Method and device for training and predicting PCB electroplating parameter prediction model
By dividing the data set and training models multiple times, the stable prediction model is selected and weighted, the overfitting and underfitting problems of PCB electroplating parameter prediction model in the prior art are solved, and the reliability of the prediction results is improved. Productivity.
Patent Information
- Application Number
- CN202411041536.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-31
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2044-07-31
AI Technical Summary
The performance of existing PCB plating parameter prediction models in the training set and test set is unstable, and they are prone to overfitting or underfitting, resulting in inaccurate prediction results, which in turn affects production quality and cost.
By dividing the data set and training model multiple times, using the random number generator seed to filter out the training set and test set that meet the threshold, construct multiple prediction models, and obtain the final prediction value through weighting, ensuring that the model is on different data sets Consistent performance.
Overfitting and underfitting problems are avoided, the reliability and accuracy of the prediction results are improved, and production quality and efficiency are improved.
Smart Images

Figure CN119092002A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of PCB production, and more specifically to a training and prediction method and device for a PCB electroplating parameter prediction model. Background Art
[0002] PCB electroplating parameters include current density, electroplating time, and the accuracy of electroplating parameter settings, which directly determine the product quality yield, production efficiency, and production cost. Set the electroplating time according to the upper and lower limits of the VCP line current density, and estimate the current density according to the PCB electroplating copper thickness requirements. At present, the industry explores the influencing factors that affect the current density, mainly PCB product characteristics, such as board thickness, aperture, copper thickness requirements, etc., and establishes mathematical models for prediction and optimization. The established mathematical models mainly include decision tree models and multivariate linear regression models. During the model training process, specify a seed of a random number generator to divide the training set and the test set, perform model training, and make predictions based on the generated model. The model obtained by dividing the data according to the specified seed will have a higher predicted yield for the training set and a lower predicted yield for the test set, or a lower predicted yield for the training set and a higher yield for the test set.
[0003] The above training methods have the following shortcomings: the training model is overfitted or underfitted, and the prediction model effect is unstable, resulting in the copper thickness of the PCB produced according to the predicted electroplating parameters being too thick or lower than the required value, thereby increasing production costs and reducing production efficiency and quality yield. Summary of the invention
[0004] The purpose of the present invention is to overcome the deficiencies of the prior art and to provide a training and prediction method and device for a PCB electroplating parameter prediction model.
[0005] To achieve the above object, the present invention adopts the following technical solutions:
[0006] In a first aspect, the present invention provides a training and prediction method for a PCB electroplating parameter prediction model, comprising:
[0007] Get the original production data of PCB;
[0008] Preprocess the original production data of PCB to obtain a data set;
[0009] Constructing a first random number generator seed database and setting a range of the first random number generator seed;
[0010] Dividing the data set into a data set and a test set according to the set range of the first random number generator seed to generate a first training and test set database;
[0011] Performing model training on the training set in the first training and test set database, and saving the trained model to obtain a first prediction model database;
[0012] Output predicted values for the training set and the data set using each prediction model in the first prediction model database, and compare the obtained predicted values with the actual values to obtain evaluation index values for the training set and the test set;
[0013] According to the electroplating parameter control error requirements, set the thresholds of the evaluation index values of the training set and the test set;
[0014] Screen out the evaluation index values of the training set and the test set that meet the threshold, and use the first random number generator seeds corresponding to the training set and the test set that meet the threshold requirements to build a second random number generator seed database;
[0015] Dividing the data set into a training set and a test set using a second random number generator seed to generate a second training and test set database;
[0016] Performing model training on the training set in the second training and test set database, and saving the trained model to obtain a second prediction model database;
[0017] The prediction model in the second prediction model database is used to predict the PCB electroplating parameters to obtain a prediction value list.
[0018] Furthermore, the calculation formulas for the evaluation index values of the training set and the test set are:
[0019] Among them, RMES is the evaluation index value, y i Indicates the actual value, represents the predicted value, and n is the number of samples.
[0020] Furthermore, the number of predicted values in the predicted value list is greater than or equal to 1.
[0021] Further, the method of predicting the PCB electroplating parameters using the prediction model in the second prediction model database to obtain a prediction value list includes:
[0022] All predicted values in the predicted value list are weighted to obtain the final predicted value.
[0023] Furthermore, the calculation formula for the weighted processing is:
[0024] Among them, predict1 is the specific predicted value in the predicted value list, n is the number of predicted values, and predict2 is the final predicted value.
[0025] Furthermore, the weighted processing of all the predicted values in the predicted value list to obtain the final predicted value includes:
[0026] The final predicted values are used to automatically update the corresponding parameters in the PCB electroplating parameters.
[0027] Furthermore, the original production data of the PCB includes current density, actual value of surface copper plating, hole copper plating requirements, surface copper plating requirements, board thickness, minimum aperture and aspect ratio.
[0028] In a second aspect, the present invention also provides a training and prediction device for a PCB electroplating parameter prediction model, comprising an acquisition unit, a preprocessing unit, a first construction unit, a first division unit, a first model training unit, a comparison unit, a setting unit, a second construction unit, a second division unit, a second model training unit and a prediction unit;
[0029] The acquisition unit is used to acquire the original production data of the PCB;
[0030] The preprocessing unit is used to preprocess the original production data of the PCB to obtain a data set;
[0031] The first construction unit is used to construct a first random number generator seed database and set a range of the first random number generator seed;
[0032] The first division unit is used to divide the data set into a data set and a test set according to a set range of the first random number generator seed to generate a first training and test set database;
[0033] The first model training unit is used to perform model training on the training set in the first training and test set database, and save the trained model to obtain a first prediction model database;
[0034] The comparison unit is used to output predicted values for the training set and the data set using each prediction model in the first prediction model database, and compare the obtained predicted values with the actual values to obtain evaluation index values for the training set and the test set;
[0035] The setting unit is used to set the thresholds of the evaluation index values of the training set and the test set according to the electroplating parameter control error requirements;
[0036] The second construction unit is used to screen out the evaluation index values of the training set and the test set that meet the threshold, and use the first random number generator seeds corresponding to the training set and the test set that meet the threshold requirements to construct a second random number generator seed database;
[0037] The second partitioning unit is used to partition the data set into a training set and a test set using a second random number generator seed to generate a second training and test set database;
[0038] The second model training unit is used to perform model training on the training set in the second training and test set database, and save the trained model to obtain a second prediction model database;
[0039] The prediction unit is used to predict the PCB electroplating parameters using the prediction model in the second prediction model database to obtain a prediction value list.
[0040] In a third aspect, the present invention also provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the training and prediction method of the PCB electroplating parameter prediction model as described above is implemented.
[0041] In a fourth aspect, the present invention further provides a computer-readable storage medium, wherein the storage medium stores a computer program, wherein the computer program includes program instructions, and when the program instructions are executed by a processor, the processor executes the training and prediction method of the PCB electroplating parameter prediction model as described above.
[0042] Compared with the prior art, the beneficial effect of the present invention is that the present invention divides the data set and the training model multiple times, and the latter is performed on the basis of the previous one, so that a model that performs well on different data sets can be found, thereby avoiding the problems of overfitting (the model performs well on the training set but poorly on the test set) or underfitting (the model performs poorly on both the training set and the test set). At the same time, a more stable model can be found through multiple attempts and screening, making the prediction results more reliable and improving the quality of production.
[0043] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the following preferred embodiments are specifically cited and described in detail as follows. BRIEF DESCRIPTION OF THE DRAWINGS
[0044] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying any creative work.
[0045] Figure 1A flowchart of a training and prediction method of a PCB electroplating parameter prediction model provided in a specific embodiment of the present invention;
[0046] Figure 2 A schematic block diagram of a training and prediction device for a PCB electroplating parameter prediction model provided in a specific embodiment of the present invention;
[0047] Figure 3 A schematic block diagram of a computer device provided for a specific embodiment of the present invention. DETAILED DESCRIPTION
[0048] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0049] It should be understood that when used in this specification and the appended claims, the terms "include" and "comprises" indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.
[0050] It should also be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.
[0051] It should be further understood that the term "and / or" used in the present description and the appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.
[0052] like Figure 1 As shown, an embodiment of the present invention provides a training and prediction method for a PCB electroplating parameter prediction model. The method is applicable to an automated production management system (such as an MES system). The method includes the following steps: S1-S11.
[0053] S1. Obtain the original production data of PCB.
[0054] Automated production management systems (such as MES systems) can monitor various parameters and status on the production line in real time. Through integration with these systems, production data can be automatically collected. Therefore, the original production data of PCB can be exported from the automated production management system.
[0055] In this embodiment, the original production data of the PCB includes current density, actual value of surface copper plating, hole copper plating requirements, surface copper plating requirements, board thickness, minimum hole diameter, and aspect ratio.
[0056] S2. Preprocess the original production data of the PCB to obtain a data set.
[0057] The preprocessing here mainly includes: removing duplicate records, records with missing fields, and records with outliers.
[0058] The purpose of removing duplicate records is to avoid the same record appearing multiple times in the data set, which may cause statistical bias and model overfitting. Specifically, methods in data frame tools (such as Pandas) (such as drop_duplicates()) are used to detect whether there are exactly the same records in the data set. Once duplicate records are detected, you can choose to keep the first or last record as needed and delete other duplicate records.
[0059] The purpose of missing field records is to ensure that each record is complete and contains all the required fields, avoiding errors or reduced accuracy when the model handles missing data. Specifically, check whether there are missing values (NaN or null values) in each field in the dataset. You can use the isna() or isnull() method. If the key field of a record is missing, you can choose to delete the entire record. For non-key fields, you can choose to fill the missing values with the mean, median, or other appropriate values.
[0060] The purpose of outlier records is to remove outliers in the data set that are significantly different from the usual situation, so as not to affect the training and prediction results of the model. Specifically, statistical methods (such as Z-score, interquartile range method) or algorithms (such as clustering, isolation forest) are used to detect outliers in the data. The detected outlier records can be deleted directly, or the outliers can be replaced with more reasonable data, such as the mean or median.
[0061] S3. Construct a first random number generator seed database and set a range of the first random number generator seed.
[0062] The random number generator seed is used to initialize the random number generator to ensure that the sequence of random numbers generated is the same each time the same seed value is used. The seed database is a structure that stores all predefined seed values. This database can be a simple list, array, or even a database table, depending on specific requirements and implementation details. The seed range refers to the possible range of seed values, such as from 0 to 100. This range is set to enable the use of multiple seed values in experiments to ensure the stability and comprehensiveness of different data partitions and model training results.
[0063] S4. Divide the data set into a training set and a test set according to the set range of the first random number generator seed to generate a first training and test set database.
[0064] By setting the range of seed values, the data set can be divided multiple times for cross-validation or evaluating the stability of the model. Different seed values will generate different combinations of training sets and test sets. The original data set is divided into training sets and test sets. The training set is used for model training, and the test set is used to evaluate the performance of the model. The data set is randomly divided according to a certain ratio (such as 80% training set, 20% test set). Multiple different sets of training sets and test sets are combined to form the first training and test set database.
[0065] S5. Perform model training on the training set in the first training and test set database, and save the trained model to obtain a first prediction model database.
[0066] Select a machine learning algorithm suitable for the data set and task to train the model for each training set in the first training and test set database. Machine learning algorithms such as lightgbm, xgboost, random forest, neural network, multivariate linear regression, etc. Store the trained model in the first prediction model database for subsequent prediction and evaluation. This database can be a simple dictionary, file system storage, or a more complex database system.
[0067] S6. Use each prediction model in the first prediction model database to output prediction values for the training set and the data set, and compare the obtained prediction values with the actual values to obtain evaluation index values for the training set and the test set.
[0068] Each prediction model in the first prediction model database predicts the training set and the test set, and calculates the corresponding RMSE index, which can effectively evaluate the performance of the model. The RMSE of the training set (train_rmse) and the RMSE of the test set (test_rmse) measure the performance of the model on the training data and unseen data respectively. These evaluation indicators can help us further understand the generalization ability of the model and select and optimize the most suitable model.
[0069] For each training set, use the corresponding model to make predictions and calculate the RMSE of the training set. For each test set, use the corresponding model to make predictions and calculate the RMSE of the test set.
[0070] The calculation formula for the evaluation index values of the training set and the test set is:
[0071] Among them, RMES is the evaluation index value, y iIndicates the actual value, represents the predicted value, and n is the number of samples.
[0072] S7. According to the electroplating parameter control error requirements, set the thresholds of the evaluation index values of the training set and the test set.
[0073] Different electroplating parameter error requirements are different, so the threshold can be set according to the specific electroplating parameters that need to be controlled and the error requirements corresponding to the parameters. For example, the electroplating thickness control error requirement is 0-3μm. According to the normal distribution rule, the upper limit of the train_rmse and test_rmse thresholds is set to t2<=0.8, and the lower limit is t1=t2-0.2. The setting of 0.8 is based on the output of the electroplated copper thickness of 1ASF*h electroplating time is <=1μm.
[0074] S8. Filter out the evaluation index values of the training set and the test set that meet the threshold, and use the first random number generator seeds corresponding to the training set and the test set that meet the threshold requirements to build a second random number generator seed database.
[0075] According to the train_rmse and test_rmse values, the random number generator seeds that meet the given interval are screened out and stored in a new random number generator seed database seed2 (the second random number generator seed database). This is a process of selecting the best model.
[0076] By screening models whose train_rmse and test_rmse values are both within the given interval [t1, t2] and storing the corresponding random number generator seeds in the new seed database seed2, the seeds corresponding to the models with better performance can be effectively selected to prevent overfitting or underfitting problems.
[0077] S9. Use the second random number generator seed to divide the data set into a training set and a test set to generate a second training and test set database.
[0078] S10: Perform model training on the training set in the second training and test set database, and save the trained model to obtain a second prediction model database.
[0079] For steps S9 and S10, in order to further optimize the model performance, the screened seeds are divided into data sets, and corresponding training and test sets are generated. Then, these data sets are used for model training to generate a new prediction model database model2 (second prediction model database).
[0080] By using the screened random number generator seed to divide the data set, a new training and test set database data2 is generated, and the ensemble learning algorithm is applied to these data sets for model training, and finally the prediction model database model2 is generated. This method combines the advantages of ensemble learning and helps to improve the generalization ability of the model, thereby obtaining more stable and higher prediction performance.
[0081] S11. Predicting PCB electroplating parameters using the prediction model in the second prediction model database to obtain a prediction value list.
[0082] According to the requirements, the prediction model in the second prediction model database is used to predict the specific PCB electroplating parameters to obtain a prediction value list. For example, the model in model2 is used to predict the current density of the PCB material number to be produced, and a prediction value list predict1 is generated.
[0083] Note that the number of predicted values in the predicted value list is greater than or equal to 1.
[0084] In one embodiment, step S11 includes the following steps: S12.
[0085] S12. Weighting all the predicted values in the predicted value list to obtain the final predicted value.
[0086] The calculation formula for weighted processing is:
[0087] Among them, predict1 is the specific predicted value in the predicted value list, n is the number of predicted values, and predict2 is the final predicted value.
[0088] By weighting all the predicted values, a more accurate and stable final predicted value predict2 can be output. This weighted processing method takes into account the different performance of each model, can make full use of the advantages of each model, and improve the accuracy of the final prediction result. The final prediction result predict2 can be saved to a file or database for subsequent analysis and use.
[0089] In one embodiment, step S12 includes the following steps: S13.
[0090] S13. Automatically update corresponding parameters in the PCB electroplating parameters using the final predicted values.
[0091] In order to realize the automatic update of parameters, the final predicted value is automatically submitted and updated through the MES system, thus realizing production automation.
[0092] The present invention divides the data set and trains the model multiple times, and the latter is performed on the basis of the previous one. Therefore, a model that performs well on different data sets can be found, thereby avoiding the problems of overfitting (the model performs well on the training set but poorly on the test set) or underfitting (the model performs poorly on both the training set and the test set). At the same time, a more stable model can be found through multiple attempts and screening, making the prediction results more reliable and improving the quality of production.
[0093] The embodiment of the present invention also provides a training and prediction device for a PCB electroplating parameter prediction model, which is used to perform the steps in any embodiment of the training and prediction method for a PCB electroplating parameter prediction model. Figure 2 , Figure 2 A schematic block diagram of a training and prediction device 1 for a PCB electroplating parameter prediction model provided in an embodiment of the present application is shown. The training and prediction device 1 for a PCB electroplating parameter prediction model specifically includes an acquisition unit 10, a preprocessing unit 20, a first construction unit 30, a first division unit 40, a first model training unit 50, a comparison unit 60, a setting unit 70, a second construction unit 80, a second division unit 90, a second model training unit 100 and a prediction unit 110.
[0094] The acquisition unit 10 is used to acquire the original production data of the PCB. The preprocessing unit 20 is used to preprocess the original production data of the PCB to obtain a data set. The first construction unit 30 is used to construct a first random number generator seed database and set the range of the first random number generator seed. The first division unit 40 is used to divide the data set into a training set and a test set according to the set range of the first random number generator seed to generate a first training and test set database. The first model training unit 50 is used to perform model training on the training set in the first training and test set database, and save the trained model to obtain a first prediction model database. The comparison unit 60 is used to use each prediction model in the first prediction model database to output the prediction value of the training set and the data set, and compare the obtained prediction value with the actual value to obtain the evaluation index value of the training set and the test set. The setting unit 70 is used to set the threshold value of the evaluation index value of the training set and the test set according to the electroplating parameter control error requirements. The second construction unit 80 is used to screen out the evaluation index values of the training set and the test set that meet the threshold, and use the first random number generator seeds corresponding to the training set and the test set that meet the threshold requirements to construct a second random number generator seed database. The second division unit 90 is used to divide the data set into a training set and a test set using the second random number generator seed to generate a second training and test set database. The second model training unit 100 is used to perform model training on the training set in the second training and test set database, and save the trained model to obtain a second prediction model database. The prediction unit 110 is used to predict PCB electroplating parameters using the prediction model in the second prediction model database to obtain a prediction value list.
[0095] In one embodiment, the training and prediction device 1 for the PCB electroplating parameter prediction model further includes a weighted processing unit.
[0096] The weighted processing unit is used to perform weighted processing on all the predicted values in the predicted value list to obtain the final predicted value.
[0097] In one embodiment, the training and prediction device 1 for the PCB electroplating parameter prediction model further includes an updating unit.
[0098] The updating unit is used to automatically update the corresponding parameters in the PCB electroplating parameters using the final predicted values.
[0099] It should be noted that technicians in the relevant field can clearly understand that the training of the above-mentioned PCB electroplating parameter prediction model, the specific implementation process of the prediction device 1 and each unit can refer to the corresponding description in the aforementioned method embodiment, and for the convenience and conciseness of the description, it will not be repeated here.
[0100] The training and prediction device of the PCB electroplating parameter prediction model can be implemented in the form of a computer program. The computer program can be used in Figure 3 Runs on the computer device shown.
[0101] See also Figure 3 , Figure 3 700 is a schematic block diagram of a computer device provided in an embodiment of the present application. The computer device 700 may be a server, wherein the server may be an independent server or a server cluster composed of multiple servers.
[0102] like Figure 3 As shown, the computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the training and prediction method steps of the PCB electroplating parameter prediction model as described above are implemented.
[0103] The computer device 700 may be a terminal or a server. The computer device 700 includes a processor 720 , a memory, and a network interface 750 connected via a system bus 710 , wherein the memory may include a non-volatile storage medium 730 and an internal memory 740 .
[0104] The non-volatile storage medium 730 can store an operating system 731 and a computer program 732. When the computer program 732 is executed, the processor 720 can execute the training and prediction method of the PCB electroplating parameter prediction model.
[0105] The processor 720 is used to provide computing and control capabilities and support the operation of the entire computer device 700.
[0106] The internal memory 740 provides an environment for the operation of the computer program 732 in the non-volatile storage medium 730. When the computer program 732 is executed by the processor 720, the processor 720 can execute the training and prediction methods of the PCB electroplating parameter prediction model.
[0107] The network interface 750 is used for network communication, such as sending assigned tasks, etc. Those skilled in the art will appreciate that Figure 3 The structure shown in the figure is only a block diagram of a part of the structure related to the present application scheme, and does not constitute a limitation on the computer device 700 to which the present application scheme is applied. The specific computer device 700 may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements. The processor 720 is used to run the program code stored in the memory to implement the training and prediction method steps of the PCB electroplating parameter prediction model:
[0108] Those skilled in the art will understand that Figure 3 The embodiments of the computer device shown in the figure do not constitute a limitation on the specific composition of the computer device. In other embodiments, the computer device may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently. For example, in some embodiments, the computer device may only include a memory and a processor. In such embodiments, the structure and function of the memory and the processor are the same as those of the embodiment of the present invention. Figure 3 The embodiments shown are consistent and will not be described again here.
[0109] It should be understood that in the embodiment of the present application, the processor 720 may be a central processing unit (CPU), and the processor 720 may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.
[0110] In another embodiment of the present invention, a computer-readable storage medium is provided. The computer-readable storage medium may be a non-volatile computer-readable storage medium. The computer-readable storage medium stores a computer program, wherein when the computer program is executed by a processor, the training and prediction method of the PCB electroplating parameter prediction model disclosed in the embodiment of the present invention is implemented.
[0111] Those skilled in the art can clearly understand that, for the convenience and simplicity of description, the specific working process of the above-described equipment, devices and units can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here. Those of ordinary skill in the art can appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented with electronic hardware, computer software or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to the function. Whether these functions are executed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.
[0112] In the several embodiments provided by the present invention, it should be understood that the disclosed equipment, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. Units with the same function may also be combined into one unit. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interfaces, devices or units, or may be electrical, mechanical or other forms of connection.
[0113] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the embodiments of the present invention.
[0114] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.
[0115] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), disk or optical disk and other media that can store program codes.
[0116] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.
Claims
1. A training and prediction method for a PCB electroplating parameter prediction model, characterized in that: include: Get the original production data of PCB; Preprocess the original production data of PCB to obtain a data set; Constructing a first random number generator seed database and setting a range of the first random number generator seed; Dividing the data set into a data set and a test set according to the set range of the first random number generator seed to generate a first training and test set database; Performing model training on the training set in the first training and test set database, and saving the trained model to obtain a first prediction model database; Output predicted values for the training set and the data set using each prediction model in the first prediction model database, and compare the obtained predicted values with the actual values to obtain evaluation index values for the training set and the test set; According to the electroplating parameter control error requirements, set the thresholds of the evaluation index values of the training set and the test set; Screen out the evaluation index values of the training set and the test set that meet the threshold, and use the first random number generator seeds corresponding to the training set and the test set that meet the threshold requirements to build a second random number generator seed database; Dividing the data set into a training set and a test set using a second random number generator seed to generate a second training and test set database; Performing model training on the training set in the second training and test set database, and saving the trained model to obtain a second prediction model database; The prediction model in the second prediction model database is used to predict the PCB electroplating parameters to obtain a prediction value list.
2. The training and prediction method of a PCB electroplating parameter prediction model according to claim 1, characterized in that: The calculation formula for the evaluation index value of the training set and the test set is: Among them, RMES is the evaluation index value, y i Indicates the actual value, represents the predicted value, and n is the number of samples.
3. The training and prediction method of a PCB electroplating parameter prediction model according to claim 1, characterized in that: The number of predicted values in the predicted value list is greater than or equal to 1.
4. The training and prediction method of a PCB electroplating parameter prediction model according to claim 3 is characterized in that: After predicting the PCB electroplating parameters using the prediction model in the second prediction model database to obtain a prediction value list, the method further comprises: All predicted values in the predicted value list are weighted to obtain the final predicted value.
5. The training and prediction method of a PCB electroplating parameter prediction model according to claim 4, characterized in that: The calculation formula for the weighted processing is: Among them, predict1 is the specific predicted value in the predicted value list, n is the number of predicted values, and predict2 is the final predicted value.
6. The training and prediction method of a PCB electroplating parameter prediction model according to claim 4, characterized in that: The weighted processing of all the predicted values in the predicted value list to obtain the final predicted value includes: The final predicted values are used to automatically update the corresponding parameters in the PCB electroplating parameters.
7. The training and prediction method of a PCB electroplating parameter prediction model according to claim 1, characterized in that: The original production data of the PCB includes current density, actual value of surface copper plating, hole copper plating requirements, surface copper plating requirements, board thickness, minimum aperture and aspect ratio.
8. A training and prediction device for a PCB electroplating parameter prediction model, characterized in that: It includes an acquisition unit, a preprocessing unit, a first construction unit, a first division unit, a first model training unit, a comparison unit, a setting unit, a second construction unit, a second division unit, a second model training unit and a prediction unit; The acquisition unit is used to acquire the original production data of the PCB; The preprocessing unit is used to preprocess the original production data of the PCB to obtain a data set; The first construction unit is used to construct a first random number generator seed database and set a range of the first random number generator seed; The first division unit is used to divide the data set into a data set and a test set according to a set range of the first random number generator seed to generate a first training and test set database; The first model training unit is used to perform model training on the training set in the first training and test set database, and save the trained model to obtain a first prediction model database; The comparison unit is used to output predicted values for the training set and the data set using each prediction model in the first prediction model database, and compare the obtained predicted values with the actual values to obtain evaluation index values for the training set and the test set; The setting unit is used to set the thresholds of the evaluation index values of the training set and the test set according to the electroplating parameter control error requirements; The second construction unit is used to screen out the evaluation index values of the training set and the test set that meet the threshold, and use the first random number generator seeds corresponding to the training set and the test set that meet the threshold requirements to construct a second random number generator seed database; The second partitioning unit is used to partition the data set into a training set and a test set using a second random number generator seed to generate a second training and test set database; The second model training unit is used to perform model training on the training set in the second training and test set database, and save the trained model to obtain a second prediction model database; The prediction unit is used to predict the PCB electroplating parameters using the prediction model in the second prediction model database to obtain a prediction value list.
9. A computer device, characterized in that: The invention comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the training and prediction method of the PCB electroplating parameter prediction model as described in any one of claims 1 to 7 is implemented.
10. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, which includes program instructions. When the program instructions are executed by a processor, the processor executes the training and prediction method of the PCB electroplating parameter prediction model as described in any one of claims 1 to 7.
Citation Information
Patent Citations
Prediction method of sample data set and hybrid prediction model
CN113780411A
PCB electroplating parameter prediction method and device based on machine learning algorithm and medium
CN117350242A
Quality prediction model generating device, quality prediction device, quality prediction model generating method, and quality prediction method
JP2023128678A
Systems and Methods for Deriving and Optimizing Classifiers from Multiple Datasets
US20200303078A1
Object detection model training
US20230282007A1
