect data include: defect size, repair area, circuit impedance after repair, and surface smoothness; The environmental data includes: temperature, humidity, cleanliness, and air pressure of the remediation area; Local preprocessing of the collected data includes: outlier removal, unit standardization, data compression, and feature extraction; The preprocessed data is encapsulated into standardized repair data frames containing carrier traceability codes, repair node identifiers, timestamps, data types, and structured data bodies, with a compression ratio controlled within 10:1 to ensure data transmission efficiency.

[0044] Cross-process data association module 2, in which: Data is transmitted via an edge-cloud dual-link architecture, real-time data is transmitted via 5G-U network, and non-real-time data is transmitted via industrial Ethernet; Establish a cross-process data association library to link the repair data with the previous process parameters; The cross-process data association module 2 specifically includes: a dual-link data transmission architecture construction submodule and a data bus repair and cross-process association library construction submodule; The aforementioned dual-link data transmission architecture construction submodule includes: Build an edge-cloud dual-link transmission architecture; The standardized repair data frames are divided into two categories for transmission by the edge acquisition unit: data with high real-time requirements (such as repair process parameters and equipment status data) are transmitted through the 5G-U low-latency network to ensure that the transmission latency is ≤5ms; non-real-time data (such as defect images before and after repair, complete data logs, etc.) are transmitted through industrial Ethernet to achieve a balance between the reliability and efficiency of data transmission. Establish a data verification mechanism and verify data integrity using the CRC-32 check algorithm to prevent data loss or tampering during data transmission. The aforementioned repair data bus and cross-process association library construction submodule includes: A unified repair data bus is built to receive data uploaded by each edge acquisition unit, and the data is transmitted and classified for storage based on the traceability code of the carrier board. A cross-process data association library is constructed, integrating the process parameters of the preceding process of the carrier board, defect detection data and repair process data. A full life cycle data association model with the carrier board traceability code as the core is established, and the influence weight of the preceding process parameters on the repair difficulty and repair effect is marked, so as to provide data support for subsequent intelligent feedback.

[0045] Repair assessment and feedback module 3, in which: Establish a scenario-based repair effect evaluation system and generate repair effect evaluation results; Establish an intelligent feedback model to generate feedback adjustment schemes for single defect repair scenarios as well as complex defect or secondary repair scenarios; The aforementioned repair assessment and feedback module 3 specifically includes: a multi-dimensional repair effect assessment system construction sub-module and a dynamic feedback parameter generation mechanism sub-module; The aforementioned sub-module for constructing a multi-dimensional repair effect evaluation system includes: Based on defect classification coding, repair process specifications and historical repair data, a scenario-based repair effect evaluation system is established to generate repair effect evaluation results. The evaluation results of the repair effect include: image segmentation and feature extraction of the image of the repaired area based on the U-Net model, obtaining feature parameters such as edge integrity, line continuity, and surface roughness. Edge integrity The normalized value of the Hausdorff distance between the repaired area and the original design outline is obtained by calculating:

[0046] in, To repair the edge point set, To design the contour point set, To prevent the constant term from being divided by zero.

[0047] Line continuity By extracting the skeleton of the repair area and performing connectivity analysis, the proportion of the longest connected path to the total design path is calculated as follows:

[0048] in, This represents the actual connected length of the line after repair. This represents the theoretical length of the original design route; Surface roughness The contrast ratio, calculated from the gray-level co-occurrence matrix, is used as a surface texture evaluation metric.

[0049] in, The gray-level co-occurrence matrix represents the gray-level co-occurrence matrix in The value at that location, Represents the gray level of an image. The square of the grayscale difference; The repair effectiveness evaluation results also include: a comprehensive repair effectiveness score calculated using the analytic hierarchy process (AHP) based on electrical performance testing data.

[0050] in, For visual weight, For electrical weights, Weights for edge integrity feature parameters, The weights are the characteristic parameters of line continuity. For the surface roughness characteristic parameter weights, Weights for line impedance characteristic parameters. Weights for insulation resistance characteristic parameters This is the standard value of the line impedance. This is the standard value for insulation resistance; The comprehensive score for the repair effect includes three levels: excellent, qualified, and unqualified. A score of 90 or above is excellent, a score between 70 and 89 is qualified, and a score less than 70 is unqualified. The traceability code for carrier boards requiring secondary repair is automatically marked based on the comprehensive score of the repair effect. The dynamic feedback parameter generation mechanism submodule includes: Based on the cross-process data association library and the repair effect evaluation results, an intelligent feedback model is established based on deep reinforcement learning methods. For single defect repair scenarios, an association rule mining algorithm is used to discover the matching rules between repair parameters and defect type and defect size, and generate a feedback adjustment scheme for parameter fine-tuning (e.g., for every 0.2μm increase in the diameter of a pinhole defect, the laser power is increased by 5-8W and the repair time is extended by 10ms). For complex defects or scenarios requiring secondary repair, the root cause of poor repair results is located by combining the parameter deviation data of the previous process and using a causal inference model. A multi-parameter collaborative feedback adjustment plan is generated, which clarifies the parameter adjustment range, timing and priority to avoid repair conflicts.

[0051] Command issuance and closed-loop control module 4, in which: The feedback adjustment scheme is transformed into standardized control commands, which are then sent to the edge acquisition units of the corresponding repair nodes and executed by the repair equipment. The repair data after the feedback adjustment scheme is executed is collected to form a closed-loop control. The repair trajectory map of each carrier board is updated based on the real-time collected repair data and synchronized to the cross-process data association library in real time. The instruction issuance and closed-loop control module 4 specifically includes: an instruction hierarchical issuance and execution submodule and a repair trajectory dynamic update and tracing submodule; The aforementioned instruction hierarchical distribution and execution submodule includes: By repairing the data bus, the parameters in the feedback adjustment scheme are converted into standardized control commands, which are then sent to the edge acquisition unit corresponding to the repair node according to the node's identity. After receiving and verifying the legality of the control command, the edge acquisition unit sends it to the repair equipment in real time to perform the adjustment according to the feedback adjustment plan, and collects the adjusted repair data to form a closed-loop control of "acquisition-analysis-feedback-adjustment-reacquisition". In case of fatal parameter deviations (such as laser power exceeding the safety threshold), the repair equipment will automatically be shut down and an alarm will be issued to ensure the safety of the repair process. The aforementioned repair trajectory dynamic update and tracing submodule includes: Based on real-time collected repair data, the repair trajectory map of each carrier board is dynamically updated; The repair trajectory map uses a time axis as a guide to fully record the defect location, changes in repair parameters, repair effect evaluation results, and operator information; The repair trajectory map is synchronized to the cross-process data association library in real time, supporting full-process data traceability and review analysis.

[0052] Data accumulation and process optimization module 5, in which: Based on historical restoration data, the intelligent feedback model and restoration effect evaluation system are iteratively updated; Based on cross-process data association database, parameter adjustment suggestions are fed back to the preceding process to achieve collaborative optimization between the preceding process and the repair process.

[0053] The data accumulation and process optimization module 5 specifically includes: The entire repair process data, feedback adjustment plans, and repair effect evaluation results are tagged by carrier board batch, defect type, and repair node, and stored in the historical database. Establish a data cleaning mechanism to regularly remove invalid and duplicate data, and retain high-quality data as training samples for process optimization; By mining the optimization patterns of various data parameters in the historical database through a multilayer perceptron network model, the intelligent feedback model and the repair effect evaluation system are iteratively updated to optimize the initial repair parameter range for different defect types. Based on data from the cross-process data association library, parameter adjustment suggestions are fed back to the upstream process (e.g., for the frequently repaired short circuit problem, it is recommended to optimize the upstream electroplating current density by ±5A / dm²), so as to achieve synergistic optimization between the upstream process and the repair process and gradually reduce the defect incidence and repair cost.

[0054] The above-disclosed embodiments are merely preferred embodiments of the present invention, but the present invention is not limited thereto. Any non-creative variations that can be conceived by those skilled in the art, as well as any improvements and modifications made without departing from the principles of the present invention, should fall within the protection scope of the present invention.

Claims

1. A method for real-time data transmission and feedback during IC substrate repair, characterized in that, Includes the following steps: Step S1: Deploy edge acquisition units to collect repair process parameters, equipment status data, defect and repair effect data, and environmental data in real time; Assign a unique traceability code to each carrier board to be repaired, linking it to information from previous processes; Step S2: Transmit data through an edge-cloud dual-link architecture, transmit real-time data through a 5G-U network, and transmit non-real-time data through an industrial Ethernet network; Establish a cross-process data association library to link the repair data with the previous process parameters; Step S3: Establish a scene-specific repair effect evaluation system and generate repair effect evaluation results; Establish an intelligent feedback model to generate feedback adjustment schemes for single defect repair scenarios as well as complex defect or secondary repair scenarios; Step S4: Convert the feedback adjustment scheme into standardized control commands, send them to the edge acquisition unit of the corresponding repair node and execute them by the repair equipment, collect the repair data after the feedback adjustment scheme is executed, and form a closed-loop control. The repair trajectory map of each carrier board is updated based on the real-time collected repair data and synchronized to the cross-process data association library in real time. Step S5: Based on historical repair data, iteratively update the intelligent feedback model and the repair effect evaluation system; Based on cross-process data association database data, provide feedback on parameter adjustment suggestions to the preceding process.

2. The method for real-time data transmission and feedback in the IC carrier board repair process according to claim 1, characterized in that, Step S1 specifically includes: Step S11, Repair node and defect type adaptation: The repair process for IC substrates can be summarized as follows: laser repair, electroplating soldering, mechanical polishing, and plasma repair. The laser repair includes: repairing pinholes on the carrier board and repairing open circuits on the carrier board. The electroplating repair welding includes: repairing defects in the coating of the carrier plate; The mechanical polishing includes: repairing scratches on the surface of the carrier plate; The plasma repair includes: removing residual impurities; Each repair process is assigned to a repair node, and each repair node is configured with a unique identifier, which is associated with the corresponding repair equipment model, repair process, and applicable defect range. Establish a defect classification and coding system, and standardize coding according to defect cause, defect form and defect level to uniformly represent defect information; The causes of the defects include: equipment-related issues, environmental factors, and substrate-related issues; The defect types mentioned include: point defects, line defects, and surface defects; The defect levels are: minor, moderate, and severe. Step S12, Edge Acquisition Unit Deployment and Data Traceability: Lightweight edge acquisition units are deployed at each repair node, and multiple communication protocols such as SECS / GEM, Modbus TCP, and EtherNet / IP are adapted through customized interfaces to establish real-time connections with repair equipment, online detection equipment, and environmental sensors. Each carrier board to be repaired is assigned a unique traceability code, which is linked to its previous process parameters, defect detection reports and repair work order information. The edge acquisition unit binds the repair data to the carrier board based on the traceability code to ensure the uniqueness of data traceability. Step S13, Real-time Acquisition and Preprocessing of Multi-Dimensional Data: The edge acquisition unit collects multi-source data during the repair process at a sampling frequency of 100Hz. The collected data includes: repair process parameters, equipment status data, defect and repair effect data, and environmental data. The repair process parameters include: laser repair power, welding temperature, polishing pressure, and repair time; The equipment status data includes: laser head wear, welding torch verticality, grinding head rotation speed, and equipment vibration value; The defect and repair effect data include: defect size, repair area, circuit impedance after repair, and surface flatness; The environmental data includes: temperature, humidity, cleanliness, and air pressure of the remediation area; Local preprocessing of the collected data includes: outlier removal, unit standardization, data compression, and feature extraction; The preprocessed data is encapsulated into standardized repair data frames containing carrier traceability codes, repair node identifiers, timestamps, data types, and structured data bodies, with a compression ratio of less than 10:

1.

3. The method for real-time data transmission and feedback in the IC carrier board repair process according to claim 1, characterized in that, Step S2 specifically includes: Step S21, Establishing a dual-link data transmission architecture: Build an edge-cloud dual-link transmission architecture; The standardized repair data frames are divided into two categories for transmission through the edge acquisition unit: data with high real-time requirements is transmitted through the 5G-U low-latency network, and non-real-time data is transmitted through the industrial Ethernet. Establish a data verification mechanism and verify data integrity using the CRC-32 checksum algorithm; Step S22, Repair the data bus and cross-process related library construction: A unified repair data bus is built to receive data uploaded by each edge acquisition unit, and the data is transmitted and classified for storage based on the traceability code of the carrier board. A cross-process data association library was constructed, integrating the process parameters of the previous process of the carrier board, defect detection data and repair process data. A data association model with the carrier board traceability code as the core was established, and the influence weight of the previous process parameters on the repair difficulty and repair effect was marked.

4. The method for real-time data transmission and feedback in the IC carrier board repair process according to claim 1, characterized in that, Step S3 specifically includes: Step S31, Construction of a multi-dimensional repair effect evaluation system: Based on defect classification coding, repair process specifications and historical repair data, a scenario-based repair effect evaluation system is established to generate repair effect evaluation results. The restoration effect evaluation results include: image segmentation and feature extraction of the image of the restoration area based on the U-Net model, obtaining feature parameters such as edge integrity, line continuity, and surface roughness of the restoration area. Edge integrity The normalized value of the Hausdorff distance between the repaired area and the original design outline is obtained by calculating: in, To repair the edge point set, To design the contour point set, To prevent the constant term from being divided by zero; Line continuity By extracting the skeleton of the repair area and performing connectivity analysis, the proportion of the longest connected path to the total design path is calculated as follows: in, This represents the actual connected length of the line after repair. This represents the theoretical length of the original design route; Surface roughness The contrast ratio, calculated from the gray-level co-occurrence matrix, is used as a surface texture evaluation metric. in, The gray-level co-occurrence matrix represents the gray-level co-occurrence matrix in The value at that location, Represents the gray level of an image. The square of the grayscale difference; The repair effectiveness evaluation results also include: a comprehensive repair effectiveness score calculated using the analytic hierarchy process (AHP) based on electrical performance testing data. in, For visual weight, For electrical weights, Weights for edge integrity feature parameters, The weights are the characteristic parameters of line continuity. For the surface roughness characteristic parameter weights, Weights for line impedance characteristic parameters. Weights for insulation resistance characteristic parameters This is the standard value of the line impedance. This is the standard value for insulation resistance; The comprehensive score for the repair effect includes three levels: excellent, qualified, and unqualified. A score of 90 or above is excellent, a score between 70 and 89 is qualified, and a score less than 70 is unqualified. The traceability code for carrier boards requiring secondary repair is automatically marked based on the comprehensive score of the repair effect. Step S32, Dynamic Feedback Parameter Generation Mechanism: Based on the cross-process data association library and the repair effect evaluation results, an intelligent feedback model is established based on deep reinforcement learning methods. For single defect repair scenarios, an association rule mining algorithm is used to discover the matching patterns between repair parameters and defect type and defect size, and to generate a feedback adjustment scheme for parameter fine-tuning. For complex defects or scenarios requiring secondary repair, the root cause of poor repair results is located by combining the parameter deviation data of the previous process and using a causal inference model. A multi-parameter collaborative feedback adjustment scheme is generated, which clarifies the parameter adjustment range, timing and priority.

5. The method for real-time data transmission and feedback in the IC carrier board repair process according to claim 1, characterized in that, Step S4 specifically includes: Step S41, hierarchical issuance and execution of instructions: By repairing the data bus, the parameters in the feedback adjustment scheme are converted into standardized control commands, which are then sent to the edge acquisition unit corresponding to the repair node according to the node's identity. After receiving and verifying the legality of the control command, the edge acquisition unit sends it to the repair equipment in real time to perform the adjustment according to the feedback adjustment plan, and collects the adjusted repair data; In case of fatal parameter deviations, the repair equipment will automatically be shut down and an alarm will be issued to ensure the safety of the repair process; Step S42, Repair the dynamic update and tracing of the trajectory: Based on real-time collected repair data, the repair trajectory map of each carrier board is dynamically updated; The repair trajectory map uses a time axis as a guide to fully record the defect location, changes in repair parameters, repair effect evaluation results, and operator information; The repair trajectory map is synchronized to the cross-process data association library in real time.

6. The method for real-time data transmission and feedback in the IC carrier board repair process according to claim 1, characterized in that, Step S5 specifically includes: The entire repair process data, feedback adjustment plans, and repair effect evaluation results are tagged by carrier board batch, defect type, and repair node, and stored in the historical database. Establish a data cleaning mechanism to regularly remove invalid and duplicate data; By mining the optimization patterns of various data parameters in the historical database through a multilayer perceptron network model, the intelligent feedback model and the repair effect evaluation system are iteratively updated to optimize the initial repair parameter range for different defect types. Based on data from the cross-process data association library, provide parameter adjustment suggestions to the preceding process.

7. A real-time data transmission and feedback system for IC substrate repair process, characterized in that, include: The multi-source data acquisition system construction module includes: Deploy edge acquisition units to collect repair process parameters, equipment status data, defect and repair effect data, and environmental data in real time; Assign a unique traceability code to each carrier board to be repaired, linking it to information from previous processes; The cross-process data association module contains: Data is transmitted via an edge-cloud dual-link architecture, real-time data is transmitted via 5G-U network, and non-real-time data is transmitted via industrial Ethernet; Establish a cross-process data association library to link the repair data with the previous process parameters; Repair assessment and feedback module, in which: Establish a scenario-based repair effect evaluation system and generate repair effect evaluation results; Establish an intelligent feedback model to generate feedback adjustment schemes for single defect repair scenarios as well as complex defect or secondary repair scenarios; The command issuance and closed-loop control module contains: The feedback adjustment scheme is transformed into standardized control commands, which are then sent to the edge acquisition units of the corresponding repair nodes and executed by the repair equipment. The repair data after the feedback adjustment scheme is executed is collected to form a closed-loop control. The repair trajectory map of each carrier board is updated based on the real-time collected repair data and synchronized to the cross-process data association library in real time. The data accumulation and process optimization module includes: Based on historical restoration data, the intelligent feedback model and restoration effect evaluation system are iteratively updated; Based on cross-process data association database data, provide feedback on parameter adjustment suggestions to the preceding process.

8. The real-time data transmission and feedback system for IC carrier board repair process according to claim 7, characterized in that, The multi-source data acquisition system construction module specifically includes: a repair node and defect type adaptation sub-module, an edge acquisition unit deployment and data tracing sub-module, and a multi-dimensional data real-time acquisition and preprocessing sub-module; The aforementioned repair node and defect type adaptation submodule includes: The repair process for IC substrates can be summarized as follows: laser repair, electroplating soldering, mechanical polishing, and plasma repair. The laser repair includes: repairing pinholes on the carrier board and repairing open circuits on the carrier board. The electroplating repair welding includes: repairing defects in the coating of the carrier plate; The mechanical polishing includes: repairing scratches on the surface of the carrier plate; The plasma repair includes: removing residual impurities; Each repair process is assigned to a repair node, and each repair node is configured with a unique identifier, which is associated with the corresponding repair equipment model, repair process, and applicable defect range. Establish a defect classification and coding system, and standardize coding according to defect cause, defect form and defect level to uniformly represent defect information; The causes of the defects include: equipment-related issues, environmental factors, and substrate-related issues; The defect types mentioned include: point defects, line defects, and surface defects; The defect levels are: minor, moderate, and severe. The edge acquisition unit deployment and data traceability submodule includes: Lightweight edge acquisition units are deployed at each repair node, and multiple communication protocols such as SECS / GEM, Modbus TCP, and EtherNet / IP are adapted through customized interfaces to establish real-time connections with repair equipment, online detection equipment, and environmental sensors. Each carrier board to be repaired is assigned a unique traceability code, which is linked to its previous process parameters, defect detection reports and repair work order information. The edge acquisition unit binds the repair data to the carrier board based on the traceability code to ensure the uniqueness of data traceability. The aforementioned multi-dimensional data real-time acquisition and preprocessing submodule includes: The edge acquisition unit collects multi-source data during the repair process at a sampling frequency of 100Hz. The collected data includes: repair process parameters, equipment status data, defect and repair effect data, and environmental data. The repair process parameters include: laser repair power, welding temperature, polishing pressure, and repair time; The equipment status data includes: laser head wear, welding torch verticality, grinding head rotation speed, and equipment vibration value; The defect and repair effect data include: defect size, repair area, circuit impedance after repair, and surface flatness; The environmental data includes: temperature, humidity, cleanliness, and air pressure of the remediation area; Local preprocessing of the collected data includes: outlier removal, unit standardization, data compression, and feature extraction; The preprocessed data is encapsulated into standardized repair data frames containing carrier traceability codes, repair node identifiers, timestamps, data types, and structured data bodies, with a compression ratio of less than 10:

1.

9. A real-time data transmission and feedback system for IC carrier board repair process according to claim 7, characterized in that, The cross-process data association module specifically includes: a dual-link data transmission architecture construction submodule, a data bus repair and cross-process association library construction submodule; The aforementioned dual-link data transmission architecture construction submodule includes: Build an edge-cloud dual-link transmission architecture; The standardized repair data frames are divided into two categories for transmission through the edge acquisition unit: data with high real-time requirements is transmitted through the 5G-U low-latency network, and non-real-time data is transmitted through the industrial Ethernet. Establish a data verification mechanism and verify data integrity using the CRC-32 checksum algorithm; The aforementioned repair data bus and cross-process association library construction submodule includes: A unified repair data bus is built to receive data uploaded by each edge acquisition unit, and the data is transmitted and classified for storage based on the traceability code of the carrier board. Construct a cross-process data association library, integrating the process parameters of the preceding process of the carrier board, defect detection data and repair process data, and establish a full life cycle data association model with the carrier board traceability code as the core, and mark the influence weight of the preceding process parameters on the repair difficulty and repair effect.

10. A real-time data transmission and feedback system for IC carrier board repair process according to claim 7, characterized in that, The aforementioned repair assessment and feedback module specifically includes: a sub-module for constructing a multi-dimensional repair effect assessment system and a sub-module for generating dynamic feedback parameters; The aforementioned multi-dimensional repair effect evaluation system construction sub-module includes: Based on defect classification coding, repair process specifications and historical repair data, a scenario-based repair effect evaluation system is established to generate repair effect evaluation results. The restoration effect evaluation results include: image segmentation and feature extraction of the restored area based on the U-Net model, obtaining feature parameters such as edge integrity, line continuity, and surface roughness. Edge integrity The normalized value of the Hausdorff distance between the repaired area and the original design outline is obtained by calculating: in, To repair the edge point set, To design the contour point set, To prevent the constant term from being divided by zero; Line continuity By extracting the skeleton of the repair area and performing connectivity analysis, the proportion of the longest connected path to the total design path is calculated as follows: in, This represents the actual connected length of the line after repair. This represents the theoretical length of the original design route; Surface roughness The contrast ratio, calculated from the gray-level co-occurrence matrix, is used as a surface texture evaluation metric. in, The gray-level co-occurrence matrix represents the gray-level co-occurrence matrix in The value at that location, Represents the gray level of an image. The square of the grayscale difference; The repair effectiveness evaluation results also include: a comprehensive repair effectiveness score calculated using the analytic hierarchy process (AHP) based on electrical performance testing data. in, For visual weight, For electrical weights, Weights for edge integrity feature parameters, The weights are the characteristic parameters of line continuity. For the surface roughness characteristic parameter weights, Weights for line impedance characteristic parameters. Weights for insulation resistance characteristic parameters This is the standard value of the line impedance. This is the standard value for insulation resistance; The comprehensive score for the repair effect includes three levels: excellent, qualified, and unqualified. A score of 90 or above is excellent, a score between 70 and 89 is qualified, and a score less than 70 is unqualified. The traceability code for carrier boards requiring secondary repair is automatically marked based on the comprehensive score of the repair effect. The dynamic feedback parameter generation mechanism submodule includes: Based on the cross-process data association library and the repair effect evaluation results, an intelligent feedback model is established based on deep reinforcement learning methods. For single defect repair scenarios, an association rule mining algorithm is used to discover the matching patterns between repair parameters and defect type and defect size, and to generate a feedback adjustment scheme for parameter fine-tuning. For complex defects or scenarios requiring secondary repair, the root cause of poor repair results is located by combining the parameter deviation data of the previous process and using a causal inference model. A multi-parameter collaborative feedback adjustment scheme is generated, which clarifies the parameter adjustment range, timing and priority.

Images