Printing machine equipment operation optimization process and system based on multi-vision analysis

Abstract

The invention discloses a printing machine equipment operation optimization process and system based on multi-vision analysis, and the method comprises the steps: obtaining multiple groups of image data in a printing process through employing a multi-vision collection scheme in a preset production period; main feature data are obtained through texture and contour feature extraction and a PCA analysis method, the recognition rate of solder paste printing is calculated through the recognition model, the recognition rate is judged, and an acquisition scheme is adjusted. And fusing the multi-image feature set by adopting a DCA feature fusion algorithm to generate a plurality of groups of feature maps, and processing the feature maps through a convolutional neural network to obtain a plurality of groups of fused feature data. And meanwhile, equipment operation parameters are obtained, defect identification is conducted on the fused feature data through the identification model, operation evaluation and optimization are conducted on the real-time equipment operation parameters based on defect information, and the printing quality and the production efficiency are ensured. According to the invention, equipment operation analysis and printing defect evaluation can be effectively carried out from multiple visual features, the intelligence of equipment operation is realized, and the production benefit is improved.

Description

Technical Field

[0001] The present invention relates to the field of multi-vision data analysis, and more specifically, to a printing press equipment operation optimization process and system based on multi-vision analysis. Background Art

[0002] With the development of the electronics manufacturing industry, automated production equipment plays an increasingly important role in improving production efficiency and product quality. As one of the key steps in the SMT production line, the quality of solder paste printing directly affects the quality of subsequent component welding and the reliability of the entire product. However, in actual production, due to the influence of various factors, such as incorrect configuration of template accuracy, scraper pressure, printing speed, etc., various defects in solder paste printing often occur, such as insufficient tin, excessive tin, offset, etc. Traditional defect detection methods mainly rely on manual visual inspection or detection by a single visual system, lack of comprehensive analysis of multiple visions, and have problems such as low efficiency, high false detection rate, and inability to provide real-time feedback. Therefore, it is of great significance to develop an efficient, accurate, and intelligent operation optimization process for printing equipment. Summary of the invention

[0003] The present invention overcomes the defects of the prior art and proposes a printing press equipment operation optimization process and system based on multi-visual analysis.

[0004] A first aspect of the present invention provides a printing press equipment operation optimization process based on multi-visual analysis, comprising: Under a preset production cycle, a multi-vision acquisition solution is configured, and multiple sets of printing image data of the solder paste printer during the printing process are obtained based on the multi-vision acquisition solution; Perform texture and contour-based feature extraction on each set of printed image data to obtain multiple image feature sets. Perform principal component feature extraction on the image feature sets based on the PCA analysis method to obtain principal feature data. Use the recognition model to calculate the solder paste printing recognition rate on the principal feature data. If the recognition rate is lower than the preset value, adjust the multi-vision acquisition scheme and re-acquire the image data until the corresponding recognition rate of each set of printed image data is higher than the preset value. Based on the DCA feature fusion algorithm, multiple image feature sets are fused with feature data. Each image feature set corresponds to a visual angle. Based on each image feature set, multiple feature maps are generated. Multiple feature maps are combined to obtain multiple groups of feature maps. Each group of feature maps is convolved and activated by a convolutional neural network to generate multiple groups of fused feature data based on multi-visual analysis. The operating parameter data of the solder paste printer equipment under a preset production cycle is obtained. Through the recognition model, defects are identified for multiple sets of fused feature data respectively. Based on the multiple sets of defect information obtained, operation evaluation and optimization of the real-time parameter data are performed.

[0005] In this solution, under a preset production cycle, a multi-vision acquisition solution is configured, and multiple sets of printing image data of the solder paste printer during the printing process are obtained based on the multi-vision acquisition solution, specifically: Set a preset production cycle and configure a multi-vision acquisition solution for the solder paste printer equipment; The multi-vision acquisition scheme includes a plurality of image acquisition setting angles, and each high-definition camera device corresponds to a setting angle; Based on the multi-vision acquisition solution, multiple sets of printing image data of the solder paste printer equipment during the printing process are obtained.

[0006] In this solution, among the multiple groups of printing image data, each group of printing image data corresponds to an image acquisition setting angle.

[0007] In this solution, the identification model is specifically: Build a CNN-based recognition model; The historical solder paste printing image data is normalized and imported into the recognition model for feature learning and recognition training. By setting the real label, the recognition output results are compared, and the recognition model parameters are updated through the back propagation algorithm to optimize the model. The recognition model training is repeated until the preset recognition standard is reached.

[0008] In this solution, the texture and contour-based feature extraction is performed on each set of printed image data to obtain multiple image feature sets. Based on the PCA analysis method, the principal component feature is extracted from the image feature set to obtain the main feature data. The solder paste printing recognition rate is calculated for the main feature data through the recognition model. If the recognition rate is lower than the preset value, the multi-vision acquisition scheme is adjusted and the image data is re-collected until the corresponding recognition rate of each set of printed image data is higher than the preset value, specifically: Through edge detection operators, feature extraction based on texture and contour is performed on each set of printed image data to form multiple image feature sets; Based on the PCA analysis method, the principal component features of each image feature set are extracted to obtain multiple groups of principal feature data; Through the recognition model, feature analysis and solder paste printing feature recognition are performed on each group of main feature data, interesting features and non-interest features are extracted and segmented, and the recognition rate based on the interesting features is calculated; Each set of main feature data corresponds to a recognition rate; If the recognition rate is lower than the preset value, the corresponding printed image data is marked, and at the same time, the setting angle of the corresponding vision is adjusted in the multi-vision acquisition scheme; Each set of main feature data is analyzed and the multi-vision acquisition scheme is adjusted, and the image data is re-collected until the corresponding recognition rate of each set of printed image data is higher than the preset value.

[0009] In this solution, among the multiple image feature sets, each image feature set corresponds to a set of printed image data.

[0010] In this solution, based on the DCA feature fusion algorithm, multiple image feature sets are fused with feature data. Each image feature set corresponds to a visual angle. Multiple feature maps are generated based on each image feature set. Multiple feature maps are combined to obtain multiple groups of feature maps. Each group of feature maps is convoluted and activated by a convolutional neural network to generate multiple groups of fused feature data based on multi-visual analysis, specifically: Import multiple image feature sets into the recognition model for feature learning and generate multiple feature maps; Combine multiple feature maps to obtain multiple groups of feature maps; Taking a set of feature maps as analysis units, based on the DCA feature fusion algorithm, setting preset weights, and performing convolution and activation operations on the set of feature maps through a convolutional neural network and the preset weights to generate multiple category activation maps; Perform spatial weighted fusion on multiple category activation maps and obtain a set of fused feature data; Multiple groups of fused feature data are obtained based on multiple groups of feature maps.

[0011] In this solution, the operating parameter data of the solder paste printer equipment under a preset production cycle is obtained, and defects are identified for multiple sets of fused feature data respectively through the recognition model. Based on the obtained multiple sets of defect information, the real-time parameter data is evaluated and optimized, specifically: Obtain the operating parameter data of the solder paste printer equipment under a preset production cycle; Import multiple groups of fused feature data into the recognition model for defect recognition, and generate multiple groups of defect information based on the recognition results; Based on the operating parameter data, multiple parameter optimization configurations are generated in combination with multiple sets of defect information; Real-time parameter tuning and evaluation of solder paste printer equipment is performed based on multiple parameter optimization configurations.

[0012] The second aspect of the present invention further provides a printing press equipment operation optimization system based on multi-visual analysis, the system comprising: a memory, a processor, the memory comprising a printing press equipment operation optimization program based on multi-visual analysis, the printing press equipment operation optimization program based on multi-visual analysis being executed by the processor to implement the following steps: Under a preset production cycle, a multi-vision acquisition solution is configured, and multiple sets of printing image data of the solder paste printer during the printing process are obtained based on the multi-vision acquisition solution; Perform texture and contour-based feature extraction on each set of printed image data to obtain multiple image feature sets. Perform principal component feature extraction on the image feature sets based on the PCA analysis method to obtain principal feature data. Use the recognition model to calculate the solder paste printing recognition rate on the principal feature data. If the recognition rate is lower than the preset value, adjust the multi-vision acquisition scheme and re-acquire the image data until the corresponding recognition rate of each set of printed image data is higher than the preset value. Based on the DCA feature fusion algorithm, multiple image feature sets are fused with feature data. Each image feature set corresponds to a visual angle. Based on each image feature set, multiple feature maps are generated. Multiple feature maps are combined to obtain multiple groups of feature maps. Each group of feature maps is convolved and activated by a convolutional neural network to generate multiple groups of fused feature data based on multi-visual analysis. The operating parameter data of the solder paste printer equipment under a preset production cycle is obtained. Through the recognition model, defects are identified for multiple sets of fused feature data respectively. Based on the multiple sets of defect information obtained, operation evaluation and optimization of the real-time parameter data are performed.

[0013] The third aspect of the present invention also provides a computer-readable storage medium, which includes a printing machine equipment operation optimization program based on multi-visual analysis. When the printing machine equipment operation optimization program based on multi-visual analysis is executed by a processor, the steps of the printing machine equipment operation optimization process based on multi-visual analysis as described in any one of the above items are implemented.

[0014] The present invention discloses a printing machine equipment operation optimization process and system based on multi-visual analysis. Within a preset production cycle, a multi-visual acquisition scheme is used to obtain multiple groups of image data in the printing process. The main feature data is obtained by texture and contour feature extraction and PCA analysis method, and the solder paste printing recognition rate is calculated using a recognition model. The recognition rate is judged and the acquisition scheme is adjusted. The DCA feature fusion algorithm is used to fuse multiple image feature sets to generate multiple groups of feature maps, which are processed by convolutional neural networks to obtain multiple groups of fused feature data. At the same time, the equipment operation parameters are obtained, and the fused feature data are identified for defects through the recognition model. The real-time equipment operation parameters are evaluated and optimized based on the defect information to ensure printing quality and production efficiency. Through the present invention, equipment operation analysis and printing defect evaluation can be effectively performed from multi-visual features, the intelligent operation of equipment can be realized, and the production efficiency can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 A flow chart showing a printing press equipment operation optimization process based on multi-visual analysis of the present invention; Figure 2 A block diagram of a printing press equipment operation optimization system based on multi-visual analysis according to the present invention is shown. DETAILED DESCRIPTION

[0016] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0017] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0018] Figure 1 A flow chart of a printing press equipment operation optimization process based on multi-visual analysis according to the present invention is shown.

[0019] like Figure 1 As shown, the first aspect of the present invention provides a printing press equipment operation optimization process based on multi-visual analysis, including: S102, configuring a multi-vision acquisition scheme in a preset production cycle, and acquiring multiple sets of printing image data of the solder paste printer during the printing process based on the multi-vision acquisition scheme; S104, performing texture and contour-based feature extraction on each set of printed image data to obtain multiple image feature sets, performing principal component feature extraction on the image feature sets based on the PCA analysis method to obtain principal feature data, and calculating the solder paste printing recognition rate on the principal feature data through the recognition model. If the recognition rate is lower than a preset value, adjusting the multi-vision acquisition scheme and re-acquiring image data until the corresponding recognition rate of each set of printed image data is higher than the preset value; S106, based on the DCA feature fusion algorithm, multiple image feature sets are fused with feature data, each image feature set corresponds to a visual angle, multiple feature maps are generated based on each image feature set, the multiple feature maps are combined to obtain multiple groups of feature maps, and each group of feature maps is convolved and activated by a convolutional neural network to generate multiple groups of fused feature data based on multi-visual analysis; S108, obtaining the operating parameter data of the solder paste printer equipment under a preset production cycle, and performing defect identification on multiple sets of fused feature data respectively through the recognition model, and performing operation evaluation and optimization on the real-time parameter data based on the obtained multiple sets of defect information.

[0020] According to an embodiment of the present invention, in a preset production cycle, a multi-vision acquisition scheme is configured, and multiple sets of printing image data of the solder paste printer device during the printing process are obtained based on the multi-vision acquisition scheme, specifically: Set a preset production cycle and configure a multi-vision acquisition solution for the solder paste printer equipment; The multi-vision acquisition scheme includes a plurality of image acquisition setting angles, and each high-definition camera device corresponds to a setting angle; Based on the multi-vision acquisition solution, multiple sets of printing image data of the solder paste printer equipment during the printing process are obtained.

[0021] It should be noted that the preset production cycle is a preset time period or a production cycle of a preset task. The multi-vision acquisition scheme includes multiple set angles. For example, for the printing of solder paste printed products, the angles can be set from the top, side, front and back. Printing defects can be comprehensively analyzed at different angles.

[0022] According to an embodiment of the present invention, among the multiple groups of printing image data, each group of printing image data corresponds to an image acquisition setting angle.

[0023] According to an embodiment of the present invention, the recognition model is specifically: Build a CNN-based recognition model; The historical solder paste printing image data is normalized and imported into the recognition model for feature learning and recognition training. By setting the real label, the recognition output results are compared, and the recognition model parameters are updated through the back propagation algorithm to optimize the model. The recognition model training is repeated until the preset recognition standard is reached.

[0024] It should be noted that the real label is the real defect information and feature information of the historical solder paste printing image data.

[0025] According to an embodiment of the present invention, the feature extraction based on texture and contour is performed on each group of printed image data to obtain multiple image feature sets, and the principal component feature is extracted on the image feature set based on the PCA analysis method to obtain the main feature data. The solder paste printing recognition rate is calculated for the main feature data through the recognition model. If the recognition rate is lower than the preset value, the multi-vision acquisition scheme is adjusted and the image data is re-collected until the corresponding recognition rate of each group of printed image data is higher than the preset value, specifically: Through edge detection operators, feature extraction based on texture and contour is performed on each set of printed image data to form multiple image feature sets; Based on the PCA analysis method, the principal component features of each image feature set are extracted to obtain multiple groups of principal feature data; Through the recognition model, feature analysis and solder paste printing feature recognition are performed on each group of main feature data, interesting features and non-interest features are extracted and segmented, and the recognition rate based on the interesting features is calculated; Each set of main feature data corresponds to a recognition rate; If the recognition rate is lower than the preset value, the corresponding printed image data is marked, and at the same time, the setting angle of the corresponding vision is adjusted in the multi-vision acquisition scheme; Each set of main feature data is analyzed and the multi-vision acquisition scheme is adjusted, and the image data is re-collected until the corresponding recognition rate of each set of printed image data is higher than the preset value.

[0026] It should be noted that the edge detection operator may be a Sobel operator, a Prewitt operator, etc. The recognition model can be used for feature analysis and recognition. The recognition rate is generally the amount of data of the feature of interest divided by the total amount of data of the feature of interest and the feature of no interest. It can be understood here that in the multi-visual feature analysis, since the data collected from multiple angles may not meet the subsequent recognition requirements, it is necessary to analyze the recognition features, judge by the recognition rate, and adjust the corresponding multi-visual acquisition scheme so that the features collected at each visual angle meet the needs of subsequent analysis.

[0027] According to an embodiment of the present invention, each of the plurality of image feature sets corresponds to a set of printed image data.

[0028] According to an embodiment of the present invention, the DCA feature fusion algorithm is used to fuse feature data of multiple image feature sets, each image feature set corresponds to a visual angle, and multiple feature maps are generated based on each image feature set. The multiple feature maps are combined to obtain multiple groups of feature maps, and each group of feature maps is convolved and activated by a convolutional neural network to generate multiple groups of fused feature data based on multi-visual analysis, specifically: Import multiple image feature sets into the recognition model for feature learning and generate multiple feature maps; Combine multiple feature maps to obtain multiple groups of feature maps; Taking a set of feature maps as analysis units, based on the DCA feature fusion algorithm, setting preset weights, and performing convolution and activation operations on the set of feature maps through a convolutional neural network and the preset weights to generate multiple category activation maps; Perform spatial weighted fusion on multiple category activation maps and obtain a set of fused feature data; Multiple groups of fused feature data are obtained based on multiple groups of feature maps.

[0029] It should be noted that the visual angle is the set angle in the multi-vision acquisition scheme. In the feature data fusion of multiple image feature sets, the previous process includes the adjustment of the multi-vision acquisition party and the re-acquisition of image data to obtain multiple image feature sets that meet expectations.

[0030] Each feature map corresponds to an image feature set, that is, each feature map corresponds to a set acquisition angle. Among multiple sets of feature maps, each set of feature maps represents a different angle combination. The specific combination can be set by the user, or each angle combination can be combined, analyzed and fused. Convolutional neural network is CNN.

[0031] According to an embodiment of the present invention, the operation parameter data of the solder paste printer equipment under a preset production cycle is obtained, and defect recognition is performed on multiple sets of fused feature data respectively through the recognition model, and operation evaluation and optimization of the real-time parameter data are performed based on the obtained multiple sets of defect information, specifically: Obtain the operating parameter data of the solder paste printer equipment under a preset production cycle; Import multiple groups of fused feature data into the recognition model for defect recognition, and generate multiple groups of defect information based on the recognition results; Based on the operating parameter data, multiple parameter optimization configurations are generated in combination with multiple sets of defect information; Real-time parameter tuning and evaluation of solder paste printer equipment is performed based on multiple parameter optimization configurations.

[0032] It should be noted that in the real-time parameter tuning and evaluation of the solder paste printer equipment according to the multiple parameter optimization configurations, it is determined whether there are conflicting parameter adjustments based on the multiple parameter optimization configurations, and adjustments are made according to the actual operating conditions. The equipment adjustments include the adjustment of printing speed, printing pressure, scraper angle, solder paste viscosity, solder paste volume, and ambient temperature. The defect information includes information such as defect characteristics, defect types, and defect occurrence time nodes.

[0033] It is worth mentioning that the traditional technology lacks multi-visual fusion analysis and is often based on simple visual images and single acquisition angle analysis, which is not comprehensive enough for feature recognition of solder paste printing. The present invention is based on multi-visual analysis and can effectively perform defect analysis from multi-dimensional visual feature data. Combined with the DCA fusion feature algorithm, it can perform different types of defect analysis on multi-visual features.

[0034] Defects include printing offset, insufficient solder paste, solder paste thickness not meeting expectations, missing, broken, misaligned printing dots, etc.

[0035] According to an embodiment of the present invention, it also includes: A group of defect information is randomly selected from multiple groups of defect information and marked as current defect information; Analyze the time node of the corresponding defect occurrence through the current defect information, and obtain the current operating parameters of the corresponding time node from the operating parameter data through the time node; Obtain defect features through current defect information, associate the defect features with current operating parameters and generate associated feature records; Generate multiple associated feature records based on multiple sets of defect information; Sort multiple associated feature records based on time nodes to form an evaluation association table; Serialize the record data in the evaluation association table to generate serialized data; Build a prediction model based on LSTM; Use the prediction model to predict serialized data, set the prediction time node, and generate prediction sequence data; Perform data analysis on the predicted sequence data to generate predicted defect features and predicted operating parameters; The printing production process is regulated and optimized based on the predicted defect characteristics and predictions.

[0036] It is worth mentioning that in traditional solder paste printing defect analysis, there is often a lack of effective prediction of defects and equipment operating parameters, and a lack of corresponding correlation analysis, which makes it difficult to improve printing quality and efficiency. Based on this, the present invention records the characteristics of defect information from multi-visual analysis, associates the corresponding operating parameters, and performs long short-term memory network (LSTM) prediction through serialized association records, effectively mining the correlation laws of printing production, and realizing the corresponding printing production correlation prediction, and optimizing the operation through correlation prediction, thereby improving the intelligent level of printing production and the monitoring optimization level.

[0037] Figure 2 A block diagram of a printing press equipment operation optimization system based on multi-visual analysis according to the present invention is shown.

[0038] The second aspect of the present invention further provides a printing press equipment operation optimization system 2 based on multi-visual analysis, the system comprising: a memory 21, a processor 22, the memory 21 comprising a printing press equipment operation optimization program based on multi-visual analysis, the printing press equipment operation optimization program based on multi-visual analysis being executed by the processor 22 to implement the following steps: Under a preset production cycle, a multi-vision acquisition solution is configured, and multiple sets of printing image data of the solder paste printer during the printing process are obtained based on the multi-vision acquisition solution; Perform texture and contour-based feature extraction on each set of printed image data to obtain multiple image feature sets. Perform principal component feature extraction on the image feature sets based on the PCA analysis method to obtain principal feature data. Use the recognition model to calculate the solder paste printing recognition rate on the principal feature data. If the recognition rate is lower than the preset value, adjust the multi-vision acquisition scheme and re-acquire the image data until the corresponding recognition rate of each set of printed image data is higher than the preset value. Based on the DCA feature fusion algorithm, multiple image feature sets are fused with feature data. Each image feature set corresponds to a visual angle. Based on each image feature set, multiple feature maps are generated. Multiple feature maps are combined to obtain multiple groups of feature maps. Each group of feature maps is convolved and activated by a convolutional neural network to generate multiple groups of fused feature data based on multi-visual analysis. The operating parameter data of the solder paste printer equipment under a preset production cycle is obtained. Through the recognition model, defects are identified for multiple sets of fused feature data respectively. Based on the multiple sets of defect information obtained, operation evaluation and optimization of the real-time parameter data are performed.

[0039] According to an embodiment of the present invention, in a preset production cycle, a multi-vision acquisition scheme is configured, and multiple sets of printing image data of the solder paste printer device during the printing process are obtained based on the multi-vision acquisition scheme, specifically: Set a preset production cycle and configure a multi-vision acquisition solution for the solder paste printer equipment; The multi-vision acquisition scheme includes a plurality of image acquisition setting angles, and each high-definition camera device corresponds to a setting angle; Based on the multi-vision acquisition solution, multiple sets of printing image data of the solder paste printer equipment during the printing process are obtained.

[0040] It should be noted that the preset production cycle is a preset time period or a production cycle of a preset task. The multi-vision acquisition scheme includes multiple set angles. For example, for the printing of solder paste printed products, the angles can be set from the top, side, front and back. Printing defects can be comprehensively analyzed at different angles.

[0041] According to an embodiment of the present invention, among the multiple groups of printing image data, each group of printing image data corresponds to an image acquisition setting angle.

[0042] According to an embodiment of the present invention, the recognition model is specifically: Build a CNN-based recognition model; The historical solder paste printing image data is normalized and imported into the recognition model for feature learning and recognition training. By setting the real label, the recognition output results are compared, and the recognition model parameters are updated through the back propagation algorithm to optimize the model. The recognition model training is repeated until the preset recognition standard is reached.

[0043] It should be noted that the real label is the real defect information and feature information of the historical solder paste printing image data.

[0044] According to an embodiment of the present invention, the feature extraction based on texture and contour is performed on each group of printed image data to obtain multiple image feature sets, and the principal component feature is extracted on the image feature set based on the PCA analysis method to obtain the main feature data. The solder paste printing recognition rate is calculated for the main feature data through the recognition model. If the recognition rate is lower than the preset value, the multi-vision acquisition scheme is adjusted and the image data is re-collected until the corresponding recognition rate of each group of printed image data is higher than the preset value, specifically: Through edge detection operators, feature extraction based on texture and contour is performed on each set of printed image data to form multiple image feature sets; Based on the PCA analysis method, the principal component features of each image feature set are extracted to obtain multiple groups of principal feature data; Through the recognition model, feature analysis and solder paste printing feature recognition are performed on each group of main feature data, interesting features and non-interest features are extracted and segmented, and the recognition rate based on the interesting features is calculated; Each set of main feature data corresponds to a recognition rate; If the recognition rate is lower than the preset value, the corresponding printed image data is marked, and at the same time, the setting angle of the corresponding vision is adjusted in the multi-vision acquisition scheme; Each set of main feature data is analyzed and the multi-vision acquisition scheme is adjusted, and the image data is re-collected until the corresponding recognition rate of each set of printed image data is higher than the preset value.

[0045] It should be noted that the edge detection operator may be a Sobel operator, a Prewitt operator, etc. The recognition model can be used for feature analysis and recognition. The recognition rate is generally the amount of data of the feature of interest divided by the total amount of data of the feature of interest and the feature of no interest. It can be understood here that in the multi-visual feature analysis, since the data collected from multiple angles may not meet the subsequent recognition requirements, it is necessary to analyze the recognition features, judge by the recognition rate, and adjust the corresponding multi-visual acquisition scheme so that the features collected at each visual angle meet the needs of subsequent analysis.

[0046] According to an embodiment of the present invention, each of the plurality of image feature sets corresponds to a set of printed image data.

[0047] According to an embodiment of the present invention, the DCA feature fusion algorithm is used to fuse feature data of multiple image feature sets, each image feature set corresponds to a visual angle, and multiple feature maps are generated based on each image feature set. The multiple feature maps are combined to obtain multiple groups of feature maps, and each group of feature maps is convolved and activated by a convolutional neural network to generate multiple groups of fused feature data based on multi-visual analysis, specifically: Import multiple image feature sets into the recognition model for feature learning and generate multiple feature maps; Combine multiple feature maps to obtain multiple groups of feature maps; Taking a set of feature maps as analysis units, based on the DCA feature fusion algorithm, setting preset weights, and performing convolution and activation operations on the set of feature maps through a convolutional neural network and the preset weights to generate multiple category activation maps; Perform spatial weighted fusion on multiple category activation maps and obtain a set of fused feature data; Multiple groups of fused feature data are obtained based on multiple groups of feature maps.

[0048] It should be noted that the visual angle is the set angle in the multi-vision acquisition scheme. In the feature data fusion of multiple image feature sets, the previous process includes the adjustment of the multi-vision acquisition party and the re-acquisition of image data to obtain multiple image feature sets that meet expectations.

[0049] Each feature map corresponds to an image feature set, that is, each feature map corresponds to a set acquisition angle. Among multiple sets of feature maps, each set of feature maps represents a different angle combination. The specific combination can be set by the user, or each angle combination can be combined, analyzed and fused. Convolutional neural network is CNN.

[0050] According to an embodiment of the present invention, the operation parameter data of the solder paste printer equipment under a preset production cycle is obtained, and defect recognition is performed on multiple sets of fused feature data respectively through the recognition model, and operation evaluation and optimization of the real-time parameter data are performed based on the obtained multiple sets of defect information, specifically: Obtain the operating parameter data of the solder paste printer equipment under a preset production cycle; Import multiple groups of fused feature data into the recognition model for defect recognition, and generate multiple groups of defect information based on the recognition results; Based on the operating parameter data, multiple parameter optimization configurations are generated in combination with multiple sets of defect information; Real-time parameter tuning and evaluation of solder paste printer equipment is performed based on multiple parameter optimization configurations.

[0051] It should be noted that in the real-time parameter tuning and evaluation of the solder paste printer equipment according to the multiple parameter optimization configurations, it is determined whether there are conflicting parameter adjustments based on the multiple parameter optimization configurations, and adjustments are made according to the actual operating conditions. The equipment adjustments include the adjustment of printing speed, printing pressure, scraper angle, solder paste viscosity, solder paste volume, and ambient temperature. The defect information includes information such as defect characteristics, defect types, and defect occurrence time nodes.

[0052] It is worth mentioning that the traditional technology lacks multi-visual fusion analysis and is often based on simple visual images and single acquisition angle analysis, which is not comprehensive enough for feature recognition of solder paste printing. The present invention is based on multi-visual analysis and can effectively perform defect analysis from multi-dimensional visual feature data. Combined with the DCA fusion feature algorithm, it can perform different types of defect analysis on multi-visual features.

[0053] Defects include printing offset, insufficient solder paste, solder paste thickness not meeting expectations, missing, broken, misaligned printing dots, etc.

[0054] The third aspect of the present invention also provides a computer-readable storage medium, which includes a printing machine equipment operation optimization program based on multi-visual analysis. When the printing machine equipment operation optimization program based on multi-visual analysis is executed by a processor, the steps of the printing machine equipment operation optimization process based on multi-visual analysis as described in any one of the above items are implemented.

[0055] The present invention discloses a printing machine equipment operation optimization process and system based on multi-visual analysis. Within a preset production cycle, a multi-visual acquisition scheme is used to obtain multiple groups of image data in the printing process. The main feature data is obtained by texture and contour feature extraction and PCA analysis method, and the solder paste printing recognition rate is calculated using a recognition model. The recognition rate is judged and the acquisition scheme is adjusted. The DCA feature fusion algorithm is used to fuse multiple image feature sets to generate multiple groups of feature maps, which are processed by convolutional neural networks to obtain multiple groups of fused feature data. At the same time, the equipment operation parameters are obtained, and the fused feature data are identified for defects through the recognition model. The real-time equipment operation parameters are evaluated and optimized based on the defect information to ensure printing quality and production efficiency. Through the present invention, equipment operation analysis and printing defect evaluation can be effectively performed from multi-visual features, the intelligent operation of equipment can be realized, and the production efficiency can be improved.

[0056] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. 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, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0057] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units; 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 present embodiment.

[0058] In addition, all functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may be separately used as a unit, or two or more units may be integrated into one unit; the above-mentioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0059] A person of ordinary skill in the art can understand that: all or part of the steps of implementing the above method embodiment can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above method embodiment; and the aforementioned storage medium includes: a mobile storage device, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and other media that can store program codes.

[0060] Alternatively, if the above-mentioned integrated unit of the present invention is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present invention can be essentially or partly reflected in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROM, RAM, magnetic disks or optical disks.

[0061] The above description is only a specific implementation mode 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 changes or substitutions within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A printing press equipment operation optimization process based on multi-visual analysis, characterized in that: include: Under a preset production cycle, a multi-vision acquisition solution is configured, and multiple sets of printing image data of the solder paste printer during the printing process are obtained based on the multi-vision acquisition solution; Perform texture and contour-based feature extraction on each set of printed image data to obtain multiple image feature sets. Perform principal component feature extraction on the image feature sets based on the PCA analysis method to obtain principal feature data. Use the recognition model to calculate the solder paste printing recognition rate on the principal feature data. If the recognition rate is lower than the preset value, adjust the multi-vision acquisition scheme and re-acquire the image data until the corresponding recognition rate of each set of printed image data is higher than the preset value. Based on the DCA feature fusion algorithm, multiple image feature sets are fused with feature data. Each image feature set corresponds to a visual angle. Based on each image feature set, multiple feature maps are generated. Multiple feature maps are combined to obtain multiple groups of feature maps. Each group of feature maps is convolved and activated by a convolutional neural network to generate multiple groups of fused feature data based on multi-visual analysis. The operating parameter data of the solder paste printer equipment under a preset production cycle is obtained. Through the recognition model, defects are identified for multiple sets of fused feature data respectively. Based on the multiple sets of defect information obtained, operation evaluation and optimization of the real-time parameter data are performed.

2. The process of optimizing the operation of a printing press based on multi-visual analysis according to claim 1, characterized in that: In a preset production cycle, a multi-vision acquisition scheme is configured, and multiple sets of printing image data of the solder paste printer during the printing process are obtained based on the multi-vision acquisition scheme, specifically: Set a preset production cycle and configure a multi-vision acquisition solution for the solder paste printer equipment; The multi-vision acquisition scheme includes a plurality of image acquisition setting angles, and each high-definition camera device corresponds to a setting angle; Based on the multi-vision acquisition solution, multiple sets of printing image data of the solder paste printer equipment during the printing process are obtained.

3. The printing press equipment operation optimization process based on multi-visual analysis according to claim 2 is characterized in that: Among the multiple groups of printing image data, each group of printing image data corresponds to an image acquisition setting angle.

4. The process of optimizing the operation of a printing press based on multi-visual analysis according to claim 3 is characterized in that: The identification model is specifically: Build a CNN-based recognition model; The historical solder paste printing image data is normalized and imported into the recognition model for feature learning and recognition training. By setting the real label, the recognition output results are compared, and the recognition model parameters are updated through the back propagation algorithm to optimize the model. The recognition model training is repeated until the preset recognition standard is reached.

5. The printing press equipment operation optimization process based on multi-visual analysis according to claim 1 is characterized in that: The method performs texture and contour-based feature extraction on each set of printed image data to obtain multiple image feature sets, performs principal component feature extraction on the image feature sets based on the PCA analysis method to obtain principal feature data, and calculates the solder paste printing recognition rate on the principal feature data through the recognition model. If the recognition rate is lower than a preset value, the multi-vision acquisition scheme is adjusted and the image data is re-acquired until the corresponding recognition rate of each set of printed image data is higher than a preset value, specifically: Through edge detection operators, feature extraction based on texture and contour is performed on each set of printed image data to form multiple image feature sets; Based on the PCA analysis method, the principal component features of each image feature set are extracted to obtain multiple groups of principal feature data; Through the recognition model, feature analysis and solder paste printing feature recognition are performed on each group of main feature data, interesting features and non-interest features are extracted and segmented, and the recognition rate based on the interesting features is calculated; Each set of main feature data corresponds to a recognition rate; If the recognition rate is lower than the preset value, the corresponding printed image data is marked, and at the same time, the setting angle of the corresponding vision is adjusted in the multi-vision acquisition scheme; Each set of main feature data is analyzed and the multi-vision acquisition scheme is adjusted, and the image data is re-collected until the corresponding recognition rate of each set of printed image data is higher than the preset value.

6. The printing press equipment operation optimization process based on multi-visual analysis according to claim 1 is characterized in that: Among the plurality of image feature sets, each image feature set corresponds to a set of printing image data.

7. The process of optimizing the operation of a printing press based on multi-visual analysis according to claim 1, characterized in that: Based on the DCA feature fusion algorithm, multiple image feature sets are fused with feature data, each image feature set corresponds to a visual angle, multiple feature maps are generated based on each image feature set, and multiple feature maps are combined to obtain multiple groups of feature maps. Each group of feature maps is convoluted and activated by a convolutional neural network to generate multiple groups of fused feature data based on multi-visual analysis, specifically: Import multiple image feature sets into the recognition model for feature learning and generate multiple feature maps; Combine multiple feature maps to obtain multiple groups of feature maps; Taking a set of feature maps as analysis units, based on the DCA feature fusion algorithm, setting preset weights, and performing convolution and activation operations on the set of feature maps through a convolutional neural network and the preset weights to generate multiple category activation maps; Perform spatial weighted fusion on multiple category activation maps and obtain a set of fused feature data; Multiple groups of fused feature data are obtained based on multiple groups of feature maps.

8. The process of optimizing the operation of a printing press based on multi-visual analysis according to claim 1, characterized in that: The operation parameter data of the solder paste printer equipment under a preset production cycle is obtained, and defects are respectively identified on multiple sets of fused feature data through the recognition model, and operation evaluation and optimization of the real-time parameter data are performed based on the obtained multiple sets of defect information, specifically: Obtain the operating parameter data of the solder paste printer equipment under a preset production cycle; Import multiple groups of fused feature data into the recognition model for defect recognition, and generate multiple groups of defect information based on the recognition results; Based on the operating parameter data, multiple parameter optimization configurations are generated in combination with multiple sets of defect information; Real-time parameter tuning and evaluation of solder paste printer equipment is performed based on multiple parameter optimization configurations.

9. A printing press equipment operation optimization system based on multi-visual analysis, characterized in that: The system includes: a memory and a processor, wherein the memory includes a printing press device operation optimization program based on multi-visual analysis, and when the printing press device operation optimization program based on multi-visual analysis is executed by the processor, the following steps are implemented: Under a preset production cycle, a multi-vision acquisition solution is configured, and multiple sets of printing image data of the solder paste printer during the printing process are obtained based on the multi-vision acquisition solution; Perform texture and contour-based feature extraction on each set of printed image data to obtain multiple image feature sets. Perform principal component feature extraction on the image feature sets based on the PCA analysis method to obtain principal feature data. Use the recognition model to calculate the solder paste printing recognition rate on the principal feature data. If the recognition rate is lower than the preset value, adjust the multi-vision acquisition scheme and re-acquire the image data until the corresponding recognition rate of each set of printed image data is higher than the preset value. Based on the DCA feature fusion algorithm, multiple image feature sets are fused with feature data. Each image feature set corresponds to a visual angle. Based on each image feature set, multiple feature maps are generated. Multiple feature maps are combined to obtain multiple groups of feature maps. Each group of feature maps is convolved and activated by a convolutional neural network to generate multiple groups of fused feature data based on multi-visual analysis. The operating parameter data of the solder paste printer equipment under a preset production cycle is obtained. Through the recognition model, defects are identified for multiple sets of fused feature data respectively. Based on the multiple sets of defect information obtained, operation evaluation and optimization of the real-time parameter data are performed.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a printing device operation optimization program based on multi-visual analysis. When the printing device operation optimization program based on multi-visual analysis is executed by a processor, the steps of the printing device operation optimization process based on multi-visual analysis as described in any one of claims 1 to 8 are implemented.

Images