A method and system for pad printing press process optimization based on automatic machine learning
By optimizing the pad printing bonding process through automated machine learning, the problems of reliance on human experience and poor model adaptability in traditional methods have been solved. This enables real-time adjustment and continuous optimization of process parameters, thereby improving production efficiency and product quality.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- LIJING PRECISION TECHNOLOGY (ZHEJIANG) CO LTD
- Filing Date
- 2026-03-04
- Publication Date
- 2026-06-05
AI Technical Summary
Traditional pad printing process parameters rely on manual experience for adjustment, resulting in unstable product quality that is difficult to meet the needs of large-scale production. Furthermore, traditional machine learning models cannot be optimized in real time, have poor adaptability, and cannot cope with complex and ever-changing production scenarios.
By employing automated machine learning methods, historical production data is collected and processed to construct an automated machine learning model framework. Process parameters are adjusted in real time to achieve iterative updates of the model and optimize the pad printing bonding process.
It achieves efficient and precise optimization of pad printing and lamination processes, reduces manual intervention, improves the accuracy and adaptability of process parameter prediction, adapts to dynamic changes in production conditions, and ensures stable product quality.
Smart Images
Figure CN122155016A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of pad printing and lamination technology, specifically to a method and system for optimizing pad printing and lamination processes based on automated machine learning. Background Technology
[0002] Pad printing and lamination are critical processes in industrial production, and the rationality of process parameters directly determines the stability of product quality. Traditional pad printing and lamination process parameter adjustments largely rely on operator experience and judgment, determining key parameters such as pad pressure, contact time, and lamination temperature through manual trial and error, lacking a scientific and systematic optimization basis. This approach not only consumes significant manpower and time costs but also leads to fluctuations in process parameters due to differences in operator experience, resulting in inconsistent product quality, appearance defects, and insufficient adhesion, making it difficult to meet the high-quality requirements of large-scale production. Furthermore, pad printing and lamination processes have an implicit coupling relationship; manual adjustments cannot accurately capture the parameter correlation between the two processes, failing to achieve synergistic process optimization.
[0003] With the application of machine learning technology in the industrial field, some production scenarios have begun to try to optimize process parameters through machine learning models. However, these methods require professional technicians to manually build model architectures, select algorithm models, and debug parameters, which places high demands on technical expertise. The manual modeling process is time-consuming, and the model has poor adaptability. When production raw materials or equipment conditions change, the model cannot be quickly adjusted and needs to be remodeled and debugged. In addition, traditional machine learning models are mostly static models after deployment, which cannot be dynamically iterated and optimized based on real-time production data. They are difficult to cope with complex and ever-changing production scenarios and cannot fully realize the value of machine learning technology in process optimization, thus limiting the overall optimization effect. Summary of the Invention
[0004] To address the aforementioned technical shortcomings, the purpose of this invention is to provide a method and system for optimizing pad printing and bonding processes based on automated machine learning, which comprehensively completes the entire process optimization and meets the high-efficiency and precise optimization requirements of pad printing and bonding processes.
[0005] To achieve the above objectives, the present invention adopts the following technical solution:
[0006] An optimization method for pad printing bonding process based on automated machine learning includes the following steps:
[0007] S1. Collect historical production data and corresponding product quality inspection data of pad printing and lamination processes, and clean and normalize the collected data.
[0008] S2. Based on the processed data, construct an automatic machine learning model framework and select an algorithm component combination that is suitable for process optimization;
[0009] S3. Input the processed data into the automatic machine learning model framework for training. Determine the optimal algorithm components and parameter configurations through automatic parameter tuning of the model to generate an optimization strategy for pad printing bonding process parameters.
[0010] S4. Deploy the trained automatic machine learning model to the production system and collect real-time operating data and product quality data of the pad printing laminating equipment;
[0011] S5. Input the real-time collected data into the model, and the model will output the corresponding pad printing and lamination process optimization parameters.
[0012] S6. Adjust the operating status of the pad printing and laminating equipment according to the output optimization parameters, and continuously monitor the product quality data after the process adjustment; supplement the newly collected data into the database and drive the automatic machine learning model to perform iterative updates.
[0013] Preferably, in step S1, data on pad printing pressure, contact time, and ink viscosity are collected, as are data on pressing temperature, pressing pressure, and holding time in the pressing process. Simultaneously, data on appearance defects, fit, and adhesion of the corresponding batch of products are collected.
[0014] Preferably, in step S1, missing values and outliers in the data are removed, and the min-max standardization method is used to map process data and quality data of different dimensions to the same numerical range to form a standardized dataset.
[0015] Preferably, in step S2, a model framework is constructed that includes a data feature engineering component, an algorithm selection component, and a parameter tuning component. Decision tree algorithm, random forest algorithm, and gradient boosting tree algorithm are selected to form an algorithm component pool for automatic selection and combination by the model.
[0016] Preferably, in step S3, the data is divided into a training set, a validation set, and a test set according to a preset ratio. The training set is used to drive the model to select algorithm components and optimize parameters. The validation set is used to verify the model's fitting effect. The test set is used to evaluate the model's process parameter prediction accuracy.
[0017] Preferably, in step S4, the operating parameters of the pad printing pressing equipment are collected in real time by the equipment sensors, and product quality data are collected by the visual inspection equipment and the mechanical inspection equipment. The collected data is transmitted to the model deployment terminal at preset time intervals.
[0018] Preferably, in step S6, a model iteration trigger threshold is set. When the newly collected quality data deviates from the model's prediction range, the model iteration process is started, the new data is integrated into the original dataset to retrain the model, and the optimal algorithm components and parameter configurations are updated.
[0019] An automated machine learning-based pad printing and laminating process optimization system includes a historical data acquisition and processing module, a model building and training module, a model deployment module, a real-time data acquisition module, a process parameter output module, a model iteration module, and an equipment control module. The historical data acquisition and processing module collects and processes historical production and quality data of the pad printing and laminating process. The model building and training module constructs an automated machine learning model framework and completes model training. The model deployment module deploys the trained model to the production system. The real-time data acquisition module collects equipment operation data and product quality data. The process parameter output module outputs process optimization parameters. The model iteration module drives the model to iteratively update. The equipment control module adjusts the operating status of the pad printing and laminating equipment according to the optimization parameters. All modules interact bidirectionally via a data bus.
[0020] Preferably, the historical data acquisition and processing module includes a data classification unit, a data verification unit, and a normalization unit; the data classification unit is used to distinguish between pad printing process data, pressing process data, and product quality data; the data verification unit is used to identify and remove missing values and outliers; and the normalization unit is used to map data of different dimensions to the same numerical range.
[0021] The real-time data acquisition module includes a sensor data acquisition unit, a quality inspection data acquisition unit, and a data transmission unit. The sensor data acquisition unit is used to collect equipment operating parameters, the quality inspection data acquisition unit is used to collect product appearance and performance inspection data, and the data transmission unit is used to transmit the collected data at preset time intervals.
[0022] Preferably, the model deployment module includes an edge computing deployment unit and a model version management unit; the edge computing deployment unit is used to deploy the trained model to the edge terminal of the production site to realize localized data computing; the model version management unit is used to record the version information of model iteration updates;
[0023] The equipment control module includes a parameter adjustment unit, an operation status monitoring unit, and a closed-loop feedback unit. The parameter adjustment unit is used to adjust the operating parameters of the pad printing pressing equipment, the operation status monitoring unit is used to monitor the equipment operation status in real time, and the closed-loop feedback unit is used to feed back the equipment operation status and product quality data to the real-time data acquisition module.
[0024] Compared with the prior art, the beneficial effects of the present invention are as follows:
[0025] The historical data acquisition and processing module is equipped with data classification, verification, and normalization processing units, which can accurately distinguish pad printing process data, pressing process data, and product quality data, effectively remove missing values and outliers, and map data of different dimensions to the same numerical range, providing high-quality data support for model training and avoiding invalid data from interfering with model accuracy.
[0026] By constructing an automated machine learning model framework and selecting multiple types of algorithms to form a component pool, the automatic selection of algorithm components and automatic parameter tuning can be achieved. This eliminates the need for manual modeling and debugging by professionals, significantly reducing the technical application threshold. At the same time, it ensures that the model can select the optimal algorithm combination that is suitable for the characteristics of pad printing and pressing processes, thereby improving the accuracy of the model's process parameter prediction.
[0027] The model deployment module is equipped with an edge computing deployment unit and a model version management unit. It can deploy the trained model to the edge terminal of the production site to realize localized data computing and improve the real-time performance of parameter output. The version management unit can record the model iteration process in a complete manner, making it easy to trace the model optimization effect at different stages.
[0028] The equipment control module is equipped with parameter adjustment, operation status monitoring, and closed-loop feedback units. It can accurately adjust the equipment operation status based on the optimized parameters output by the model, monitor the equipment operation in real time, and feed back the equipment status and product quality data to the data acquisition module, forming a closed-loop control for process optimization and timely correcting parameter deviations.
[0029] The model iteration module drives continuous model updates, integrating new production data into the original dataset to retrain the model. This allows the model to adapt to dynamic changes in production conditions and continuously optimize process parameters. All system modules work collaboratively via a data bus, enabling automated operation of the entire process optimization process. This reduces manual intervention, improves optimization efficiency, and caters to diverse needs for both individual and collaborative optimization of pad printing lamination processes. Attached Figure Description
[0030] Figure 1 This is a flowchart of the method of the present invention;
[0031] Figure 2 This is a structural block diagram of the system of the present invention. Detailed Implementation
[0032] The invention will now be further described with reference to the accompanying drawings.
[0033] like Figure 1 As shown, a method for optimizing pad printing bonding process based on automated machine learning includes the following steps:
[0034] S1. Collect historical production data and corresponding product quality inspection data for pad printing lamination process:
[0035] Collect data on pad printing process, including pad pressure, contact time, and ink viscosity; collect data on pressing process, including pressing temperature, pressing pressure, and holding time; and simultaneously collect data on appearance defects, fit, and adhesion of corresponding batches of products.
[0036] The collected data was cleaned and normalized.
[0037] Missing and outlier values in the data are removed, and the min-max standardization method is used to map process data and quality data of different dimensions to the same numerical range to form a standardized dataset.
[0038] S2. Based on the processed data, construct an automated machine learning model framework and select algorithm components that are suitable for process optimization:
[0039] A model framework is constructed that includes data feature engineering components, algorithm selection components, and parameter tuning components. Decision tree algorithm, random forest algorithm, and gradient boosting tree algorithm are selected to form an algorithm component pool for automatic selection and combination by the model.
[0040] S3. Input the processed data into the automatic machine learning model framework for training. The model automatically tunes parameters to determine the optimal algorithm components and parameter configurations, generating an optimization strategy for pad printing bonding process parameters.
[0041] The standardized dataset is divided into training set, validation set and test set according to a preset ratio. The training set is used to drive the model to select algorithm components and optimize parameters. The validation set is used to verify the model fitting effect. The test set is used to evaluate the model's process parameter prediction accuracy.
[0042] S4. Deploy the trained automated machine learning model to the production system to collect real-time operational data and product quality data from the pad printing laminating equipment:
[0043] The operating parameters of the pad printing and pressing equipment are collected in real time by the equipment sensors, and product quality data are collected by the visual inspection equipment and mechanical inspection equipment. The collected data is transmitted to the model deployment terminal at preset time intervals.
[0044] S5. Input the real-time collected data into the model, and the model will output the corresponding pad printing and lamination process optimization parameters.
[0045] S6. Adjust the operating status of the pad printing laminating equipment according to the output optimized parameters, and continuously monitor the product quality data after the process adjustment; supplement the newly collected data into the database to drive the automatic machine learning model to perform iterative updates:
[0046] Set a threshold to trigger model iteration. When newly collected quality data deviates from the model's prediction range, start the model iteration process, integrate the new data into the original dataset, retrain the model, and update the optimal algorithm components and parameter configurations.
[0047] like Figure 2 As shown, a pad printing and pressing process optimization system based on automated machine learning includes a historical data acquisition and processing module, a model building and training module, a model deployment module, a real-time data acquisition module, a process parameter output module, a model iteration module, and an equipment control module. The historical data acquisition and processing module is used to collect and process historical production and quality data of the pad printing and pressing process. The model building and training module is used to build an automated machine learning model framework and complete model training. The model deployment module is used to deploy the trained model to the production system. The real-time data acquisition module is used to collect equipment operation data and product quality data. The process parameter output module is used to output process optimization parameters. The model iteration module is used to drive the model to perform iterative updates. The equipment control module is used to adjust the operating status of the pad printing and pressing equipment according to the optimization parameters. All modules achieve bidirectional data interaction through a data bus.
[0048] Furthermore, the historical data acquisition and processing module includes a data classification unit, a data verification unit, and a normalization unit; the data classification unit is used to distinguish between pad printing process data, lamination process data, and product quality data; the data verification unit is used to identify and remove missing values and outliers; and the normalization unit is used to map data of different dimensions to the same numerical range.
[0049] The real-time data acquisition module includes a sensor data acquisition unit, a quality inspection data acquisition unit, and a data transmission unit. The sensor data acquisition unit is used to collect equipment operating parameters, the quality inspection data acquisition unit is used to collect product appearance and performance inspection data, and the data transmission unit is used to transmit the collected data at preset time intervals.
[0050] Furthermore, the model deployment module includes an edge computing deployment unit and a model version management unit; the edge computing deployment unit is used to deploy the trained model to the edge terminal in the production site to realize localized data computing; the model version management unit is used to record the version information of model iteration updates;
[0051] The equipment control module includes a parameter adjustment unit, an operation status monitoring unit, and a closed-loop feedback unit. The parameter adjustment unit is used to adjust the operating parameters of the pad printing pressing equipment, the operation status monitoring unit is used to monitor the equipment operation status in real time, and the closed-loop feedback unit is used to feed back the equipment operation status and product quality data to the real-time data acquisition module.
[0052] Example 1
[0053] This embodiment is applied to the separate optimization of the pad printing process. The corresponding method is executed using the aforementioned pad printing bonding process optimization system based on automatic machine learning. The specific process is as follows:
[0054] The historical data acquisition and processing module collects historical production data and corresponding product quality inspection data for pad printing. Production data includes information on pad pressure, contact time, and ink viscosity, while quality data includes information on product appearance defects and pattern adhesion. The data classification unit categorizes and organizes the collected data, the data verification unit identifies and removes missing and outlier values, and the normalization unit uses the min-max standardization method to map data of different dimensions to the same numerical range, forming a standardized dataset.
[0055] An automated machine learning model framework is constructed based on a standardized dataset, selecting decision tree, random forest, and gradient boosting tree algorithms to form an algorithm component pool. The standardized dataset is divided into training, validation, and test sets. The training set is used to drive the model in selecting algorithm components and optimizing parameters. The validation set is used to verify the model's fitting effect, and the test set is used to evaluate the model's process parameter prediction accuracy. The optimal algorithm components and parameter configurations are determined, and an optimization strategy for pad printing process parameters is generated.
[0056] The edge computing deployment unit of the model deployment module deploys the trained model to the edge terminal on the production site, and the model version management unit records the current model version information. The real-time data acquisition module is activated. The sensor data acquisition unit collects the operating parameters of the pad printing equipment in real time through the equipment's sensors, and the quality inspection data acquisition unit collects product quality data through visual inspection and mechanical inspection equipment. The data transmission unit transmits the collected data to the model deployment terminal at preset time intervals.
[0057] The real-time collected data is input into the model, and the process parameter output module outputs the corresponding pad printing process optimization parameters. The parameter adjustment unit of the equipment control module adjusts the operating parameters of the pad printing equipment, such as the pad pressure and contact time, according to the optimization parameters. The operation status monitoring unit monitors the equipment operation status in real time, and the closed-loop feedback unit feeds back the equipment operation status and product quality data to the real-time data acquisition module.
[0058] Continuously monitor product quality data after process adjustments. When newly collected quality data deviates from the model's prediction range, the model iteration module initiates the model iteration process, integrates the new data into the original dataset, retrains the model, updates the optimal algorithm components and parameter configurations, and completes the continuous optimization of the pad printing process.
[0059] Example 2
[0060] This embodiment applies to the collaborative optimization of the pad printing and lamination process. The corresponding method is executed using the aforementioned pad printing and lamination process optimization system based on automatic machine learning. The specific process is as follows:
[0061] The historical data acquisition and processing module synchronously collects historical production data and corresponding product quality inspection data for pad printing and lamination processes. Pad printing production data includes data related to pad pressure and contact time; lamination production data includes data related to lamination temperature, lamination pressure, and holding time; and quality data includes data related to product pattern adhesion and structural strength. The data classification unit categorizes and labels the three types of data, the data verification unit removes outlier data, and the normalization unit standardizes the data, forming a standardized dataset that includes the coupling relationship between the two processes.
[0062] An automated machine learning model framework is constructed based on a standardized dataset. Decision tree, random forest, and gradient boosting tree algorithms are selected as the algorithm component pool to enhance the model's ability to learn the coupling relationship of the pad printing and bonding process. The standardized dataset is divided into training, validation, and test sets. The training set is used to select algorithm components and optimize parameters. The validation set verifies the model's adaptability to the combined process. The test set evaluates the model's co-optimization accuracy and generates optimization strategies for the pad printing and bonding process parameters.
[0063] The trained model is deployed to the production system via the model deployment module, and the model version management unit records the model version information. The real-time data acquisition module is activated to synchronously collect operating parameters of the pad printing and pressing equipment, as well as comprehensive product quality data, and transmits them to the model deployment terminal at preset time intervals.
[0064] Real-time data is input into the model, and the process parameter output module outputs the collaborative optimization parameters for the pad printing and pressing combined process. The parameter adjustment unit of the equipment control module synchronously adjusts the operating parameters of the pad printing equipment and the pressing equipment, the operation status monitoring unit monitors the operating status of both types of equipment simultaneously, and the closed-loop feedback unit feeds back the equipment operating status and product quality data to the real-time data acquisition module.
[0065] Continuously monitor product quality data after the joint process adjustment. When the newly collected data reaches the model iteration trigger threshold, start the model iteration process, integrate the newly added joint process data into the original dataset to retrain the model, update the optimal algorithm components and parameter configuration, and realize dynamic collaborative optimization of the pad printing and lamination joint process.
Claims
1. A method for optimizing pad printing bonding process based on automated machine learning, characterized in that, Includes the following steps: S1. Collect historical production data and corresponding product quality inspection data of pad printing and lamination processes, and clean and normalize the collected data. S2. Based on the processed data, construct an automatic machine learning model framework and select an algorithm component combination that is suitable for process optimization; S3. Input the processed data into the automatic machine learning model framework for training. Determine the optimal algorithm components and parameter configurations through automatic parameter tuning of the model to generate an optimization strategy for pad printing bonding process parameters. S4. Deploy the trained automatic machine learning model to the production system and collect real-time operating data and product quality data of the pad printing laminating equipment; S5. Input the real-time collected data into the model, and the model will output the corresponding pad printing and lamination process optimization parameters. S6. Adjust the operating status of the pad printing and laminating equipment according to the output optimization parameters, and continuously monitor the product quality data after the process adjustment; supplement the newly collected data into the database and drive the automatic machine learning model to perform iterative updates.
2. The method for optimizing pad printing bonding process based on automated machine learning as described in claim 1, characterized in that, In step S1, data on pad printing pressure, contact time, and ink viscosity are collected; data on pressing temperature, pressing pressure, and holding time are collected for the pressing process; and data on appearance defects, fit, and adhesion of the corresponding batch of products are collected simultaneously.
3. The method for optimizing pad printing bonding process based on automated machine learning as described in claim 2, characterized in that, In step S1, missing and outlier values in the data are removed, and the min-max standardization method is used to map process data and quality data of different dimensions to the same numerical range to form a standardized dataset.
4. The method for optimizing pad printing bonding process based on automated machine learning as described in claim 1, characterized in that, In step S2, a model framework is constructed that includes data feature engineering components, algorithm selection components, and parameter tuning components. Decision tree algorithm, random forest algorithm, and gradient boosting tree algorithm are selected to form an algorithm component pool for automatic selection and combination by the model.
5. The method for optimizing pad printing bonding process based on automated machine learning as described in claim 1, characterized in that, In step S3, the data is divided into a training set, a validation set, and a test set according to a preset ratio. The training set is used to drive the model to select algorithm components and optimize parameters. The validation set is used to verify the model's fitting effect, and the test set is used to evaluate the model's process parameter prediction accuracy.
6. The method for optimizing pad printing bonding process based on automated machine learning as described in claim 1, characterized in that, In step S4, the operating parameters of the pad printing and pressing equipment are collected in real time through the equipment sensors, and product quality data are collected through visual inspection equipment and mechanical inspection equipment. The collected data are transmitted to the model deployment terminal at preset time intervals.
7. The method for optimizing pad printing bonding process based on automated machine learning as described in claim 1, characterized in that, In step S6, a model iteration trigger threshold is set. When the newly collected quality data deviates from the model's prediction range, the model iteration process is started, the new data is integrated into the original dataset to retrain the model, and the optimal algorithm components and parameter configurations are updated.
8. A pad printing bonding process optimization system based on automated machine learning, used to implement the method described in any one of claims 1 to 7, characterized in that, It includes a historical data acquisition and processing module, a model building and training module, a model deployment module, a real-time data acquisition module, a process parameter output module, a model iteration module, and an equipment control module; the historical data acquisition and processing module is used to collect and process historical production data and quality data of the pad printing and laminating process; The model building and training module is used to build an automated machine learning model framework and complete model training; the model deployment module is used to deploy the trained model to the production system; and the real-time data acquisition module is used to collect equipment operation data and product quality data. The process parameter output module is used to output process optimization parameters; The model iteration module is used to drive the model to perform iterative updates; the equipment control module is used to adjust the operating status of the pad printing and pressing equipment according to the optimization parameters. The modules achieve bidirectional data interaction through the data bus.
9. The pad printing lamination process optimization system based on automated machine learning as described in claim 8, characterized in that, The historical data acquisition and processing module includes a data classification unit, a data verification unit, and a normalization unit. The data classification unit is used to distinguish between pad printing process data, lamination process data, and product quality data. The data verification unit is used to identify and remove missing values and outliers. The normalization unit is used to map data of different dimensions to the same numerical range. The real-time data acquisition module includes a sensor data acquisition unit, a quality inspection data acquisition unit, and a data transmission unit. The sensor data acquisition unit is used to collect equipment operating parameters, the quality inspection data acquisition unit is used to collect product appearance and performance inspection data, and the data transmission unit is used to transmit the collected data at preset time intervals.
10. The pad printing lamination process optimization system based on automated machine learning as described in claim 8, characterized in that, The model deployment module includes an edge computing deployment unit and a model version management unit; The edge computing deployment unit is used to deploy the trained model to the edge terminal in the production site to realize localized data computing; the model version management unit is used to record the version information of model iteration updates; The equipment control module includes a parameter adjustment unit, an operation status monitoring unit, and a closed-loop feedback unit. The parameter adjustment unit is used to adjust the operating parameters of the pad printing pressing equipment, the operation status monitoring unit is used to monitor the equipment operation status in real time, and the closed-loop feedback unit is used to feed back the equipment operation status and product quality data to the real-time data acquisition module.