A method for full-process traceability in circuit board manufacturing systems

Abstract

This invention relates to the field of circuit board manufacturing technology, specifically disclosing a full-process traceability method for circuit board production systems. It calculates the stability characteristic values ​​of the electroplating solution using a sliding window difference algorithm combined with K-means clustering to evaluate the stability of the electroplating process. It also employs empirical mode decomposition to extract multi-scale energy features of the lamination temperature difference signal, constructing abnormal temperature difference characteristic values ​​to determine their impact on structural integrity. These two types of characteristic values ​​are then fused into a comprehensive process feature vector, which is input into a trained and optimized random forest model for anomaly scoring output. This identifies abnormal process nodes and triggers quality warnings and traceability mechanisms. This invention achieves dynamic monitoring, intelligent analysis, and closed-loop traceability of key process parameters throughout the entire circuit board production process, improving product quality consistency and the level of intelligent manufacturing.

Description

Technical Field

[0001] This invention relates to the field of circuit board manufacturing technology, and more specifically to a method for full-process traceability of a circuit board production system. Background Technology

[0002] In circuit board manufacturing, electroplating and lamination processes have a decisive impact on the electrical performance and structural integrity of the product. Fluctuations in the concentration of trace elements such as copper, nickel, gold, and silver in the electroplating solution directly affect the plating thickness and adhesion, while abnormal changes in the temperature difference at the lamination interface can lead to thermal stress imbalance in the material, causing delamination or warping defects. Therefore, real-time monitoring and analysis of these key parameters are crucial for ensuring product quality stability. Currently, the industry generally uses monitoring methods based on fixed threshold judgments or simple statistical methods, lacking in-depth analysis of the dynamic characteristics of the process, making it difficult to achieve early identification and precise control of potential anomalies.

[0003] The existing technology has the following shortcomings:

[0004] Existing circuit board manufacturing process monitoring systems often neglect local state modeling of electroplating solution concentration change rates and multi-scale feature extraction of lamination temperature signals, resulting in an inability to effectively capture minute but persistent process disturbances. For example, traditional methods typically only use moving averages or variance as stability indicators, failing to introduce K-means clustering to classify concentration change rates into stable or fluctuating states, and failing to utilize empirical mode decomposition to extract nonlinear and non-stationary features from temperature difference data, thus missing a large amount of implicit information reflecting abnormal process trends. This technical limitation makes the system have low warning sensitivity and a high false alarm rate when facing complex operating conditions, making it difficult to meet the dual requirements of real-time performance and accuracy in intelligent manufacturing. Summary of the Invention

[0005] The purpose of this invention is to provide a method for full-process traceability of circuit board production systems to solve the problems mentioned above.

[0006] The objective of this invention can be achieved through the following technical solutions:

[0007] A method for end-to-end traceability in a circuit board manufacturing system includes the following steps:

[0008] S1: During the circuit board manufacturing process, monitor the concentration data of trace elements in the electroplating solution and the temperature difference data of the lamination interface in real time;

[0009] S2: Analyze the concentration data of trace elements, calculate the stability characteristic value of the electroplating solution based on the stability of its rate of change, and use it to evaluate the stability of the electroplating process.

[0010] S3: Analyze the temperature difference data of the lamination interface, calculate the abnormal characteristic value of the temperature difference based on its stability, and use it to assess the impact of the abnormal temperature of the lamination interface on the integrity of the circuit board structure.

[0011] S4: Construct a comprehensive process feature vector by combining the stability feature value of the electroplating solution with the abnormal feature value of temperature difference. Input the vector into the artificial intelligence analysis model for fusion analysis. Based on the analysis results, identify abnormal process nodes and trigger corresponding quality warning and traceability mechanisms to achieve refined control and traceability of the entire production process.

[0012] As a further aspect of the present invention: the evaluation of the stability of the electroplating process specifically includes:

[0013] During the circuit board manufacturing process, the concentration data of trace elements in the electroplating solution is obtained, and the data is analyzed. Based on the stability of the change rate, the stability characteristic value of the electroplating solution is calculated. It is then determined whether the stability characteristic value of the electroplating solution is greater than or equal to a preset threshold. If it is, the electroplating process is unstable; otherwise, the electroplating process is stable.

[0014] As a further aspect of the present invention: the process for obtaining the stability characteristic value of the electroplating solution is as follows:

[0015] The concentration data of trace elements in the electroplating solution during the electroplating process are collected in real time according to the time series and constructed into time series data.

[0016] A sliding window differencing process is applied to the time series data to obtain the concentration change rate series;

[0017] The concentration change rates at all collected time points were integrated into a concentration change rate sequence.

[0018] The concentration change rate sequence was subjected to binary clustering using the K-means clustering algorithm, with a set number of clusters. , representing stable change and violent fluctuation states respectively, and calculate the corresponding cluster centers. and ;

[0019] Calculate the local stability factor based on the cluster centers;

[0020] The stability characteristic value of the electroplating solution was calculated using a weighted average method.

[0021] As a further aspect of the present invention: the assessment of the impact of abnormal lamination interface temperature on the structural integrity of the circuit board specifically includes:

[0022] During the circuit board manufacturing process, the temperature difference data of the lamination interface is acquired, analyzed, and the temperature difference abnormality characteristic value is calculated based on its stability. The abnormality characteristic value of the temperature difference is then calculated to determine whether it is greater than or equal to a preset threshold. If it is, the abnormal temperature of the lamination interface has a serious impact on the structural integrity of the circuit board; if not, the abnormal temperature of the lamination interface has a slight impact on the structural integrity of the circuit board.

[0023] As a further aspect of the present invention: the process for obtaining the abnormal temperature difference characteristic value is as follows:

[0024] Acquire temperature difference data sequence at the lamination interface during circuit board manufacturing;

[0025] The temperature difference data is denoised and smoothed to eliminate measurement errors and noise interference, resulting in a preprocessed temperature difference data sequence.

[0026] The empirical mode decomposition method is applied to decompose the preprocessed temperature difference data sequence into several intrinsic mode functions;

[0027] For each intrinsic mode function, calculate the energy of each intrinsic mode function;

[0028] The ratio of the mean energy of all intrinsic mode functions to their sum is calculated to obtain the abnormal characteristic value of temperature difference.

[0029] As a further aspect of the present invention: the step of constructing a comprehensive process feature vector from the stability characteristic values ​​and temperature difference abnormality characteristic values ​​of the electroplating solution, and inputting it into an artificial intelligence analysis model for fusion analysis, specifically includes:

[0030] The stability feature value and temperature difference anomaly feature value of the electroplating solution during the circuit board production process are obtained. The stability feature value and temperature difference anomaly feature value of the electroplating solution are used to construct a comprehensive process feature vector, which is used as the input of the artificial intelligence analysis model to minimize the error between the predicted process node anomaly score and the actual process node anomaly score. This is used as the training target of the model to train the artificial intelligence analysis model. Based on the trained artificial intelligence analysis model, the process node anomaly score is output. The artificial intelligence analysis model is a random forest model.

[0031] As a further aspect of the present invention: the training process of the artificial intelligence analysis model is as follows:

[0032] Historical electroplating solution stability features and lamination interface temperature difference anomaly features were collected and combined with manually labeled actual process node anomaly scores as target labels to construct a historical dataset. The entire dataset was divided into training and validation sets according to a certain ratio for model training and performance evaluation. A random forest algorithm was used for modeling, with hyperparameters such as the number of base learners, maximum tree depth, and minimum sample splitting threshold set. During training, each decision tree was split by randomly selecting samples and features, using information gain to select the optimal split point, gradually generating a complete decision tree structure. The prediction results of all decision trees were evaluated through voting. Alternatively, a weighted average method can be used for integration to obtain the final predicted scores for process node anomalies. After each training iteration, the mean squared error between the predicted score and the actual score is calculated as a performance evaluation metric for the model. Grid search and Bayesian optimization strategies are introduced to automatically tune the model's hyperparameters to further improve model accuracy. When the mean squared error on the validation set tends to stabilize and reaches its minimum value, model training is complete. The trained random forest model is deployed to the actual production system, receiving in real time a comprehensive process feature vector composed of electroplating solution stability feature values ​​and temperature difference anomaly feature values, and outputting the corresponding process node anomaly scores.

[0033] As a further aspect of the present invention: the step of identifying abnormal process nodes based on the analysis results specifically includes:

[0034] Determine whether the abnormal score of a process node in the circuit board production process is greater than or equal to a preset threshold. If yes, mark it as an abnormal process node; otherwise, mark it as a normal process node.

[0035] As a further aspect of the present invention: the triggering of the corresponding quality early warning and traceability mechanism to achieve refined control and traceability of the entire production process specifically includes:

[0036] If the abnormal score of the process node output by the artificial intelligence analysis model is greater than or equal to the preset quality warning threshold, the system will automatically trigger a multi-level quality warning mechanism and send warning information to the operator or the central control system. The warning information includes the name of the process node where the abnormality occurred, the time of occurrence, and the current score value. The system will record the abnormal event in the database and generate a unique identifier for subsequent quality traceability.

[0037] The beneficial effects of this invention are:

[0038] (1) This invention innovatively adopts a method combining sliding window difference and K-means clustering to perform refined analysis on the time series data of the concentration of trace elements such as copper, nickel, gold and silver in the electroplating solution. By extracting the concentration change rate sequence and clustering it to generate two types of states: stable and fluctuating, the local stability factor at each time point is calculated based on the cluster center. Finally, the stability characteristic value of the electroplating solution is obtained by weighted averaging, thereby realizing the quantitative assessment of the dynamic stability of the electroplating process. At the same time, in terms of lamination interface temperature monitoring, this invention introduces empirical mode decomposition technology to adaptively decompose the preprocessed temperature difference signal into multiple intrinsic mode functions. By calculating the statistical characteristics of the energy distribution of each intrinsic mode function, a temperature difference abnormality characteristic value with the ratio of the energy mean to the sum is constructed, which effectively reveals the multi-scale characteristics and potential abnormal trends of temperature fluctuations. On this basis, the above two types of characteristic values ​​are fused to form a comprehensive process feature vector, which is used as input to a random forest model trained and optimized by historical data. Through the information gain splitting strategy and ensemble learning mechanism, the process node abnormality score is output, realizing high-precision identification and intelligent early warning of the abnormal state of key process nodes in the entire circuit board production process.

[0039] (2) This invention constructs a comprehensive feature vector with process representativeness by combining the stability feature value of the electroplating solution with the abnormal feature value of the temperature difference at the lamination interface. This vector is then used as input data to import a random forest analysis model trained with historical data and optimized with hyperparameters. The model utilizes the ability of multiple decision trees in its ensemble learning structure to model the sample features from multiple perspectives, thereby achieving high-precision scoring output for the abnormal status of key process nodes in the circuit board production process. The model takes minimizing the mean square error between the predicted score and the manually labeled score as the objective function. During the training process, it combines grid search and Bayesian optimization strategies to automatically adjust hyperparameters such as the number of base learners, maximum depth, and splitting threshold, ensuring that the model has good generalization ability and robustness. In practical applications, the system intelligently judges whether there is an abnormality in the current process node based on whether the output abnormal score exceeds the preset threshold, and triggers a multi-level quality early warning mechanism accordingly, sending an early warning notification containing key information such as the type of abnormality, the time of occurrence, and the scope of impact to the operator or the central control system. Attached Figure Description

[0040] The invention will now be further described with reference to the accompanying drawings.

[0041] Figure 1 This is a flowchart of a full-process traceability method for a circuit board production system according to the present invention. Detailed Implementation

[0042] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0043] Please see Figure 1 As shown, this invention provides a method for full-process traceability in a circuit board manufacturing system, comprising the following steps:

[0044] S1: During the circuit board manufacturing process, monitor the concentration data of trace elements in the electroplating solution and the temperature difference data of the lamination interface in real time;

[0045] S2: Analyze the concentration data of trace elements, calculate the stability characteristic value of the electroplating solution based on the stability of its rate of change, and use it to evaluate the stability of the electroplating process.

[0046] S3: Analyze the temperature difference data of the lamination interface, calculate the abnormal characteristic value of the temperature difference based on its stability, and use it to assess the impact of the abnormal temperature of the lamination interface on the integrity of the circuit board structure.

[0047] S4: Construct a comprehensive process feature vector by combining the stability feature value of the electroplating solution with the abnormal feature value of temperature difference. Input the vector into the artificial intelligence analysis model for fusion analysis. Based on the analysis results, identify abnormal process nodes and trigger corresponding quality warning and traceability mechanisms to achieve refined control and traceability of the entire production process.

[0048] In S1, during the circuit board manufacturing process, real-time monitoring of trace element concentration data and lamination interface temperature difference data in the electroplating solution is performed, specifically including:

[0049] During circuit board manufacturing, real-time monitoring of trace element concentrations in the electroplating solution is crucial, including precise measurements of several key elements. These trace elements primarily include copper (Cu), nickel (Ni), gold (Au), and silver (Ag), which are essential for ensuring the quality and performance of the electroplated layer. Using advanced electrochemical analysis instruments, such as atomic absorption spectrometry (AAS), concentration data for these elements can be acquired in real-time on the production line. This process involves automatically introducing trace electroplating solution samples into the analytical instrument and processing and monitoring the data through a software system, ensuring that any changes that may affect product quality are detected promptly.

[0050] Simultaneously, during circuit board manufacturing, real-time monitoring of the lamination interface temperature difference is necessary. This step typically involves using high-precision temperature sensors, such as thermocouples or resistance temperature detectors (RTDs), installed in critical locations to accurately record temperature changes during lamination. First, multiple temperature sensors are installed in different areas of the laminator to ensure comprehensive coverage and monitoring of the temperature distribution across the entire lamination interface. Then, a data acquisition system collects temperature readings from each sensor, and filtering and smoothing algorithms remove noise interference to ensure data accuracy. Finally, this processed temperature difference data is used in subsequent analysis steps to assess the stability of the lamination process and its impact on the structural integrity of the circuit board.

[0051] In S2, the trace element concentration data are analyzed, and the stability characteristic value of the electroplating solution is calculated based on the stability of its rate of change. This value is used to evaluate the stability of the electroplating process, specifically including:

[0052] During the circuit board manufacturing process, the concentration data of trace elements in the electroplating solution is obtained, and the data is analyzed. Based on the stability of the change rate, the stability characteristic value of the electroplating solution is calculated. It is then determined whether the stability characteristic value of the electroplating solution is greater than or equal to a preset threshold. If it is, the electroplating process is unstable; otherwise, the electroplating process is stable.

[0053] The process for obtaining the stability characteristic value of the electroplating solution is as follows:

[0054] The concentration data of trace elements in the electroplating solution during the electroplating process are collected in real time according to the time series and constructed into time series data.

[0055] By performing sliding window differencing on the time series data, the concentration change rate series is obtained, and the calculation expression is as follows: ;

[0056] in, This indicates the set length of the sliding window. Indicates the first Each data collection time point Indicates the first Concentration values ​​at each collection time point, Indicates the first Concentration change rate at each collection time point;

[0057] The concentration change rates at all collected time points were integrated into a concentration change rate sequence.

[0058] The concentration change rate sequence was subjected to binary clustering using the K-means clustering algorithm, with a set number of clusters. , representing stable change and violent fluctuation states respectively, and calculate the corresponding cluster centers. and ;

[0059] The local stability factor is calculated based on the cluster centers, and the expression for the calculation is as follows: ;

[0060] in, Indicates the first The local stability factor at each acquisition time point, if A value close to 0 indicates that the corresponding sample point is more likely to be in a stable state. A value close to 1 indicates that the corresponding sample point is more likely to be in an unstable state; Cluster centers representing a stable state, Cluster centers representing unstable states This represents a small constant that prevents division by zero.

[0061] The stability characteristic value of the electroplating solution is calculated using a weighted average method, and the calculation expression is as follows: ;

[0062] in, Indicates the stability characteristic value of the electroplating solution. Indicates the weighting coefficient. This represents the total number of data collection points.

[0063] It should be noted that this invention introduces a method for calculating the stability characteristic value of electroplating solution based on sliding window difference and K-means clustering in the evaluation of electroplating process stability. It innovatively maps time-series data of trace element concentration change rates to local stability factors through cluster analysis, and further weights and fuses these factors to obtain a global stability index. This method can effectively capture the trend characteristics and sudden disturbances of trace component fluctuations during electroplating, thereby achieving a quantitative judgment of process stability. Compared with traditional methods that rely solely on fixed thresholds or simple variance analysis, this invention has stronger dynamic adaptability and anomaly identification capabilities, accurately distinguishing between stable and unstable process states, improving the precision of quality control in circuit board production, and possessing significant technological advancement and practical application value.

[0064] In S3, the lamination interface temperature difference data is analyzed, and based on its stability, abnormal temperature difference characteristic values ​​are calculated to assess the impact of lamination interface temperature anomalies on the integrity of the circuit board structure. Specifically, this includes:

[0065] During the circuit board manufacturing process, the temperature difference data of the lamination interface is acquired, analyzed, and the temperature difference abnormality characteristic value is calculated based on its stability. The abnormality characteristic value of the temperature difference is then calculated to determine whether it is greater than or equal to a preset threshold. If it is, the abnormal temperature of the lamination interface has a serious impact on the structural integrity of the circuit board; if not, the abnormal temperature of the lamination interface has a slight impact on the structural integrity of the circuit board.

[0066] The process for obtaining the abnormal temperature difference characteristic value is as follows:

[0067] Acquire temperature difference data sequence at the lamination interface during circuit board manufacturing;

[0068] The temperature difference data is denoised and smoothed to eliminate measurement errors and noise interference, resulting in a preprocessed temperature difference data sequence.

[0069] The empirical mode decomposition method is applied to decompose the preprocessed temperature difference data sequence into several intrinsic mode functions;

[0070] For each intrinsic mode function, the energy of each intrinsic mode function is calculated using the following expression: ;

[0071] in, This indicates the number of intrinsic mode functions. This represents the total number of intrinsic mode functions. Indicates the first The th intrinsic mode function One sample point, Indicates the number of sample points. Indicates the first The energy of an intrinsic mode function;

[0072] The ratio of the mean energy of all intrinsic mode functions to their sum is calculated to obtain the abnormal characteristic value of temperature difference.

[0073] It should be noted that this invention innovatively constructs a method for calculating temperature difference anomaly eigenvalues ​​based on the energy distribution of intrinsic mode functions by introducing empirical mode decomposition (EMD) technology to extract multi-scale features from lamination interface temperature difference data. This method can adaptively decompose complex temperature difference signals into multiple stable modal components, and reflect the overall stability and local abrupt changes of temperature fluctuations through the ratio of the energy mean to the sum, thereby achieving a more accurate and quantitative assessment of the stability of the lamination process. Compared with traditional threshold judgment or variance analysis methods, this invention significantly improves the ability to identify minute but potentially destructive temperature anomalies, effectively distinguishes different process states that affect the structural integrity of circuit boards, and enhances the sensitivity and reliability of quality early warning in the production process, demonstrating good engineering application prospects and technological innovation value.

[0074] In S4, the stability characteristic values ​​of the electroplating solution and the abnormal temperature difference characteristic values ​​are constructed into a comprehensive process feature vector, which is then input into an artificial intelligence analysis model for fusion analysis. Based on the analysis results, abnormal process nodes are identified, and corresponding quality warnings and traceability mechanisms are triggered to achieve refined control and traceability throughout the entire production process. Specifically, this includes:

[0075] The stability feature value and temperature difference anomaly feature value of the electroplating solution during the circuit board production process are obtained. The stability feature value and temperature difference anomaly feature value of the electroplating solution are used to construct a comprehensive process feature vector, which is used as the input of the artificial intelligence analysis model to minimize the error between the predicted process node anomaly score and the actual process node anomaly score. This is used as the training target of the model to train the artificial intelligence analysis model. Based on the trained artificial intelligence analysis model, the process node anomaly score is output. The artificial intelligence analysis model is a random forest model.

[0076] The training process of the artificial intelligence analysis model is as follows:

[0077] Historical electroplating solution stability features and lamination interface temperature difference anomaly features were collected and combined with manually labeled actual process node anomaly scores as target labels to construct a historical dataset. The entire dataset was divided into training and validation sets according to a certain ratio for model training and performance evaluation. A random forest algorithm was used for modeling, with hyperparameters such as the number of base learners, maximum tree depth, and minimum sample splitting threshold set. During training, each decision tree was split by randomly selecting samples and features, using information gain to select the optimal split point, gradually generating a complete decision tree structure. The prediction results of all decision trees were evaluated through voting. Alternatively, a weighted average method can be used for integration to obtain the final predicted scores for process node anomalies. After each training iteration, the mean squared error between the predicted score and the actual score is calculated as a performance evaluation metric for the model. Grid search and Bayesian optimization strategies are introduced to automatically tune the model's hyperparameters to further improve model accuracy. When the mean squared error on the validation set tends to stabilize and reaches its minimum value, model training is complete. The trained random forest model is deployed to the actual production system, receiving in real time a comprehensive process feature vector composed of electroplating solution stability feature values ​​and temperature difference anomaly feature values, and outputting the corresponding process node anomaly scores.

[0078] Determine whether the abnormal score of a process node in the circuit board production process is greater than or equal to a preset threshold. If yes, mark it as an abnormal process node; otherwise, mark it as a normal process node.

[0079] If the abnormal score of the process node output by the artificial intelligence analysis model is greater than or equal to the preset quality warning threshold, the system will automatically trigger a multi-level quality warning mechanism and send warning information to the operator or the central control system. The warning information includes the name of the process node where the abnormality occurred, the time of occurrence, and the current score value. The system will record the abnormal event in the database and generate a unique identifier for subsequent quality traceability.

[0080] The working principle of this invention: This invention provides a technical solution for full-process traceability in circuit board production systems. It constructs a multi-dimensional process parameter acquisition system by real-time monitoring of the concentration data of trace elements such as copper, nickel, gold, and silver in the electroplating solution and the temperature difference data at the lamination interface during the production process. For the electroplating solution concentration data, a sliding window difference combined with K-means clustering is used to extract the concentration change rate feature, and the local stability factor and weighted average are calculated to obtain the stability feature value of the electroplating solution, achieving dynamic evaluation of the stability of the electroplating process. For the lamination interface temperature difference data, noise reduction and smoothing are first performed, and then empirical mode decomposition is applied to extract the energy distribution of each intrinsic mode function. The ratio of the energy mean to the sum is used to obtain the temperature difference anomaly feature value, which is used to determine the degree of influence of temperature fluctuations on structural integrity. The above two types of feature values ​​are constructed into a comprehensive process feature vector, which is input into a random forest model trained and optimized with historical data for fusion analysis. The output is a process node anomaly score, identifying abnormal process nodes and triggering quality warnings and traceability mechanisms, thereby achieving refined control and traceability management of the entire circuit board production process. This technical solution integrates signal processing, cluster analysis, and machine learning algorithms, improving the accuracy and response speed of process anomaly identification.

[0081] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0082] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0083] It should be understood that the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. A and B can be singular or plural. Additionally, the character " / " in this article generally indicates an "or" relationship between the preceding and following related objects, but it can also represent an "and / or" relationship. Please refer to the context for a more accurate understanding.

[0084] It should be understood that in the various embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0085] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A method for full-process traceability in a circuit board manufacturing system, characterized in that, Includes the following steps: S1: During the circuit board manufacturing process, monitor the concentration data of trace elements in the electroplating solution and the temperature difference data of the lamination interface in real time; S2: Analyze the concentration data of trace elements, calculate the stability characteristic value of the electroplating solution based on the stability of its rate of change, and use it to evaluate the stability of the electroplating process. The process for obtaining the stability characteristic value of the electroplating solution is as follows: The concentration data of trace elements in the electroplating solution during the electroplating process are collected in real time according to the time series and constructed into time series data. A sliding window differencing process is applied to the time series data to obtain the concentration change rate series; The concentration change rates at all collected time points were integrated into a concentration change rate sequence. The concentration change rate sequence was subjected to binary clustering using the K-means clustering algorithm, with a set number of clusters. , representing stable change and violent fluctuation states respectively, and calculate the corresponding cluster centers. and ; Calculate the local stability factor based on the cluster centers; The stability characteristic value of the electroplating solution was calculated using a weighted average method. S3: Analyze the temperature difference data of the lamination interface, calculate the abnormal characteristic value of the temperature difference based on its stability, and use it to assess the impact of the abnormal temperature of the lamination interface on the integrity of the circuit board structure. The process for obtaining the abnormal temperature difference characteristic value is as follows: Acquire temperature difference data sequence at the lamination interface during circuit board manufacturing; The temperature difference data is denoised and smoothed to eliminate measurement errors and noise interference, resulting in a preprocessed temperature difference data sequence. The empirical mode decomposition method is applied to decompose the preprocessed temperature difference data sequence into several intrinsic mode functions; For each intrinsic mode function, calculate the energy of each intrinsic mode function; Calculate the ratio of the mean energy to the sum of all intrinsic mode functions to obtain the abnormal characteristic value of temperature difference; S4: Construct a comprehensive process feature vector by combining the stability feature value of the electroplating solution with the abnormal feature value of temperature difference. Input the vector into the artificial intelligence analysis model for fusion analysis. Based on the analysis results, identify abnormal process nodes and trigger corresponding quality warning and traceability mechanisms to achieve refined control and traceability of the entire production process.

2. The method for full-process traceability of a circuit board production system according to claim 1, characterized in that, The evaluation of the stability of the electroplating process specifically includes: During the circuit board manufacturing process, the concentration data of trace elements in the electroplating solution is obtained, and the data is analyzed. Based on the stability of the change rate, the stability characteristic value of the electroplating solution is calculated. It is then determined whether the stability characteristic value of the electroplating solution is greater than or equal to a preset threshold. If it is, the electroplating process is unstable; otherwise, the electroplating process is stable.

3. The method for full-process traceability of a circuit board production system according to claim 1, characterized in that, The assessment of the impact of abnormal lamination interface temperature on the structural integrity of the circuit board specifically includes: During the circuit board manufacturing process, the temperature difference data of the lamination interface is acquired, analyzed, and the temperature difference abnormality characteristic value is calculated based on its stability. It is then determined whether the temperature difference abnormality characteristic value is greater than or equal to a preset threshold. If it is, the abnormal temperature of the lamination interface has a serious impact on the structural integrity of the circuit board; otherwise, the abnormal temperature of the lamination interface has a slight impact on the structural integrity of the circuit board.

4. The method for full-process traceability of a circuit board production system according to claim 1, characterized in that, The process of constructing a comprehensive process feature vector from the stability characteristic values ​​and temperature difference abnormality characteristic values ​​of the electroplating solution, and inputting it into an artificial intelligence analysis model for fusion analysis, specifically includes: The stability feature value and temperature difference anomaly feature value of the electroplating solution during the circuit board production process are obtained. The stability feature value and temperature difference anomaly feature value of the electroplating solution are used to construct a comprehensive process feature vector, which is used as the input of the artificial intelligence analysis model to minimize the error between the predicted process node anomaly score and the actual process node anomaly score. This is used as the training target of the model to train the artificial intelligence analysis model. Based on the trained artificial intelligence analysis model, the process node anomaly score is output. The artificial intelligence analysis model is a random forest model.

5. The method for full-process traceability of a circuit board production system according to claim 4, characterized in that, The training process of the artificial intelligence analysis model is as follows: Historical electroplating solution stability features and lamination interface temperature difference anomaly features were collected and combined with manually labeled actual process node anomaly scores as target labels to construct a historical dataset. The entire dataset was divided into training and validation sets according to a certain ratio for model training and performance evaluation. A random forest algorithm was used for modeling, and hyperparameters such as the number of base learners, maximum tree depth, and minimum sample splitting threshold were set. During training, each decision tree was split by randomly selecting samples and features, and information gain was used to select the optimal split point to gradually generate a complete decision tree structure. The prediction results of all decision trees were integrated by voting or weighted averaging to obtain the final process node anomaly score prediction value. After each training iteration, the mean squared error between the predicted score and the actual score is calculated as a model performance evaluation metric. Grid search and Bayesian optimization strategies are introduced to automatically tune the model hyperparameters to improve model accuracy. When the mean squared error on the validation set tends to stabilize and reaches its minimum value, model training is completed. The trained random forest model is deployed to the actual production system, receiving in real time a comprehensive process feature vector composed of electroplating solution stability feature values ​​and temperature difference anomaly feature values, and outputting the corresponding process node anomaly score.

6. The method for full-process traceability of a circuit board production system according to claim 1, characterized in that, The process of identifying abnormal process nodes based on the analysis results specifically includes: Determine whether the abnormal score of a process node in the circuit board production process is greater than or equal to a preset threshold. If yes, mark it as an abnormal process node; otherwise, mark it as a normal process node.

7. The method for full-process traceability of a circuit board production system according to claim 1, characterized in that, The aforementioned triggering of corresponding quality early warning and traceability mechanisms enables refined control and traceability throughout the entire production process, specifically including: If the abnormal score of the process node output by the artificial intelligence analysis model is greater than or equal to the preset quality warning threshold, the system will automatically trigger a multi-level quality warning mechanism and send warning information to the operator or the central control system. The warning information includes the name of the process node where the abnormality occurred, the time of occurrence, and the current score value. The system will record the abnormal event in the database and generate a unique identifier for subsequent quality traceability.

Images