Method, device and equipment for reconstructing unstructured influencing factors of solder paste printing on SMT production lines

Abstract

An embodiment of the present invention provides a method, a reconstruction device, a reconstruction equipment and a computer-readable storage medium for reconstructing unstructured influencing factors of solder paste printing on an SMT production line. The method generates an original data set by collecting original production data; performs feature interaction on the influencing factors in the original data set to obtain an influencing factor set; converts the influencing factor set into unstructured data to obtain a reconstructed data packet, establishes the relationship between the influencing factors through feature interaction, and then converts the structured data into unstructured data. On the one hand, it can solve the problem of high repeatability of data in the solder paste printing stage of the SMT production line. At the same time, the unstructured data processing method is applied to the data in the solder paste printing stage of the SMT production line, which solves the problem that the printing parameters brought by the original production data of the solder paste printing link are complicated, and the parameters affect each other and have interactive relationships, making it impossible to accurately locate the influencing factors, and thus it is impossible to identify solder paste printing defects based on the influencing factors, resulting in a high printing defect rate and waste of resources.

Description

Technical Field

[0001] The embodiments of the present invention relate to, but are not limited to, the field of intelligent manufacturing technology. Specifically, they relate to, but are not limited to, a method for reconstructing unstructured influencing factors of solder paste printing on an SMT production line, a reconstruction device, a reconstruction equipment, and a computer-readable storage medium. Background Art

[0002] Surface Mount Technology (SMT) is a popular technology and process in the current electronics assembly industry. It is a circuit assembly technology that mounts leadless or short-lead surface mount components (abbreviated as SMC / SMD, called chip components in Chinese) on the surface of a printed circuit board (PCB) or other substrate, and assembles them by soldering through methods such as reflow soldering or dip soldering.

[0003] With economic development, market demands for electronic products are becoming increasingly demanding. Components are becoming flatter and more miniaturized, printed circuit board assembly dimensions are shrinking, and pads are becoming denser and denser. This places higher demands on surface mount technology (SMT). Surface mount technology primarily involves three processes: solder paste printing, component placement, and reflow soldering. Solder paste printing is the primary and most critical step. Analysis shows that approximately 70% of SMT product quality issues are caused by poor solder paste printing performance.

[0004] The solder paste printing process of printed circuit boards in surface mount technology is complex. The printing parameters brought by the original production data are complicated and diverse. The parameters affect each other and have interactive relationships, making the factors leading to poor printing more complex and hidden, difficult to analyze, and impossible to accurately locate. As a result, it is impossible to identify solder paste printing defects based on the influencing factors, which wastes resources. Summary of the Invention

[0005] The embodiments of the present invention provide a method, a device, a reconstruction equipment and a computer-readable storage medium for reconstructing unstructured influencing factors of solder paste printing on SMT production lines. The main technical problem to be solved is that in related technologies, the printing parameters brought by the original production data of the solder paste printing link are complicated, and the parameters affect each other and have interactive relationships. It is impossible to accurately locate the influencing factors, and thus it is impossible to identify solder paste printing defects based on the influencing factors, resulting in a high printing defect rate and waste of resources.

[0006] To solve the above technical problems, an embodiment of the present invention provides a method for reconstructing unstructured influencing factors of solder paste printing on an SMT production line, comprising:

[0007] Collect original production data to generate original data sets;

[0008] Perform feature interaction on each influencing factor in the original data set to obtain an influencing factor set;

[0009] The influencing factor set is converted into unstructured data to obtain a reconstructed data packet.

[0010] An embodiment of the present invention further provides a reconstruction device, comprising:

[0011] The acquisition module is used to collect original production data to generate original data sets;

[0012] Feature interaction module, the feature interaction module is used to perform feature interaction on each influencing factor in the original data set to obtain an influencing factor set;

[0013] Reconstruction module,The reconstruction module is used to convert the influencing factor set into unstructured data,to obtain the reconstructed data packet.

[0014] An embodiment of the present invention also provides a reconstruction device, which includes a processor, a memory, and a communication bus;

[0015] The communication bus is used to realize the connection and communication between the processor and the memory;

[0016] The processor is used to execute one or more computer programs stored in the memory to implement the steps of the above-mentioned method for reconstructing unstructured influencing factors of solder paste printing in an SMT production line.

[0017] An embodiment of the present invention also provides a computer storage medium, wherein the computer-readable storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps of the above-mentioned method for reconstructing unstructured influencing factors of SMT production line solder paste printing.

[0018] According to the unstructured influencing factor reconstruction method, reconstruction device, reconstruction equipment and computer-readable storage medium for SMT production line solder paste printing provided by the embodiments of the present invention, an original data set is generated by collecting original production data; feature interaction is performed on the influencing factors in the original data set to obtain an influencing factor set; the influencing factor set is converted into unstructured data to obtain a reconstructed data packet, and the relationship between the influencing factors is established through feature interaction. Then, the structured data is converted into unstructured data. On the one hand, it can solve the problem of high data repeatability in the solder paste printing stage of the SMT production line. At the same time, the unstructured data processing method is applied to the data of the solder paste printing stage of the SMT production line, which solves the problem that the printing parameters brought by the original production data of the solder paste printing link are complicated, and the parameters affect each other and have interactive relationships, making it impossible to accurately locate the influencing factors, and thus it is impossible to identify solder paste printing defects based on the influencing factors, resulting in a high printing defect rate and waste of resources.

[0019] Other features and corresponding beneficial effects of the present invention are described in the latter part of the specification, and it should be understood that at least some of the beneficial effects become obvious from the description in the specification of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 Schematic diagram of the basic process of the method for reconstructing unstructured influencing factors of solder paste printing in an SMT production line according to the first embodiment of the present invention;

[0021] Figure 2 This is a schematic diagram of the basic process of converting one-dimensional structured data into unstructured data according to the first embodiment of the present invention;

[0022] Figure 3 This is a schematic diagram of the basic process of the method for reconstructing unstructured influencing factors of solder paste printing on an SMT production line according to the second embodiment of the present invention;

[0023] Figure 4 A histogram of the importance scores of influencing factors provided in the second embodiment of the present invention;

[0024] Figure 5 This is a schematic diagram of the basic structure of a reconstruction device according to the third embodiment of the present invention;

[0025] Figure 6 This is a schematic diagram of the basic structure of a reconstruction device according to the third embodiment of the present invention. DETAILED DESCRIPTION

[0026] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the following is a further detailed description of the embodiments of the present invention through specific implementation methods in conjunction with the accompanying drawings. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0027] Example 1:

[0028] In order to solve the problem in the related art that the original production data of the solder paste printing process brings about a variety of complex printing parameters, and the parameters affect each other and have interactive relationships, which makes it impossible to accurately locate the influencing factors, and then it is impossible to identify solder paste printing defects based on the influencing factors, resulting in a high printing defect rate and waste of resources; the embodiment of the present invention proposes a method for reconstructing unstructured influencing factors of SMT production line solder paste printing, see Figure 1 , Figure 1 The figure shows the basic process flow of the reconstruction method of the unstructured influencing factors of solder paste printing on SMT production lines, which includes but is not limited to:

[0029] S101, collecting original production data to generate an original data set;

[0030] In some embodiments, collecting raw production data involves collecting data related to solder paste printing on pads during the solder paste printing phase of an SMT production line. Specifically, the collected raw production data includes, but is not limited to, at least one of the following: batch data, process parameter data, printing process parameter data, and serial peripheral interface (SPI) inspection data. Batch data refers to product attribute parameters for a particular batch, including but not limited to PCB board length, width, and height. Process parameter data refers to printing process parameters manually set before printing a particular batch of products, including but not limited to board length, board width, board height, squeegee speed, squeegee pressure, cleaning speed, worktable separation speed, worktable separation distance, worktable printing height compensation, etc. Printing process parameter data refers to process data generated during the printing process, including but not limited to average pressure, minimum pressure, maximum pressure, cleaning supply time, squeegee separation speed, automatic cleaning, automatic cleaning count, manual cleaning, etc. SPI inspection data refers to data generated after a printed single board passes through an SPI inspection machine, including but not limited to solder paste volume, solder paste area, solder paste height, etc.

[0031] It should be understood that the original data set includes multiple original production data, and each original production data corresponds to one influencing factor.

[0032] S102, performing feature interaction on each influencing factor in the original data set to obtain an influencing factor set;

[0033] It should be understood that feature interaction is to obtain the features (influencing factor set) of the higher-dimensional space of the sample data by performing composite function operations on multiple column vectors of the influencing factors in the original data set. There are many parameters in the solder paste printing stage of the SMT production line, and the parameters affect each other and the logical relationship is unclear. Feature interaction can establish the relationship between the parameters (influencing factors). For example, for X={x1,x2,…,x p} for feature interaction, all vectors in X are column vectors, X n Represents an influencing factor, and obtains the interactive feature data set (influencing factor set) X′={x 1,1 ,x 1,2 ,x 1,3 ,…x p-1,p ,x p,p}.

[0034] In some embodiments, the method of feature interaction includes but is not limited to any of the following: product interaction, OR interaction, Sum interaction, Graph interaction, etc.; wherein, product interaction refers to the product of two features; OR interaction refers to the maximum value of two features; Sum interaction refers to the sum of two features; Graph interaction refers to the graph of two features.

[0035] S103, converting the influencing factor set into unstructured data to obtain a reconstructed data packet;

[0036] In some embodiments, before converting the influencing factor set into unstructured data to obtain a reconstructed data packet, the following steps are also included: calculating the importance score of each influencing factor in the original data set; and deleting the influencing factors with lower scores in the influencing factor set. It should be understood that when the number of influencing factors in the new influencing factor set formed after feature interaction is N*N+M, in order to convert the data into N*N unstructured data, it is necessary to delete the M influencing factors with lower scores based on the importance score. It should be understood that the M influencing factors deleted in this embodiment are deleted from the product interaction features of the influencing factors themselves in the deleted original feature set, and the data obtained after the interaction of different influencing factors is not deleted. For example, in some examples, the deleted influencing factor is "cleaning supply time * cleaning supply time". For example, the importance score of the influencing factor is determined by the Random Forest (RF) integration algorithm, and then the influencing factors are sorted in descending order according to their importance scores, and finally the M influencing factors with lower scores are deleted from low to high.

[0037] In some embodiments, before the influencing factor set is converted into unstructured data to obtain a reconstructed data packet, the following steps are also included: performing data reduction on the influencing factor set. It should be understood that the data dimensions of the influencing factors in the SMT solder paste printing stage are too different. For example, the magnitude and unit of each influencing factor are too different, which in turn affects the subsequent SPI defect recognition model. If data reduction is not performed, the defect recognition model training time will be too long and the model parameter magnitude will be too large. Therefore, it is necessary to perform data reduction on the influencing factor set to eliminate the impact of the magnitude and unit of the influencing factor set on the data mining work, thereby speeding up the training speed of certain models. For example, min-max normalization is used to perform data reduction on the influencing factor set.

[0038] In some embodiments, converting the influencing factor set into unstructured data to obtain a reconstructed data packet includes: converting the influencing factor set into unstructured data to obtain a reconstructed data packet through a one-dimensional loop; for example, Figure 2 As shown, when the influencing factor set is When , the influencing factor set to be processed is taken as one-dimensional data. First, the starting position of the loop is determined, and the data before the starting position is spliced ​​at the end of the original one-dimensional data. Then, a sliding window is set with n as the loop interval, and data is intercepted by interval to obtain n data segments. Finally, the data segments are spliced ​​by column to obtain n*n unstructured data.

[0039] In some embodiments, the method for reconstructing unstructured factors affecting solder paste printing on an SMT production line further includes inputting the reconstructed data packet into a defect recognition model to perform defect recognition. For example, the reconstructed data packet is input into a deep learning-based SPI defect recognition model to identify various solder paste printing defects, such as large volume, small volume, large area, small area, no solder paste, positive X offset, positive Y offset, negative Y offset, and excessive height.

[0040] The embodiment of the present invention provides a method for reconstructing unstructured influencing factors of SMT production line solder paste printing, which generates an original data set by collecting original production data; performs feature interaction on each influencing factor in the original data set to obtain an influencing factor set; converts the influencing factor set into unstructured data to obtain a reconstructed data packet, establishes the relationship between the influencing factors through feature interaction, and then converts the structured data into unstructured data. On the one hand, it can solve the problem of high repeatability of data in the solder paste printing stage of the SMT production line. At the same time, the unstructured data processing method is applied to the data of the solder paste printing stage of the SMT production line. Compared with the analysis method for structured data, the unstructured data processing method can better explore the intrinsic relationship between production factors after feature interaction, thereby greatly improving the accuracy of the SMT printing quality defect recognition type prediction model, and solves the problem that the printing parameters brought by the original production data of the solder paste printing link are complicated, and the parameters affect each other and have interactive relationships, making it impossible to accurately locate the influencing factors, and thus it is impossible to identify solder paste printing defects based on the influencing factors, resulting in a high printing defect rate and waste of resources.

[0041] Example 2:

[0042] In order to better understand the present invention, the embodiment of the present invention provides a more specific example to illustrate the method for reconstructing the unstructured influencing factors of MT production line solder paste printing; see Figure 3 , the method comprising:

[0043] S301, collect the original production data of the solder paste printing stage of the SMT production line to generate the original data set;

[0044] In some embodiments, collecting raw production data involves collecting data related to solder paste printing on pads during the solder paste printing phase of an SMT production line. Specifically, the collected raw production data includes, but is not limited to, at least one of the following: batch data, process parameter data, printing process parameter data, and serial peripheral interface (SPI) inspection data. Batch data refers to product attribute parameters for a particular batch, including but not limited to PCB board length, width, and height. Process parameter data refers to printing process parameters manually set before printing a particular batch of products, including but not limited to board length, board width, board height, squeegee speed, squeegee pressure, cleaning speed, worktable separation speed, worktable separation distance, worktable printing height compensation, etc. Printing process parameter data refers to process data generated during the printing process, including but not limited to average pressure, minimum pressure, maximum pressure, cleaning supply time, squeegee separation speed, automatic cleaning, automatic cleaning count, manual cleaning, etc. SPI inspection data refers to data generated after a printed single board passes through an SPI inspection machine, including but not limited to solder paste volume, solder paste area, solder paste height, etc. It should be understood that the original data set includes multiple original production data, and each original production data corresponds to one influencing factor. For example, the original data set formed by collecting original production data is shown in Table 1:

[0045] Table 1: Original dataset

[0046]

[0047] S302, performing feature interaction on each influencing factor in the original data set to obtain an influencing factor set;

[0048] In some embodiments, feature interaction is used to perform composite function operations on multiple column vectors of influencing factors in the original data set to obtain features (influencing factor set) in a higher-dimensional space of the sample data. It should be understood that there are many parameters in the solder paste printing stage of the SMT production line, and the parameters affect each other and the logical relationship is unclear. Feature interaction can establish the relationship between the parameters (influencing factors). For example, for X={x1,x2,…,x p} for feature interaction, all vectors in X are column vectors, X n Represents an influencing factor, and obtains the interactive feature data set (influencing factor set) X′={x 1,1 ,x 1,2 ,x 1,3 ,…x p-1,p ,x p,p For example, the feature interaction is performed by product interaction, and the calculation formula is: i,j =x i ×x j, where x ij Influencing factor x i and characteristic influencing factors x j The results of product interaction; the influencing factor set obtained after feature interaction of each influencing factor in the original data set is shown in Table 2:

[0049] Table 2, Influencing Factors

[0050]

[0051] S303, calculating the importance score of each influencing factor in the original data set, and deleting the influencing factors with lower scores in the influencing factor set;

[0052] In some embodiments, when the number of influencing factors in the new influencing factor set formed after feature interaction is N*N+M, in order to convert the data into N*N unstructured data, it is necessary to delete the M influencing factors with lower scores according to the importance score. For example, the feature importance score is determined by the Random Forest (RF) integration algorithm. Among them, the random forest belongs to the integration algorithm, which uses the decision tree as the base learner and adopts the bootstrap aggregating algorithm (Bagging) integration strategy. Since the SPI detection results are different defect types, the SMT solder paste printing scenario is a classification problem. For classification problems, the voting mechanism can be used to determine the category to which the sample belongs and perform feature evaluation according to the principle of minority obeys majority to obtain the feature importance score ranking, that is, to determine the importance score ranking of each influencing factor. Among them, the detailed process of random forest importance score calculation is as follows:

[0053] The feature importance score is represented by VIM, and the Gini index is represented by GI. When there are m features X1, X2, ...X m , that is, when there are m influencing factors, calculate each feature X j Gini index score That is, the calculation formula of the Gini index of the average change in node split impurity of the j-th feature in all RF decision trees is:

[0054]

[0055] Among them, K means there are K categories, p mk It represents the proportion of category k in node m, that is, the probability that two samples randomly selected from node m have inconsistent category labels.

[0056] Feature X j The importance of node m, that is, the change in the Gini index before and after the node m branches is:

[0057]

[0058] Among them, GI l and GI r Respectively represent the Gini index of the two new nodes after branching.

[0059] Finally, normalize all the importance scores obtained: the calculation process is as follows:

[0060]

[0061] in, is the sum of the gains of all features, is feature X j The Gini index.

[0062] It is important to understand that by inputting the influencing factors of the solder paste printing stage of the SMT production line (that is, the influencing factors in the original dataset) and the SPI test results into the random forest algorithm, the importance score of each influencing factor is obtained, as shown in Table 3:

[0063] Table 3. Importance scores of various influencing factors

[0064] Influencing factors Importance score Maximum pressure 0.19485 Width 0.149867 Minimum pressure 0.136368 average pressure 0.097049 long 0.0968 Workbench printing height compensation 0.088443 Automatic cleaning count 0.056794 high 0.045406 Workbench separation distance 0.025585 Scraper separation speed 0.023946 Scraper speed 0.019906 Cleaning speed 0.017801 Scraper pressure 0.015503 Scraper separation distance 0.015039 Automatic cleaning 0.009349 Workbench separation speed 0.003696 Cleaning supply time 0.003598

[0065] It should be understood that in order to more directly display the importance scores of each influencing factor, they can be sorted, such as Figure 4 As shown; according to the importance score, M influencing factors with lower scores are deleted. For example, the influencing factor set obtained by performing feature interaction on the influencing factors in the original data set includes 230 columns of influencing factors. In order to meet the input requirements of the unstructured data processing model, the 5 columns of influencing factors with the lowest random forest feature importance scores are deleted, and 225 columns of influencing factors remain, thereby reconstructing the data into 15*15 unstructured data; it should be understood that in some examples, the influencing factor supplementation method can be used to reconstruct the influencing factor set into unstructured data. For example, the influencing factor set obtained by performing feature interaction on the influencing factors in the original data set includes 230 columns of influencing factors. In order to meet the input requirements of the unstructured data processing model, the method of adding some original influencing factors is used to supplement the influencing factors, thereby obtaining 256 columns of influencing factors, thereby reconstructing the data into 16*16 unstructured data.

[0066] S304, performing data reduction on the influencing factor set;

[0067] In some embodiments, the data dimensions of the influencing factors in the SMT solder paste printing stage are too different. For example, the magnitude and unit of each influencing factor are too different, which in turn affects the SPI defect recognition model. Therefore, it is necessary to perform data reduction on the influencing factor set to eliminate the impact of the magnitude and unit of the influencing factor set on the data mining work, thereby speeding up the training speed of certain models. For example, min-max normalization is used to perform data reduction on the influencing factor set. Min-max normalization is also called minimum-maximum normalization or deviation standardization. This method compresses the data of each influencing factor to the [0,1] interval by performing a linear transformation on the influencing factors in the influencing factor set. The min-max conversion formula is: The formula is: The data obtained after data reduction of the influencing factor set is shown in Table 4:

[0068] Table 4. Data obtained after data reduction of the influencing factor set

[0069]

[0070] S305, converting the influencing factor set into unstructured data to obtain a reconstructed data packet;

[0071] In some embodiments, converting the influencing factor set into unstructured data to obtain a reconstructed data packet includes: converting the influencing factor set into unstructured data to obtain a reconstructed data packet in a one-dimensional loop; for example, when the influencing factor set to be processed is When processing the influencing factor set as one-dimensional data, first determine the starting point of the loop and splice the data before the starting point to the end of the original one-dimensional data. Then, set a sliding window with n as the loop interval, intercept the data by interval, and obtain n data segments. Finally, splice the data segments by column to obtain n*n unstructured data. For example, select the data in the first row of Table 4 for unstructured data conversion. Use the long*long in the first row as the starting point, and splice the data before this point to the end of the original one-dimensional data. After the one-dimensional loop is transformed, the unstructured data is obtained, as shown in Table 5. Table 5 shows the result of converting the data in the first row of Table 4.

[0072] Table 5. Unstructured data obtained after changing the one-dimensional loop method

[0073]

[0074] It is important to understand that the data of each row in the influencing factor set is subjected to a one-dimensional cyclic transformation in turn, thereby obtaining an unstructured reconstructed data packet;

[0075] S306, inputting the reconstructed data packet into the defect recognition model to perform defect recognition;

[0076] In some embodiments, the reconstructed data packet is input into the deep learning-based SPI defect recognition model to identify various solder paste printing defects, such as large volume, small volume, large area, small area, no solder paste, positive X offset, positive Y offset, negative Y offset, and high height. As shown in Table 6, Table 6 shows the recognition accuracy of an example defect.

[0077] Table 6, recognition accuracy of an example defect;

[0078]

[0079] The embodiment of the present invention provides a method for reconstructing unstructured influencing factors of SMT production line solder paste printing, which generates an original data set by collecting original production data; performs feature interaction on each influencing factor in the original data set to obtain an influencing factor set; converts the influencing factor set into unstructured data to obtain a reconstructed data packet, establishes the relationship between the influencing factors through feature interaction, and then converts the structured data into unstructured data. On the one hand, it can solve the problem of high repeatability of data in the solder paste printing stage of the SMT production line. At the same time, the unstructured data processing method is applied to the data of the solder paste printing stage of the SMT production line. Compared with the analysis method for structured data, the unstructured data processing method can better explore the intrinsic relationship between production factors after feature interaction, thereby greatly improving the accuracy of the SMT printing quality defect recognition type prediction model, and solves the problem that the printing parameters brought by the original production data of the solder paste printing link are complicated, and the parameters affect each other and have interactive relationships, making it impossible to accurately locate the influencing factors, and thus it is impossible to identify solder paste printing defects based on the influencing factors, resulting in a high printing defect rate and waste of resources.

[0080] Example 3:

[0081] This embodiment also provides a reconstruction device, such as Figure 5 As shown, it includes:

[0082] The acquisition module is used to collect original production data to generate original data sets;

[0083] Feature interaction module, the feature interaction module is used to perform feature interaction on each influencing factor in the original data set to obtain an influencing factor set;

[0084] Reconstruction module, which is used to convert the influencing factor set into unstructured data to obtain a reconstructed data packet

[0085] This embodiment also provides a reconstruction device, see Figure 6 As shown, it includes a processor 601, a memory 602 and a communication bus 603, wherein:

[0086] The communication bus 603 is used to realize the connection and communication between the processor 601 and the memory 602;

[0087] The processor 601 is configured to execute one or more computer programs stored in the memory 602 to implement at least one step of the method for reconstructing unstructured influencing factors of solder paste printing in an SMT production line in the first and second embodiments above.

[0088] The present embodiment also provides a computer-readable storage medium, which includes volatile or non-volatile, removable or non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, computer program modules or other data). Computer-readable storage media include, but are not limited to, RAM (Random Access Memory), ROM (Read-Only Memory), EEPROM (Electrically Erasable Programmable read only memory), flash memory or other memory technology, CD-ROM (Compact Disc Read-Only Memory), digital versatile disk (DVD) or other optical disk storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store the desired information and can be accessed by a computer.

[0089] The computer-readable storage medium in this embodiment can be used to store one or more computer programs, and the one or more computer programs stored therein can be executed by a processor to implement at least one step of the method for reconstructing unstructured influencing factors of SMT production line solder paste printing in the above-mentioned embodiments one and two.

[0090] It can be seen that those skilled in the art should understand that all or some of the steps, systems, and functional modules / units in the methods disclosed above can be implemented as software (which can be implemented using computer program code executable by a computing device), firmware, hardware, and appropriate combinations thereof. In hardware implementations, the division between the functional modules / units mentioned in the above description does not necessarily correspond to the division of physical components; for example, a physical component can have multiple functions, or a function or step can be performed by several physical components in cooperation. Some or all physical components can be implemented as software executed by a processor, such as a central processing unit, a digital signal processor, or a microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit.

[0091] In addition, it is well known to those skilled in the art that communication media generally contain computer-readable instructions, data structures, computer program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media. Therefore, the present invention is not limited to any specific hardware and software combination.

[0092] The above content is a further detailed description of the embodiments of the present invention in conjunction with specific implementation methods, and the specific implementation of the present invention cannot be considered to be limited to these descriptions. For those skilled in the art of the present invention, without departing from the concept of the present invention, several simple deductions or substitutions can be made, which should be considered to fall within the scope of protection of the present invention.

Claims

1. A method for reconstructing unstructured influencing factors of solder paste printing on an SMT production line, comprising: Collecting a plurality of original production data to generate an original data set, wherein one original production data corresponds to one influencing factor and one original production data includes a plurality of data values; Performing feature interaction on each of the influencing factors in the original data set to obtain an influencing factor set, wherein the influencing factor set includes a plurality of interaction feature data, and the interaction feature data is obtained by performing a composite function operation on data values ​​of the original data set corresponding to at least two of the influencing factors that perform feature interaction; The influencing factor set is converted into unstructured data in N*N format to obtain a reconstructed data packet.

2. The method for reconstructing unstructured influencing factors of solder paste printing in an SMT production line according to claim 1, characterized in that: The feature interactions include: product interaction, OR interaction, Sum interaction, and graph interaction.

3. The method for reconstructing unstructured influencing factors of solder paste printing in an SMT production line according to claim 1, characterized in that: Before converting the influencing factor set into unstructured data in N*N form to obtain a reconstructed data packet, the following steps are also included: Calculating the importance score of each influencing factor in the original data set; Delete the influencing factors with lower scores in the influencing factor set.

4. The method for reconstructing unstructured influencing factors of solder paste printing in an SMT production line according to claim 1, characterized in that: Before converting the influencing factor set into unstructured data in N*N format to obtain a reconstructed data packet, the method further includes: performing data reduction on the influencing factor set.

5. The method for reconstructing unstructured influencing factors of solder paste printing in an SMT production line according to any one of claims 1 to 4, characterized in that: The step of converting the influencing factor set into unstructured data in an N*N format to obtain a reconstructed data packet includes: The influencing factor set is converted into unstructured data in a one-dimensional loop manner to obtain a reconstructed data packet.

6. The method for reconstructing unstructured influencing factors of solder paste printing in an SMT production line according to claim 5, characterized in that: The original production data includes: batch data, process parameter data, printing process parameter data, and SPI detection data.

7. The method for reconstructing unstructured influencing factors of solder paste printing in an SMT production line according to claim 5, characterized in that: The method for reconstructing unstructured influencing factors of solder paste printing on an SMT production line further includes: The reconstructed data packet is input into a defect recognition model to perform defect recognition.

8. A reconstruction device comprising: A collection module, the collection module is used to collect multiple original production data to generate an original data set, wherein one original production data corresponds to one influencing factor, and one original production data includes multiple data values; a feature interaction module, configured to perform feature interaction on each of the influencing factors in the original data set to obtain an influencing factor set, wherein the influencing factor set includes a plurality of interaction feature data, and the interaction feature data is obtained by performing a composite function operation on data values ​​of the original data set corresponding to at least two of the influencing factors that perform feature interaction; A reconstruction module is used to convert the influencing factor set into unstructured data in N*N format to obtain a reconstructed data packet.

9. A reconstruction device, comprising a processor, a memory, and a communication bus; The communication bus is used to realize the connection and communication between the processor and the memory; The processor is configured to execute one or more computer programs stored in the memory to implement the steps of the method for reconstructing unstructured influencing factors of solder paste printing in an SMT production line according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium One or more computer programs are stored, and the one or more computer programs can be executed by one or more processors to implement the steps of the SMT production line solder paste printing unstructured influencing factor reconstruction method according to any one of claims 1 to 7.

Images