Printed circuit board manufacturing optimization method and system based on machine learning
By applying machine learning-based optimization methods in printed circuit board manufacturing, the problems of low accuracy, low efficiency and serious waste in traditional manufacturing processes are solved, and a more efficient and stable manufacturing process is achieved, reducing costs and improving competitiveness.
Patent Information
- Application Number
- CN202411912515.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-24
- Publication Date
- 2025-05-30
AI Technical Summary
There are problems such as design defects, low manufacturing accuracy, low production efficiency, large quality fluctuations, serious material waste and difficult process optimization in the manufacturing process of traditional printed circuit boards.
Using machine learning-based printed circuit board manufacturing optimization method, an etch path optimization model is established through reinforcement learning algorithms, and a drilling and plating parameter prediction model is established based on historical data and real-time monitoring data, and the manufacturing process parameters are optimized to improve production efficiency and quality.
It improves the design accuracy, manufacturing accuracy, production efficiency and quality stability of printed circuit boards, reduces material waste, reduces manufacturing costs, and enhances corporate competitiveness.
Smart Images
Figure CN120068758A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of printed circuit board manufacturing, and particularly relates to an optimization method and system for printed circuit board manufacturing based on machine learning. Background Art
[0002] In the traditional printed circuit board manufacturing process, there are many problems such as design defects, low manufacturing precision, low production efficiency, large quality fluctuations, serious material waste, and difficult process optimization. These problems seriously affect the development and competitiveness of the printed circuit board manufacturing industry. Therefore, there is an urgent need for an innovative optimization method for printed circuit board manufacturing based on machine learning to improve design accuracy, manufacturing precision, production efficiency, and quality stability, thereby reducing costs, reducing material waste, and realizing intelligent optimization of process parameters, bringing new development opportunities and competitive advantages to the printed circuit board manufacturing industry. Use machine learning algorithms to establish an etching path optimization model, optimize the etching path according to the collected data, improve production efficiency and stability, and reduce material waste at the same time. Establish a drilling and electroplating parameter prediction model based on historical data and real-time monitoring data, optimize parameter settings such as drilling position, hole diameter size, electroplating time, etc., and improve production efficiency and product quality. This optimization method for printed circuit board manufacturing based on machine learning aims to improve production efficiency, reduce costs, and improve product quality, thus solving the problems in the traditional printed circuit board manufacturing process. Summary of the Invention
[0003] An optimization method for printed circuit board manufacturing based on machine learning includes the following steps: S1. Design the circuit schematic diagram and the printed circuit board layout: Use CAD software to design the circuit schematic diagram and the printed circuit board layout, determine the layout, connection, and component installation positions of the circuit board, and optimize the layout design to reduce the line length and increase the wiring density; S2. Data collection and preprocessing: Collect monitoring data during the manufacturing process: real-time data such as printed quality images and ink usage, and perform data cleaning, standardization, and feature extraction on the data; S3. Make a printing stencil: Make a printing stencil according to the printed circuit board layout for the subsequent printing manufacturing process. First, cover the copper foil on the substrate, and then transfer the required circuit pattern to the copper foil through photolithography technology. According to the collected data, use the reinforcement learning algorithm to establish an etching path optimization model and integrate to obtain the etching path optimization formula , where F is the optimal path, L is the path length, E is the energy consumption, S is the path smoothness, C is the obstacle constraint (which can be represented by a penalty term), w 1 , w 2 , w 3 ,w 4 is the weight coefficient, which is used to balance the importance of different objectives. The copper foil without the covered pattern is removed by chemical etching method, leaving the required wire pattern. The fabricated inner layer board is laminated with the pre-fabricated glass fiber cloth, and heated and pressed to cure it, forming a printed circuit board with a multi-layer structure; S4. Drilling and electroplating: The printed circuit board is drilled using a drilling machine to prepare for subsequent component assembly and circuit connection. Then, a layer of copper is plated on the surface and hole walls of the printed circuit board through electroplating technology. Based on historical data and real-time monitoring data, a prediction model for parameters such as drilling position, hole diameter size, and electroplating time is established to help optimize the parameter settings of the drilling and electroplating processes, so as to improve production efficiency and stability; S5. Pattern coating: A protective pattern coating layer is covered on the surface of the printed circuit board, including treatment methods such as spraying tin, spraying nickel-gold, or spraying silver, which is used to protect the printed circuit and identification; S6. Assembly and testing: The components are soldered by wave soldering or hot air soldering iron, etc. to complete the assembly of the printed circuit board, and then circuit connectivity testing, function testing, and reliability testing are carried out to ensure the normal operation of the printed circuit board.
[0004] Further, an optimization method for manufacturing printed circuit boards based on machine learning, In step S3, according to the collected data, a reinforcement learning algorithm is used to establish an etching path optimization model. The copper foil without the covered pattern is removed by chemical etching method, leaving the required wire pattern. The specific steps are as follows: S31. Data collection: First, a large amount of relevant data needs to be collected, and the printed circuit board design file is parsed to obtain component layout and wiring information, and determine the parameters of the etching machine, including etching speed and liquid flow rate; S32. Data preprocessing: The collected data is cleaned and normalized to ensure data quality and consistency, and prepare for subsequent modeling; S33. Feature extraction: According to the problem characteristics and requirements, appropriate features are selected for extraction, such as the length, angle, and density of the etching path, as well as features related to etching quality; S34. Model training: The selected decision tree model is trained using the processed dataset. By continuously adjusting the model parameters and optimizing the algorithm, the model can accurately learn the rules of etching path optimization; S35. Model evaluation: The trained model is evaluated through cross-validation and other methods to check the generalization ability and accuracy of the model, ensure that the model can effectively predict and optimize the etching path, and integrate to obtain the etching path optimization formula , where F is the optimal path, L is the path length, E is the energy consumption, S is the path smoothness, and C is the obstacle constraint, represented by a penalty term. w 1 , w 2 , w 3 , w 4 are weight coefficients used to balance the importance of different objectives; S36. Path Optimization: Determine the optimal etching path using the model formula, considering factors such as etching sequence, distance, angle, etc., to reduce the etching time and the usage amount of chemical solution. The algorithm applies collision detection to prevent interference and damage between components during etching and ensure the etching quality. S37. Real-time Monitoring and Adjustment: During the etching process, real-time monitor parameters such as etching speed and depth, and according to the monitoring data, adjust the etching path in real time to achieve the best effect. S38. Model Application: Apply the trained machine learning model to the actual etching process, and continuously optimize the etching path according to the real-time feedback to achieve efficient and precise copper foil etching and leave the required wire pattern.
[0005] Furthermore, a machine learning-based optimization method for printed circuit board manufacturing In step S4, based on historical data and real-time monitoring data, establish a prediction model for parameters such as drilling position, hole diameter size, electroplating time, etc., to help optimize the parameter settings of the drilling and electroplating processes. The specific steps are as follows: S41. Data Preparation: Collect historical data and real-time monitoring data, including parameters such as drilling position, hole diameter size, electroplating time, etc., as well as corresponding circuit board design information and process parameters, and clean, process, and transform the data to ensure data quality and consistency. S42. Feature Engineering: Extract relevant features: coordinates of the drilling position, hole diameter size, circuit board design, and perform feature extraction, transformation, and combination to facilitate the learning and prediction of machine learning algorithms. S43. Data Partitioning: Divide the dataset into a training set and a test set, and usually use cross-validation to ensure the generalization ability of the model. S44. Model Training: Use the training set to train the regression model, and continuously adjust the model parameters to optimize the model performance. S45. Model Evaluation and Optimization: Use the test set to evaluate the trained model, examine the accuracy, generalization ability, and stability of the model, and adjust and optimize the model according to the evaluation results to improve the prediction accuracy and reliability. S46. Model Deployment and Application: Deploy the trained model to the actual production environment, monitor data in real time, and predict parameters such as drilling location, hole diameter size, electroplating time, etc., to help optimize the parameter settings of the drilling and electroplating processes and improve production efficiency and quality.
[0006] Furthermore, a printed circuit board manufacturing optimization system based on machine learning, which is used to implement a printed circuit board manufacturing optimization method based on machine learning; the printed circuit board manufacturing optimization system based on machine learning includes: a data acquisition module, a data preprocessing module, a feature extraction and selection module, a model training module, a real-time monitoring module, a manufacturing optimization module, and an anomaly detection and warning module; Among them, the data acquisition module: is responsible for collecting and organizing various data related to printed circuit board manufacturing, including design data, process parameters, production process data, etc.; The data preprocessing module: performs preprocessing operations such as cleaning, denoising, and normalization on the collected data to ensure the quality and consistency of the data; The feature extraction and selection module: extracts and selects features related to the manufacturing process from the original data to provide effective input for subsequent modeling and optimization; The model training module: uses machine learning algorithms to train optimization models for the printed circuit board manufacturing process, such as prediction models, optimization models, etc.; The real-time monitoring module: monitors the data in the production process in real time and transmits the monitored data to other modules for processing and analysis; The manufacturing optimization module: optimizes and adjusts manufacturing parameters according to the model training results and real-time monitoring data to improve production efficiency and product quality; The anomaly detection and warning module: monitors abnormal situations in the production process, performs anomaly detection and warning through machine learning algorithms, and processes problems in a timely manner to avoid production failures.
[0007] Advantages of the present invention: By optimizing the layout design, etching path, and drilling parameters, the line length is reduced, the wiring density is increased, and the parameter settings of the manufacturing process are optimized to improve production efficiency. Using machine learning algorithms for data processing and prediction model establishment can accurately control the key parameters in the manufacturing process and ensure the stable quality of printed circuit board manufacturing. Optimizing the manufacturing process parameters can reduce material waste, and improving production efficiency helps to reduce manufacturing costs, thereby enhancing the competitiveness of enterprises. Combining machine learning technology makes the entire printed circuit board manufacturing process more intelligent, improving the intelligent level and automation degree of the production line. Description of the Drawings
[0008] Figure 1Flowchart of an optimization method for printed circuit board manufacturing based on machine learning; Detailed implementation mode
[0009] An optimization method for printed circuit board manufacturing based on machine learning, the method process is as Figure 1 shown, including the following steps: S1. Design the circuit schematic diagram and printed circuit board layout: Use CAD software to design the circuit schematic diagram and printed circuit board layout, determine the layout, connection and component installation position of the circuit board, and optimize the layout design to reduce the line length and increase the wiring density; S2. Data collection and preprocessing: Collect monitoring data during the manufacturing process: real-time data such as printed quality images, ink usage, etc., and clean, standardize and extract features from the data; S3. Make a printing screen: Make a printing screen according to the printed circuit board layout for the subsequent printing manufacturing process. First, cover the copper foil on the substrate, and then transfer the required circuit pattern to the copper foil through photolithography technology. According to the collected data, use the reinforcement learning algorithm to establish an etching path optimization model and integrate it to obtain the etching path optimization formula , where F is the optimal path, L is the path length, E is the energy consumption, S is the path smoothness, C is the obstacle constraint (which can be represented by a penalty term), w 1 , w 2 , w 3 , w 4 are weight coefficients used to balance the importance of different objectives. Use the chemical etching method to remove the copper foil without the covered pattern, leaving the required wire pattern. Laminate the made inner layer board with the pre-made fiberglass cloth, heat and press it to cure, forming a multi-layer printed circuit board; S4. Drilling and electroplating: Use a drill press to drill the printed circuit board to prepare for subsequent component assembly and circuit connection. Then, deposit a layer of copper on the surface and hole walls of the printed circuit board through electroplating technology. Based on historical data and real-time monitoring data, establish a prediction model for parameters such as drilling position, hole diameter size, electroplating time, etc., to help optimize the parameter settings of the drilling and electroplating processes to improve production efficiency and stability; S5. Graphical coating: Cover a protective graphical coating layer on the surface of the printed circuit board, including treatment methods such as tin spraying, nickel-gold spraying or silver spraying, etc., for protecting the printed circuit and identification; S6. Assembly and testing: Solder the components by means of wave soldering or hot air soldering iron, etc., complete the assembly of the printed circuit board, and then conduct circuit connectivity testing, function testing and reliability testing to ensure the normal operation of the printed circuit board.
[0010] Furthermore, an optimization method for printed circuit board manufacturing based on machine learning In step S3, according to the collected data, use the reinforcement learning algorithm to establish an etching path optimization model, and use the chemical etching method to remove the copper foil without the covered pattern, leaving the required wire pattern. The specific steps are as follows: S31. Data collection: First, a large amount of relevant data needs to be collected, and the printed circuit board design file is parsed to obtain component layout and wiring information, and determine the parameters of the etching machine, including etching speed and liquid flow rate; S32. Data preprocessing: Clean and normalize the collected data to ensure data quality and consistency, and prepare for subsequent modeling; S33. Feature extraction: According to the characteristics and requirements of the problem, select appropriate features for extraction, such as the length, angle, density of the etching path, and features related to etching quality; S34. Model training: Use the processed dataset to train the selected decision tree model, and by continuously adjusting the model parameters and optimizing the algorithm, make the model accurately learn the rules of etching path optimization; S35. Model evaluation: Evaluate the trained model by means of cross-validation, etc., check the generalization ability and accuracy of the model, ensure that the model can effectively predict and optimize the etching path, and integrate to obtain the etching path optimization formula , where F is the optimal path, L is the path length, E is the energy consumption, S is the path smoothness, C is the obstacle constraint, represented by a penalty term, w 1 , w 2 , w 3 , w 4 are weight coefficients used to balance the importance of different objectives; S36. Path optimization: Use the model formula to determine the optimal etching path, consider factors such as etching order, distance, angle, etc., reduce the etching time and the usage amount of chemical solution, and apply collision detection in the algorithm to prevent interference and damage between components during etching and ensure etching quality; S37. Real-time monitoring and adjustment: During the etching process, real-time monitor parameters such as etching speed and depth, and according to the monitoring data, real-time adjust the etching path to achieve the best effect; S38. Model Application: Apply the trained machine learning model to the actual etching process, and continuously optimize the etching path according to real-time feedback to achieve efficient and precise copper foil etching and leave the required wire pattern.
[0011] Furthermore, an optimization method for printed circuit board manufacturing based on machine learning, In step S4, based on historical data and real-time monitoring data, establish a prediction model for drilling position, hole diameter size, and electroplating time parameters to help optimize the parameter settings of the drilling and electroplating processes. The specific steps are as follows: S41. Data Preparation: Collect historical data and real-time monitoring data, including parameters such as drilling position, hole diameter size, electroplating time, and corresponding circuit board design information and process parameters. Clean, process, and transform the data to ensure data quality and consistency; S42. Feature Engineering: Extract relevant features: coordinates of the drilling position, hole diameter size, and circuit board design, and perform feature extraction, transformation, and combination to facilitate the learning and prediction of machine learning algorithms; S43. Data Partitioning: Divide the data set into a training set and a test set, and usually use cross-validation to ensure the generalization ability of the model; S44. Model Training: Use the training set to train the regression model and continuously adjust the model parameters to optimize the model performance; S45. Model Evaluation and Optimization: Use the test set to evaluate the trained model, examine the accuracy, generalization ability, and stability of the model, and adjust and optimize the model according to the evaluation results to improve the prediction accuracy and reliability; S46. Model Deployment and Application: Deploy the trained model to the actual production environment, monitor the data in real time, and predict parameters such as drilling position, hole diameter size, and electroplating time to help optimize the parameter settings of the drilling and electroplating processes and improve production efficiency and quality.
[0012] Furthermore, an optimization system for printed circuit board manufacturing based on machine learning, the optimization system for printed circuit board manufacturing based on machine learning is used to implement an optimization method for printed circuit board manufacturing based on machine learning; the optimization system for printed circuit board manufacturing based on machine learning includes: a data acquisition module, a data preprocessing module, a feature extraction and selection module, a model training module, a real-time monitoring module, a manufacturing optimization module, and an anomaly detection and warning module; Among them, the data acquisition module: is responsible for collecting and organizing various data related to printed circuit board manufacturing, including design data, process parameters, production process data, etc.; The data preprocessing module: performs preprocessing operations such as cleaning, denoising, and normalization on the collected data to ensure data quality and consistency; Feature extraction and selection module: Extract and select features related to the manufacturing process from the original data to provide effective input for subsequent modeling and optimization; Model training module: Use machine learning algorithms to train optimization models for the printed circuit board manufacturing process, such as prediction models, optimization models, etc.; Real-time monitoring module: Real-time monitor the data in the production process and transmit the monitored data to other modules for processing and analysis; Manufacturing optimization module: Optimize and adjust manufacturing parameters according to the model training results and real-time monitoring data to improve production efficiency and product quality; Abnormality detection and warning module: Monitor abnormal situations in the production process and perform abnormality detection and warning through machine learning algorithms to promptly handle problems to avoid production failures.
Claims
1. A printed circuit board manufacturing optimization method based on machine learning, characterized in that: The following steps are involved: S1. Design circuit schematics and printed circuit board layouts: Use CAD software to design circuit schematics and printed circuit board layouts, determine the layout, connections and component installation locations of the circuit boards, and optimize the layout design to reduce line length and increase wiring density; S2. Data collection and preprocessing: Collect monitoring data during the manufacturing process: printing quality images, real-time data on ink usage, and clean, standardize and extract features from the data; S3. Make a printing screen: Make a printing screen according to the printed circuit board layout for the subsequent printing manufacturing process. First, cover the copper foil on the substrate, and then transfer the required circuit pattern to the copper foil through photolithography technology. Based on the collected data, use the reinforcement learning algorithm to establish an etching path optimization model and integrate it to obtain the etching path optimization formula , where F is the optimal path, L is the path length, E is the energy consumption, S is the path smoothness, and C is the obstacle constraint, expressed as a penalty term. w 1, w 2, w 3, w 4 is the weight coefficient, which is used to balance the importance of different goals. The copper foil not covered with the pattern is removed by chemical etching method, leaving the required conductor pattern. The prepared inner layer board is laminated with the pre-made glass fiber cloth, and heated and pressed to solidify it to form a multi-layer printed circuit board. S4. Drilling and electroplating: Use a drilling machine to drill holes in the printed circuit board to prepare for the subsequent assembly of components and circuit connections. Then use electroplating technology to plate a layer of copper on the surface of the printed circuit board and the hole wall. Based on historical data and real-time monitoring data, a prediction model for drilling position, hole size, and electroplating time parameters is established to help optimize the parameter settings of the drilling and electroplating processes to improve production efficiency and stability. S5. Graphic coating: A protective graphic coating is applied to the surface of the printed circuit board, including tin plating, nickel-gold plating, and silver plating, to protect the printed circuits and logos; S6. Assembly and testing: Solder the components by wave soldering and hot air soldering iron to complete the assembly of printed circuit boards, and conduct circuit connectivity tests, functional tests and reliability tests to ensure that the printed circuit boards work properly.
2. The method for optimizing printed circuit board manufacturing based on machine learning according to claim 1, characterized in that: In step S3, based on the collected data, a reinforcement learning algorithm is used to establish an etching path optimization model, and a chemical etching method is used to remove the copper foil not covered with the pattern, leaving the desired conductor pattern. The specific steps are: S31. Data collection: First, a large amount of relevant data needs to be collected and the printed circuit board design file needs to be parsed to obtain component layout and wiring information and determine the parameters of the etching machine, including etching speed and liquid flow rate; S32. Data preprocessing: Clean and normalize the collected data to ensure data quality and consistency, and prepare for subsequent modeling; S33. Feature extraction: According to the characteristics and requirements of the problem, select appropriate features for extraction, such as the length, angle, density of the etching path, and features related to etching quality; S34. Model training: Use the processed data set to train the selected decision tree model, and continuously adjust the model parameters and optimization algorithm so that the model can accurately learn the rules of etching path optimization; S35. Model evaluation: Evaluate the trained model through cross-validation and other methods to check the generalization ability and accuracy of the model, ensure that the model effectively predicts and optimizes the etching path, and integrate the etching path optimization formula , where F is the optimal path, L is the path length, E is the energy consumption, S is the path smoothness, and C is the obstacle constraint, expressed as a penalty term. w 1, w 2, w 3, w 4 is the weight coefficient, which is used to balance the importance of different goals; S36. Path optimization: Use the model formula to determine the optimal etching path, consider the etching sequence, distance, and angle factors, reduce the etching time and the amount of chemical solution used, and use the algorithm to apply collision detection to prevent interference and damage between components during etching to ensure etching quality; S37. Real-time monitoring and adjustment: During the etching process, the etching speed, depth and other parameters are monitored in real time, and the etching path is adjusted in real time according to the monitoring data to achieve the best effect; S38. Model application: Apply the trained machine learning model to the actual etching process, and continuously optimize the etching path based on real-time feedback to achieve efficient and accurate copper foil etching, leaving the desired conductor pattern.
3. The method for optimizing printed circuit board manufacturing based on machine learning according to claim 1, characterized in that: In step S4, based on historical data and real-time monitoring data, a prediction model of drilling position, hole size, and electroplating time parameters is established to help optimize the parameter settings of the drilling and electroplating processes. The specific steps are: S41. Data preparation: Collect historical data and real-time monitoring data, including drilling location, aperture size, electroplating time parameters and corresponding circuit board design information and process parameters, and clean, process and convert the data to ensure data quality and consistency; S42. Feature Engineering: Extract relevant features: coordinates of drilling locations, hole size, circuit board design, perform feature extraction, transformation and combination to facilitate learning and prediction of machine learning algorithms; S43. Data partitioning: Divide the data set into training set and test set, and use cross-validation to ensure the generalization ability of the model; S44. Model training: Use the training set to train the regression model and continuously adjust the model parameters to optimize the model performance; S45. Model evaluation and optimization: Use the test set to evaluate the trained model, examine the accuracy, generalization ability and stability of the model, adjust and optimize the model according to the evaluation results, and improve the prediction accuracy and reliability; S46. Model deployment and application: Deploy the trained model to the actual production environment, monitor the data in real time and predict the drilling position, hole size, and electroplating time parameters to help optimize the parameter settings of the drilling and electroplating processes and improve production efficiency and quality.
4. A printed circuit board manufacturing optimization system based on machine learning, characterized in that: The printed circuit board manufacturing optimization system based on machine learning is used to implement the printed circuit board manufacturing optimization method based on machine learning as described in any one of claims 1 to 3; The printed circuit board manufacturing optimization system based on machine learning includes: a data acquisition module, a data preprocessing module, a feature extraction and selection module, a model training module, a real-time monitoring module, a manufacturing optimization module, and an anomaly detection and early warning module; The data acquisition module is responsible for collecting and sorting various data related to printed circuit board manufacturing, including design data, process parameters, and production process data; Data preprocessing module: performs cleaning, denoising, and normalization preprocessing operations on the collected data to ensure data quality and consistency; Feature extraction and selection module: extracts and selects features related to the manufacturing process from raw data to provide effective input for subsequent modeling and optimization; Model training module: Use machine learning algorithms to train optimization models for the printed circuit board manufacturing process; Real-time monitoring module: monitors the data in the production process in real time and transmits the monitored data to other modules for processing and analysis; Manufacturing optimization module: optimizes and adjusts manufacturing parameters based on model training results and real-time monitoring data to improve production efficiency and product quality; Anomaly detection and early warning module: monitors abnormal situations in the production process, and performs anomaly detection and early warning through machine learning algorithms, and handles problems in a timely manner to avoid production failures.