Printing production information management method and system

Through machine learning models and sensor technology, the problems of inaccurate data and difficulty in regulation in traditional printing production information management are solved, accurate matching of printed product feature data and real-time control of equipment parameters are achieved, the quality and consistency of printed products are improved, and the quality and consistency of printed products are reduced. production costs.

CN118342892BActive Publication Date: 2024-11-15HUNAN HUACHENG PACKAGING & PRINTING TECHNOLOGY CO LTD
View PDF 11 Cites 0 Cited by

Patent Information

Application Number
CN202410544078.3
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-05-06
Publication Date
2024-11-15
Estimated Expiration
2044-05-06

AI Technical Summary

Technical Problem

Traditional printing and production information management methods rely on manual operations, resulting in inaccurate data, untimely information transmission, low management efficiency, resulting in color deviations in print products and difficulty in equipment regulation, affecting the consistency of print products.

Method used

The machine learning model is used to combine photoelectric sensors and camera equipment, and the printing equipment is accurately matched and adjusted by collecting historical printed material feature data and environmental parameters, building a database, and real-time control of printing equipment parameters, so as to achieve accurate matching and adjustment of printed material feature data.

Benefits of technology

It improves the quality and consistency of printed materials, reduces the time and cost of testing and adjustment, improves production efficiency and economic benefits, and realizes real-time regulation and precise control of printing equipment.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118342892B_ABST
    Figure CN118342892B_ABST
Patent Text Reader

Abstract

The present invention belongs to the field of printing control technology, and discloses a printing production information management method and system, wherein the method includes: S1, collecting the original image and the historical equipment parameters and historical environmental parameters of the historical printed product corresponding to the original image as the first historical printing data, and obtaining the historical printed product feature data by preprocessing the historical printed product; S2, constructing a database according to the first historical printing data and the historical printed product feature data. The present invention obtains the feature extraction area by threshold segmentation of the historical printed product, reduces the amount of calculation for obtaining the feature data of the historical printed product, makes the segmentation of the historical printed product more accurate, and improves the quality and speed of preprocessing the historical printed product; uses a machine learning model to predict the equipment parameters in real time, realizes the instant regulation of the printing equipment, and meets different printing needs and environmental conditions; accurately adjusts the equipment parameters to make the quality of the printed product more stable and improve the consistency.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of printing control technology, and more specifically, to a printing production information management method and system. Background Art

[0002] With the intensification of market competition and the diversification of customer needs, printing companies need to continuously improve production efficiency, optimize product quality, and reduce production costs; however, traditional printing production information management methods often rely on manual operations and paper records, and there are problems such as inaccurate data, untimely information transmission, and low management efficiency.

[0003] During the printing process of existing printing equipment, when the same printed product is printed for the second time, the staff often manually records the printing equipment parameters at that time. The manually recorded content is easy to be improperly saved, and when the second printing or subsequent multiple printings are carried out, it is easy to produce large color deviations. In addition, during the printing process, it is inconvenient for the staff to discover the printing color deviation problem in time, which is easy to cause the production of multiple defective prints, and it is inconvenient to adjust the printing equipment parameters in time to make the printed products printed at different times consistent.

[0004] In view of this, the present invention proposes a printing production information management method and system to solve the above problems. Summary of the invention

[0005] In order to overcome the above-mentioned defects of the prior art and to achieve the above-mentioned purpose, the present invention provides the following technical solution: a printing production information management method, comprising:

[0006] S1, collecting an original image and historical equipment parameters and historical environment parameters of a historical printed product corresponding to the original image as first historical printing data, and obtaining historical printed product feature data by preprocessing the historical printed product;

[0007] S2. construct a database based on the first historical printing data and the historical printed product feature data, and construct and train a machine learning model through the database;

[0008] S3, collect current environmental parameters and obtain historical printed product feature data to input into the machine learning model to obtain the corresponding current device parameters;

[0009] S4. Perform preliminary adjustment on the printing device according to the current device parameters to obtain the current printed product, obtain the feature data difference between the current printed product and the historical printed product, and adjust the device parameters according to the feature data difference until the feature data difference is less than the maximum feature data difference.

[0010] Furthermore, the step of collecting the original image and the historical equipment parameters and historical environment parameters of the historical printed product corresponding to the original image as the first historical printing data includes:

[0011] S10, collecting historical equipment parameters, the historical equipment parameters including printing speed, printing pressure and printing inkjet volume;

[0012] By installing a photoelectric sensor in the transmission system of the printing device, the printing speed is determined by detecting the rotation of the transmission shaft or the printing roller to obtain the printing speed in the historical device parameters;

[0013] Printing pressure is the sum of upper pressure and lower pressure. Upper pressure is the pressure applied on the printing head, printing roller or printing mold during printing, and lower pressure is the pressure applied on the printing material or printing bed during printing. Both upper pressure and lower pressure are obtained by the pressure system on the printing equipment.

[0014] The printing inkjet volume includes the jetting frequency and the distance between the nozzle and the printing material. The jetting frequency is obtained by the jetting frequency sensor, and the distance between the nozzle and the printing material is obtained by the distance sensor. The formula for the printing inkjet volume is: ,

[0015] in, is the printing inkjet volume, is a constant, is the injection frequency, is the distance between the nozzle and the printing material;

[0016] S11. Collect historical environmental parameters, which include ambient temperature, ambient humidity and air dust density; the ambient temperature is obtained by a temperature sensor, the ambient humidity is obtained by a humidity sensor, and the air dust density is obtained by a laser particle sensor.

[0017] Furthermore, the step of obtaining characteristic data of historical printed products by preprocessing the historical printed products includes:

[0018] S12, photographing the printed product of the original image through the printing device by using a camera device to obtain an initial printed image, and correcting the initial printed image to obtain a historical printed product;

[0019] S13, segmenting the historical printed matter using an improved adaptive threshold algorithm to obtain a feature extraction region of the historical printed matter;

[0020] S14, acquiring historical printed product feature data of the feature extraction area, the historical printed product feature data including color distribution value, color gradient value, brightness distribution value and ink dot density value;

[0021] The method of correcting the initial printed image includes:

[0022] S121, cutting and adjusting the initial printed image, cutting or enlarging the initial printed image to the same specifications as the original image and making the pattern content in the initial printed image correspond to the pattern content in the original image;

[0023] S122, obtaining an influence area of ​​the initial printed image, dividing the image area of ​​the original image and the initial printed image according to the influence area, and obtaining an image area division influence map;

[0024] S123, obtaining the influence coefficient corresponding to the influence area in the image area division influence map; performing color correction on the initial printed image according to the influence coefficient corresponding to the influence area;

[0025] The image area division method includes:

[0026] An original image and an initial printing image are set, a virtual grid is preset according to the size specification of the original image, the original image is a pure color image, the gray value of the pure color image is obtained, and the same gray value in the preset virtual grid is defined as an influence area, so as to perform image area division until an image area division influence map is obtained;

[0027] The method of obtaining the influence coefficient includes:

[0028] The color difference between the initial printed image and the original image of the same affected area is obtained, and the influence coefficient of the affected area is obtained according to the color difference.

[0029] Furthermore, the method of obtaining the color distribution value includes:

[0030] S140, using K clustering algorithm to obtain feature extraction area color clusters, each color cluster forms a The color vector of is used to count the occurrence frequencies of different color clusters in the feature extraction area, and the occurrence frequencies of different color clusters are weighted averaged to obtain the color distribution value;

[0031] The method of obtaining the color gradient value includes:

[0032] S141, dividing the feature extraction area into color regions according to different color clusters to obtain color division regions, calculating gradient values ​​between different color division regions, and performing weighted average of the gradient values ​​between different color division regions to obtain a color gradient value;

[0033] The method of obtaining the brightness distribution value includes:

[0034] S142, calculating the grayscale histogram of the feature extraction area, counting the number of pixels at different brightness levels in the grayscale histogram, and performing weighted average on the number of pixels at different brightness levels to obtain a brightness distribution value;

[0035] The method of obtaining the ink dot density value includes:

[0036] S143, counting the number of pixels in the feature extraction area as the total number of the feature extraction area, counting the number of ink dot pixels in the feature extraction area as the number of ink dots, and taking the ratio of the number of ink dots to the total number of the feature extraction area as the ink dot density value.

[0037] Furthermore, the step of segmenting the historical printed matter using the improved adaptive threshold algorithm to obtain the feature extraction area of ​​the historical printed matter includes:

[0038] S130, improving the adaptive threshold algorithm by using a particle swarm algorithm and grayscale entropy to obtain a segmentation threshold for the feature extraction area;

[0039] S1301, initializing a particle swarm, each particle represents a threshold vector, the length of the threshold vector matches the number of blocks to be segmented in the historical printed product, and each element in the threshold vector represents a threshold of a block to be segmented;

[0040] S1302, segmenting the historical printed matter according to the threshold vector corresponding to each particle, calculating the inter-class variance of the historical printed matter after segmentation, and selecting the highest inter-class variance to calculate the fitness of the segmentation threshold;

[0041] S1303, update the position and speed of each particle, find the individual optimal position and the global optimal position corresponding to the fitness,

[0042] The formula for updating the particle position is:

[0043] ;

[0044] in, It is a particle In time location, It is a particle In time location, It is a particle In time speed;

[0045] The formula for updating particle velocity is:

[0046] ;

[0047] in, is the inertia weight, and are the first acceleration factor and the second acceleration factor, and are the first random number and the second random number respectively, It is a particle The individual optimal position of is the global optimal position of the entire particle swarm, It is a particle In time speed;

[0048] S1304, check whether the termination condition is met, if the maximum number of iterations is reached or the fitness function value converges to a preset threshold;

[0049] S1305, selecting a threshold vector with the best fitness from all particles as the final segmentation threshold;

[0050] S1306 , using the optimal threshold vector to segment the historical printed product to obtain a feature extraction region of the historical printed product.

[0051] Furthermore, the inter-class variance of the historical printed materials is:

[0052] ;

[0053] in, is the between-class variance, is the number of blocks to be split, No. The weight of the blocks to be split, It is The grayscale mean of the blocks to be segmented, is the grayscale mean of the entire historical print;

[0054] The fitness of the segmentation threshold is:

[0055] ;

[0056] in, It is The probability of gray levels, is the total number of gray levels.

[0057] Furthermore, the step of constructing a database according to the first historical printing data and the historical printed product feature data, and constructing and training a machine learning model through the database includes:

[0058] S20, taking the first historical printing data and the historical printed product characteristic data corresponding to the same original image and the historical printed product as a data group, wherein the data in the data group are mutually related, and the data group is named after the original image; obtaining n data groups within the historical period, and constructing the n data groups into a database;

[0059] S21, combining historical printed product feature data and historical printing device parameters and historical environment parameters corresponding to the historical printed product feature data into a first sample set, and numerically numbering the historical printing device parameters;

[0060] S22, constructing the historical printed material feature data and historical environmental parameters of each sample into a feature vector as the input of the machine learning model;

[0061] S23, use keras to define the input layer and output layer and build a machine learning framework, with the feature vector as the input layer and the historical printing equipment parameter number as the output layer;

[0062] S24, dividing the first sample set into a training set and a test set, constructing a machine learning model, training the machine learning model to obtain the machine learning model, testing the machine learning model using the test set, and outputting a machine learning model that meets a preset accuracy, wherein the machine learning model is one of a logistic regression model, a naive Bayes model, or a support vector machine model;

[0063] The calculation formula for the preset accuracy is: ,in, is the total number of samples in the first sample set, The minimum accuracy is preset for the number of samples that are correctly predicted. When the accuracy reaches the minimum accuracy, the training of the machine learning model can be stopped.

[0064] Furthermore, the step of collecting current environmental parameters and obtaining historical printed product feature data and inputting them into a machine learning model to obtain corresponding current device parameters includes:

[0065] S30, searching the database for the original image corresponding to the to-be-printed product by name to obtain a data group corresponding to the to-be-printed product;

[0066] S31, obtaining historical printed product characteristic data corresponding to the product to be printed according to the data group;

[0067] S32. Input historical printed feature data and current environmental parameters into the machine learning model, and the machine learning model outputs the current device parameter number.

[0068] Further, the step of preliminarily adjusting the printing device according to the current device parameters to obtain the current printed product, obtaining the feature data difference between the current printed product and the historical printed product, and adjusting the device parameters according to the feature data difference until the feature data difference is less than the maximum feature data difference includes:

[0069] S40, applying the current device parameters to the printing device, and performing preliminary regulation on the printing device;

[0070] S41, the printing device prints the original image using the current device parameters to obtain a current printed product;

[0071] S42, photographing and preprocessing the current printed product by a camera device to obtain feature data of the current printed product, and obtaining a feature data difference between the feature data of the current printed product and feature data of historical printed products;

[0072] S43, inputting the feature data difference and the current environmental parameters into the machine learning model, and the machine learning model obtains the difference device parameters;

[0073] S44, adding the difference device parameter to the current device parameter and adjusting the printing device until the current printed product feature data is less than the maximum feature data difference.

[0074] A printing production information management method and system, comprising:

[0075] A data collection module collects the original image and the historical equipment parameters and historical environment parameters of the historical printed product corresponding to the original image as the first historical printing data, and obtains the historical printed product characteristic data by pre-processing the historical printed product;

[0076] A database construction module is used to construct a database according to the first historical printing data and the historical printed product feature data, and a machine learning model is constructed and trained through the database;

[0077] The real-time acquisition output module collects current environmental parameters and obtains historical printed product feature data to input into the machine learning model to obtain the corresponding current device parameters;

[0078] The control and optimization module performs preliminary control on the printing equipment according to the current equipment parameters, obtains the current printed product, obtains the feature data difference between the current printed product and the historical printed product, and controls the equipment parameters according to the feature data difference until the feature data difference is less than the maximum feature data difference.

[0079] The technical effects and advantages of a printing production information management method and system of the present invention are as follows:

[0080] 1. By performing threshold segmentation on historical printed materials to obtain feature extraction areas, the amount of calculation required to obtain feature data of historical printed materials is reduced, making the segmentation of historical printed materials more accurate and improving the quality and speed of preprocessing of historical printed materials.

[0081] 2. By predicting and optimizing printing equipment parameters, the printing process can be better controlled and printing quality and efficiency can be improved; by instantly collecting current environmental parameters and characteristic data of the printed product, and predicting the current equipment parameters based on the machine learning model, real-time regulation of the printing equipment can be achieved, so as to better adapt to different printing needs and environmental conditions. By adjusting the equipment parameters according to the difference in characteristic data, precise regulation of the printing equipment can be achieved, so that the characteristic data of the printed product is closer to the historical printed product, thereby improving the quality and consistency of the printed product. Compared with traditional testing and adjustment methods, this method can reduce the time and cost of testing and adjustment, and improve production efficiency and economic benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0082] Figure 1 A schematic diagram of a process of a printing production information management method of the present invention;

[0083] Figure 2 It is a schematic diagram of the regulation cycle of the present invention;

[0084] Figure 3 It is a structural schematic diagram of a printing production information management system of the present invention. DETAILED DESCRIPTION

[0085] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0086] Example 1

[0087] See also Figure 1 As shown, this embodiment provides a printing production information management method, including:

[0088] S1, collecting an original image and historical equipment parameters and historical environment parameters of a historical printed product corresponding to the original image as first historical printing data, and obtaining historical printed product feature data by preprocessing the historical printed product;

[0089] S10, collecting historical equipment parameters, the historical equipment parameters including printing speed, printing pressure and printing inkjet volume;

[0090] By installing a photoelectric sensor in the transmission system of the printing device, the printing speed is determined by detecting the rotation of the transmission shaft or the printing roller to obtain the printing speed in the historical device parameters;

[0091] Printing pressure is the sum of upper pressure and lower pressure. Upper pressure is the pressure applied on the printing head, printing roller or printing mold during printing, and lower pressure is the pressure applied on the printing material or printing bed during printing. Both upper pressure and lower pressure are obtained by the pressure system on the printing equipment.

[0092] The printing inkjet volume includes the jetting frequency and the distance between the nozzle and the printing material. The jetting frequency is obtained by the jetting frequency sensor, and the distance between the nozzle and the printing material is obtained by the distance sensor. The formula for the printing inkjet volume is: ,

[0093] in, is the printing inkjet volume, is a constant, is the injection frequency, is the distance between the nozzle and the printing material,

[0094] It should be noted that experimental tests, The printing inkjet volume under different jetting frequencies and distances between the nozzle and the printing material can be measured experimentally, and the constant can be determined based on the experimental results. This method needs to be carried out in a laboratory or actual printing scene, and the influence of various factors on the experimental results should be taken into account; data fitting, if there are some relevant data, the constant can be determined by using the data fitting method The value of , Distance between printhead and printing material And the corresponding printing inkjet volume By fitting the data, we can get an optimal value, so that the model fits the actual data best; theoretical analysis, in some specific cases, theoretical analysis can be used to determine the constant For example, the ink flow behavior is analyzed based on fluid mechanics theory, and the printing inkjet volume and jet frequency are derived by combining the design parameters of the inkjet head and the characteristics of the printing material. , Distance between printhead and printing material The relationship between The value of .

[0095] S11, collecting historical environmental parameters, the historical environmental parameters include ambient temperature, ambient humidity and air dust density; the ambient temperature is obtained by a temperature sensor, the ambient humidity is obtained by a humidity sensor, and the air dust density is obtained by a laser particle sensor;

[0096] The above content monitors and records the operating status and environmental conditions of printing equipment in real time through various sensors, and can obtain historical equipment parameter and environmental parameter data in a timely manner, thereby improving the accuracy and reliability of the data. The application of sensors realizes the automatic collection of data, reduces the time and cost of manual operation, and can also avoid the influence of human factors on data collection, thereby improving the efficiency and accuracy of data collection. The application of sensors helps to realize the intelligence and automation of the printing process, and improve production efficiency and product quality.

[0097] The steps of obtaining characteristic data of historical printed products by preprocessing the historical printed products include:

[0098] S12, photographing the printed product produced by the printing device using the original image by means of a camera (the light source of the camera is fixed) to obtain an initial printed image, and performing correction processing on the initial printed image to obtain a historical printed product.

[0099] The methods for correcting the initial printed image include:

[0100] S121, cutting and adjusting the initial printed image, cutting or enlarging the initial printed image to the same specifications as the original image and making the pattern content in the initial printed image correspond to the pattern content in the original image;

[0101] S122, obtaining an influence area of ​​the initial printed image, dividing the image area of ​​the original image and the initial printed image according to the influence area, and obtaining an image area division influence map;

[0102] S123, obtaining the influence coefficient corresponding to the influence area in the image area division influence map; performing color correction on the initial printed image according to the influence coefficient corresponding to the influence area;

[0103] The above contents can more accurately match the pattern content and color of the original image through preprocessing of the initial printed image, including cropping adjustment and color correction, which helps to improve the quality and accuracy of the printed product. Compared with the traditional simple color correction, the effect is more significant; by obtaining the influence coefficient of the affected area and correcting the color of the initial printed image according to the influence coefficient, the color deviation caused by other environmental factors such as the light of the camera equipment can be reduced, which can improve the consistency and stability of the printed product.

[0104] The method of image area division is as follows: an original image and an initial printing image are set, a virtual grid is preset according to the size specification of the original image, the original image is a pure color image, the gray value of the pure color image is obtained, and the same gray value in the preset virtual grid is defined as an influence area, so as to perform image area division until an image area division influence map is obtained, wherein the size of a single grid in the preset virtual grid is a single color in each grid;

[0105] The method of obtaining the influence coefficient is as follows: obtaining the color difference between the initial printed image and the original image of the same influence area, making the initial printed image with the same gray value correspond to the preset virtual grid, adjusting the original image to pure color images of different colors (which can be the four basic printing pure colors of CMYK respectively), normalizing the color difference so that it is within a specific value range, which can be set by the staff, and obtaining the influence coefficient corresponding to the influence area, so as to obtain the influence coefficient of different colors in the influence area;

[0106] By using a preset virtual grid, the original image and the initial printed image are divided into regions so that the grayscale value in each region is the same. This division method can more accurately capture the characteristics and changes of different regions in the image, which helps to more accurately analyze the influencing factors in the printing process; the same grayscale value in the preset virtual grid is defined as an impact region, which can effectively locate different impact regions in the image, help to accurately identify possible problem areas in the printing process, and provide an accurate reference for subsequent color correction. Using a virtual grid to divide the impact region can make the image processing process more efficient; the virtual grid can divide the image into relatively uniform regions, simplifying the complexity of image processing, thereby improving the speed and efficiency of processing; according to the impact coefficient corresponding to the impact region, the color of the initial printed image is corrected. This correction method can make targeted color adjustments according to the size and degree of the impact region, making the correction more accurate, and can effectively improve the color quality and consistency of the printed product.

[0107] S13, segmenting the historical printed matter using an improved adaptive threshold algorithm to obtain a feature extraction region of the historical printed matter;

[0108] S14, acquiring historical printed product feature data of the feature extraction area, wherein the historical printed product feature data includes color distribution value, color gradient value, brightness distribution value and ink dot density value.

[0109] The methods for obtaining color distribution values ​​include:

[0110] S140, using K clustering algorithm to obtain feature extraction area color clusters, each color cluster forms a The color vector of is used to count the occurrence frequencies of different color clusters in the feature extraction area, and the occurrence frequencies of different color clusters are weighted averaged to obtain the color distribution value;

[0111] The methods for obtaining color gradient values ​​include:

[0112] S141, dividing the feature extraction area into color regions according to different color clusters to obtain color division regions, calculating gradient values ​​between different color division regions, and performing weighted average of the gradient values ​​between different color division regions to obtain a color gradient value;

[0113] The ways to obtain brightness distribution values ​​include:

[0114] S142, calculating the grayscale histogram of the feature extraction area, counting the number of pixels at different brightness levels in the grayscale histogram, and performing weighted average on the number of pixels at different brightness levels to obtain a brightness distribution value;

[0115] The methods for obtaining ink dot density values ​​include:

[0116] S143, counting the number of pixels in the feature extraction area as the total number of the feature extraction area, counting the number of ink dot pixels in the feature extraction area as the number of ink dots, and taking the ratio of the number of ink dots to the total number of the feature extraction area as the ink dot density value.

[0117] The steps of segmenting the historical printed matter using the improved adaptive threshold algorithm to obtain the feature extraction area of ​​the historical printed matter include:

[0118] S130, improving the adaptive threshold algorithm by using a particle swarm algorithm and grayscale entropy to obtain a segmentation threshold for the feature extraction area;

[0119] S1301, initializing a particle swarm, each particle represents a threshold vector, the length of the threshold vector matches the number of blocks to be segmented in the historical printed product, and each element in the threshold vector represents a threshold of a block to be segmented;

[0120] S1302, segmenting the historical printed matter according to the threshold vector corresponding to each particle, calculating the inter-class variance of the historical printed matter after segmentation, and selecting the highest inter-class variance to calculate the fitness of the segmentation threshold;

[0121] The between-class variance of historical prints is:

[0122] ;

[0123] in, is the between-class variance, is the number of blocks to be split, It is The weight of the blocks to be split, It is The grayscale mean of the blocks to be segmented, is the grayscale mean of the entire historical print;

[0124] The fitness of the segmentation threshold is:

[0125] ;

[0126] in, It is The probability of gray levels, is the total number of gray levels.

[0127] S1303, update the position and speed of each particle, find the individual optimal position and the global optimal position corresponding to the fitness,

[0128] The formula for updating the particle position is:

[0129] ;

[0130] in, It is a particle In time location, It is a particle In time location, It is a particle In time speed;

[0131] The formula for updating particle velocity is:

[0132] ;

[0133] in, is the inertia weight, and are the first acceleration factor and the second acceleration factor, Used to control particles to adjust toward the individual optimal solution. Used to control the particles to adjust towards the global optimal solution. and are the first random number and the second random number respectively, and yes A random number uniformly distributed in the interval, Used to introduce the randomness of individual experience, Used to introduce randomness of group experience, It is a particle The individual optimal position of is the global optimal position of the entire particle swarm, It is a particle In time speed;

[0134] S1304, check whether the termination condition is met, if the maximum number of iterations is reached or the fitness function value converges to a preset threshold;

[0135] S1305, selecting a threshold vector with the best fitness from all particles as the final segmentation threshold;

[0136] S1306, segmenting the historical printed matter using the optimal threshold vector to obtain a feature extraction region of the historical printed matter;

[0137] In addition, the segmentation threshold may be post-processed as needed to remove small areas or merge adjacent areas to obtain a more accurate segmentation threshold.

[0138] The above content improves the adaptive threshold algorithm by using particle swarm algorithm and grayscale entropy, which can more accurately determine the segmentation threshold of historical prints, overcome the problem of poor adaptability of traditional threshold algorithm to different historical prints, and improve the accuracy and stability of segmentation; by segmenting historical prints through the improved adaptive threshold algorithm, more accurate feature extraction areas are obtained; these areas can better reflect the local features and detail information of historical prints, which is helpful for subsequent feature analysis and processing; the adaptive threshold algorithm improved by particle swarm algorithm and grayscale entropy can automatically determine the optimal segmentation threshold without manual intervention, which improves the degree of automation of the preprocessing process of historical prints and reduces labor and time costs.

[0139] The improved adaptive threshold algorithm can quickly and accurately segment historical prints, improving the efficiency of historical print preprocessing. After obtaining the segmentation threshold, post-processing can be performed as needed to remove small areas or merge adjacent areas to further improve the accuracy and stability of the segmentation threshold. This post-processing capability makes the final feature extraction more accurate and reliable.

[0140] By performing threshold segmentation on historical printed materials to obtain feature extraction areas, the amount of calculation required to obtain feature data of historical printed materials is reduced, the segmentation of historical printed materials has higher accuracy, and the quality and speed of preprocessing of historical printed materials are improved.

[0141] S2. Building a database based on the first historical printing data and the historical printed product feature data, and the steps of building and training a machine learning model through the database include:

[0142] S20, taking the first historical printing data and the historical printed product characteristic data corresponding to the same original image and the historical printed product as a data group, wherein the data in the data group are mutually related, and the data group is named after the original image; obtaining n data groups within the historical period, and constructing the n data groups into a database;

[0143] By building a database of the same original image and its corresponding historical printing data, historical data can be better managed and utilized; the construction of a comprehensive database makes the storage and retrieval of historical data more efficient and convenient, which is helpful for subsequent data analysis and model training.

[0144] S21, combining historical printed product feature data and historical printing device parameters and historical environment parameters corresponding to the historical printed product feature data into a first sample set, and numerically numbering the historical printing device parameters;

[0145] S22, constructing the historical printed material feature data and historical environmental parameters of each sample into a feature vector as the input of the machine learning model;

[0146] S23, use keras to define the input layer and output layer and build a machine learning framework, with the feature vector as the input layer and the historical printing equipment parameter number as the output layer;

[0147] S24, dividing the first sample set into a training set and a test set, constructing a machine learning model, training the machine learning model to obtain the machine learning model, testing the machine learning model using the test set, and outputting a machine learning model that meets a preset accuracy, wherein the machine learning model is one of a logistic regression model, a naive Bayes model, or a support vector machine model;

[0148] The calculation formula for the preset accuracy is: ,in, is the total number of samples in the first sample set, The minimum accuracy is preset for the number of samples that are correctly predicted. When the accuracy reaches the minimum accuracy, the training of the machine learning model can be stopped.

[0149] The above content builds a more accurate and efficient machine learning model through historical printing equipment parameters, historical environmental parameters and historical printed product feature data, which helps to improve the quality and efficiency of the printing process.

[0150] In this embodiment, feature extraction areas are obtained by performing threshold segmentation on historical printed materials, thereby reducing the amount of calculation for obtaining feature data of historical printed materials, making the segmentation of historical printed materials more accurate, and improving the quality and speed of preprocessing of historical printed materials. By predicting and optimizing printing equipment parameters, the printing process can be better controlled, and the printing quality and efficiency can be improved. The automated parameter optimization process can reduce the impact of human errors and subjective factors on printing quality. By predicting parameters through machine learning models, the time and cost of testing and adjustment can be reduced, thereby improving production efficiency.

[0151] Example 2

[0152] See also Figure 1 and Figure 2 As shown, this embodiment provides a printing production information management method, including:

[0153] S3, the step of collecting current environmental parameters and obtaining historical printed product feature data to input into a machine learning model to obtain corresponding current device parameters includes:

[0154] S30, searching the database for the original image corresponding to the to-be-printed product by name, the original image is named as x year x month x day + xx image, and obtaining a data group corresponding to the to-be-printed product;

[0155] S31, obtaining historical printed product characteristic data corresponding to the product to be printed according to the data group;

[0156] S32. Input historical printed feature data and current environmental parameters into the machine learning model, and the machine learning model outputs the current device parameter number.

[0157] S4, preliminarily adjusting the printing device according to the current device parameters to obtain the current printed product, obtaining the feature data difference between the current printed product and the historical printed product, and adjusting the device parameters according to the feature data difference until the feature data difference is less than the maximum feature data difference, the step includes:

[0158] S40, applying the current device parameters to the printing device, and performing preliminary regulation on the printing device;

[0159] S41, the printing device prints the original image using the current device parameters to obtain a current printed product (printing is stopped when the current printed product is obtained);

[0160] S42, photographing and preprocessing the current printed product by a camera device to obtain feature data of the current printed product, and obtaining a feature data difference between the feature data of the current printed product and feature data of historical printed products;

[0161] S43, inputting the feature data difference and the current environmental parameters into the machine learning model, and the machine learning model obtains the difference device parameters;

[0162] S44, adding the difference device parameter to the current device parameter and adjusting the printing device until the current printed product feature data is less than the maximum feature data difference.

[0163] The method of training the second machine learning model through feature data difference and difference device parameters is as follows:

[0164] The characteristic data difference, the difference device parameter, and the environmental parameter are combined into a second sample set, and the difference device parameter is numerically numbered;

[0165] The feature data difference of each sample and the environmental parameters are constructed into a feature vector as the input of the machine learning model;

[0166] Use keras to define the input layer and output layer and build a machine learning framework, with the feature vector as the input layer and the historical printing equipment parameter number as the output layer;

[0167] Dividing the second sample set into a training set and a test set to train the machine learning model to obtain the machine learning model, using the test set to test the machine learning model, and outputting a machine learning model that meets a preset accuracy, wherein the machine learning model is one of a logistic regression model, a naive Bayes model, or a support vector machine model;

[0168] The feature data difference is obtained by subtracting the feature data of the current printed product from the feature data of the historical printed product;

[0169] The difference device parameters are obtained by subtracting the current device parameters from the historical device parameters.

[0170] Methods for obtaining the maximum feature data difference include:

[0171] Obtain printed product feature data corresponding to n printed products of the same original image under different device parameters and environmental parameters;

[0172] Obtain n feature data differences between n printed feature data and original image feature data, and sort the n feature data differences from small to large;

[0173] The feature data of the printed products that are significantly different according to manual evaluation are taken as the maximum feature data difference.

[0174] In this embodiment, by instantly collecting the current environmental parameters and the characteristic data of the product to be printed, and predicting the current equipment parameters according to the machine learning model, real-time regulation of the printing equipment can be achieved, so as to better adapt to different printing needs and environmental conditions. By adjusting the equipment parameters according to the difference in characteristic data, precise regulation of the printing equipment can be achieved, so that the characteristic data of the printed product is closer to the historical printed product, thereby improving the quality and consistency of the printed product. Compared with the traditional test and adjustment method, this method can reduce the time and cost of test and adjustment, and improve production efficiency and economic benefits.

[0175] Example 2

[0176] See also Figure 3 As shown, this embodiment provides a printing production information management system, including:

[0177] A data collection module collects the original image and the historical equipment parameters and historical environment parameters of the historical printed product corresponding to the original image as the first historical printing data, and obtains the historical printed product characteristic data by pre-processing the historical printed product;

[0178] A database construction module is used to construct a database according to the first historical printing data and the historical printed product feature data, and a machine learning model is constructed and trained through the database;

[0179] The real-time acquisition output module collects current environmental parameters and obtains historical printed product feature data to input into the machine learning model to obtain the corresponding current device parameters;

[0180] The control and optimization module performs preliminary control on the printing equipment according to the current equipment parameters, obtains the current printed product, obtains the feature data difference between the current printed product and the historical printed product, and controls the equipment parameters according to the feature data difference until the feature data difference is less than the maximum feature data difference.

[0181] This embodiment can improve printing quality and efficiency, reduce human errors and subjective influences, reduce costs, and improve production efficiency by predicting and optimizing printing equipment parameters; use machine learning models to predict equipment parameters in real time to achieve instant regulation of printing equipment to meet different printing needs and environmental conditions; accurately adjust equipment parameters to make the quality of printed products more stable and improve consistency.

[0182] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters and thresholds in the formula are set by technicians in this field according to actual conditions.

[0183] The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technical users in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A printing production information management method, characterized in that: include: S1, collecting an original image and historical equipment parameters and historical environment parameters of a historical printed product corresponding to the original image as first historical printing data, and obtaining historical printed product feature data by preprocessing the historical printed product; S2. construct a database based on the first historical printing data and the historical printed product feature data, and construct and train a machine learning model through the database; S3, collect current environmental parameters and obtain historical printed product feature data to input into the machine learning model to obtain the corresponding current device parameters; S4, preliminarily adjusting the printing device according to the current device parameters, obtaining the current printed product, obtaining the feature data difference between the current printed product and the historical printed product, and adjusting the device parameters according to the feature data difference until the feature data difference is less than the maximum feature data difference; The step of collecting the original image and the historical equipment parameters and historical environment parameters of the historical printed product corresponding to the original image as the first historical printing data includes: S10, collecting historical equipment parameters, the historical equipment parameters including printing speed, printing pressure and printing inkjet volume; By installing a photoelectric sensor in the transmission system of the printing device, the printing speed is determined by detecting the rotation of the transmission shaft or the printing roller to obtain the printing speed in the historical device parameters; Printing pressure is the sum of upper pressure and lower pressure. Upper pressure is the pressure applied on the printing head, printing roller or printing mold during printing, and lower pressure is the pressure applied on the printing material or printing bed during printing. Both upper pressure and lower pressure are obtained by the pressure system on the printing equipment. The printing inkjet volume includes the jetting frequency and the distance between the nozzle and the printing material. The jetting frequency is obtained by the jetting frequency sensor, and the distance between the nozzle and the printing material is obtained by the distance sensor. The formula for the printing inkjet volume is: Among them, pml is the printing inkjet volume, k1 is a constant, f is the jetting frequency, and d is the distance between the nozzle and the printing material; S11. Collect historical environmental parameters, which include ambient temperature, ambient humidity and air dust density; the ambient temperature is obtained by a temperature sensor, the ambient humidity is obtained by a humidity sensor, and the air dust density is obtained by a laser particle sensor.

2. The printing production information management method according to claim 1, characterized in that: The step of obtaining characteristic data of historical printed products by preprocessing the historical printed products comprises: S12, photographing the printed product of the original image through the printing device by using a camera device to obtain an initial printed image, and correcting the initial printed image to obtain a historical printed product; S13, segmenting the historical printed matter using an improved adaptive threshold algorithm to obtain a feature extraction region of the historical printed matter; S14, acquiring historical printed product feature data of the feature extraction area, the historical printed product feature data including color distribution value, color gradient value, brightness distribution value and ink dot density value; The method of correcting the initial printed image includes: S121, cutting and adjusting the initial printed image, cutting or enlarging the initial printed image to the same specifications as the original image and making the pattern content in the initial printed image correspond to the pattern content in the original image; S122, obtaining an influence area of ​​the initial printed image, dividing the image area of ​​the original image and the initial printed image according to the influence area, and obtaining an image area division influence map; S123, obtaining the influence coefficient corresponding to the influence area in the image area division influence map; performing color correction on the initial printed image according to the influence coefficient corresponding to the influence area; The image area division method includes: An original image and an initial printing image are set, a virtual grid is preset according to the size specification of the original image, the original image is a pure color image, the gray value of the pure color image is obtained, and the same gray value in the preset virtual grid is defined as an influence area, so as to perform image area division until an image area division influence map is obtained; Methods for obtaining the influence coefficient include: The color difference between the initial printed image and the original image of the same affected area is obtained, and the influence coefficient of the affected area is obtained according to the color difference.

3. The printing production information management method according to claim 2, characterized in that: Methods for obtaining the color distribution value include: S140, using a K clustering algorithm to obtain k color clusters in the feature extraction area, each color cluster forms a color vector with a length of k, counting the occurrence frequencies of different color clusters in the feature extraction area, and performing weighted average of the occurrence frequencies of different color clusters to obtain a color distribution value; Methods for obtaining the color gradient value include: S141, dividing the feature extraction area into color regions according to different color clusters to obtain color division regions, calculating gradient values ​​between different color division regions, and performing weighted average of the gradient values ​​between different color division regions to obtain a color gradient value; The method of obtaining the brightness distribution value includes: S142, calculating the grayscale histogram of the feature extraction area, counting the number of pixels at different brightness levels in the grayscale histogram, and performing weighted average on the number of pixels at different brightness levels to obtain a brightness distribution value; The method of obtaining the ink dot density value includes: S143, counting the number of pixels in the feature extraction area as the total number of the feature extraction area, counting the number of ink dot pixels in the feature extraction area as the number of ink dots, and taking the ratio of the number of ink dots to the total number of the feature extraction area as the ink dot density value.

4. The printing production information management method according to claim 3, characterized in that: The step of segmenting the historical printed matter using the improved adaptive threshold algorithm to obtain the feature extraction area of ​​the historical printed matter comprises: S130, improving the adaptive threshold algorithm by using a particle swarm algorithm and grayscale entropy to obtain a segmentation threshold for the feature extraction area; S1301, initializing a particle swarm, each particle represents a threshold vector, the length of the threshold vector matches the number of blocks to be segmented in the historical printed product, and each element in the threshold vector represents a threshold of a block to be segmented; S1302, segmenting the historical printed matter according to the threshold vector corresponding to each particle, calculating the inter-class variance of the segmented historical printed matter, and selecting the highest inter-class variance to calculate the fitness of the segmentation threshold; S1303, update the position and speed of each particle, find the individual optimal position and the global optimal position corresponding to the fitness, The formula for updating the particle position is: x i (t+1)=x i (t)+v i (t+1), Among them, x i (t) is the position of particle i at time t, x i (t+1) is the position of particle i at time t+1, v i (t+1) is the velocity of particle i at time t+1; The formula for updating the particle velocity is: v i (t+1)=ω×v i (t)+c1×r1×(pbest-x i (t))+c2×r2×(gbest-x i (t)), Among them, ω is the inertia weight, c1 and c2 are the first acceleration coefficient and the second acceleration coefficient, r1 and r2 are the first random number and the second random number, pbest is the individual optimal position of particle i, gbest is the global optimal position of the entire particle swarm, v i (t) is the velocity of particle i at time t; S1304, check whether the termination condition is met, if the maximum number of iterations is reached or the fitness function value converges to a preset threshold; S1305, selecting a threshold vector with the best fitness from all particles as the final segmentation threshold; S1306 , using the optimal threshold vector to segment the historical printed product to obtain a feature extraction region of the historical printed product.

5. The printing production information management method according to claim 4, characterized in that: The inter-class variance of the historical prints is: Among them, ICV is the inter-class variance, N is the number of blocks to be segmented, ω z is the weight of the zth block to be segmented, μ z is the grayscale mean of the zth block to be segmented, μ total is the grayscale mean of the entire historical print; The fitness of the segmentation threshold is: Among them, p j is the probability of the jth gray level, and L is the total number of gray levels.

6. The printing production information management method according to claim 1, characterized in that: The step of constructing a database according to the first historical printing data and the historical printed product feature data, and constructing and training a machine learning model through the database comprises: S20, taking the first historical printing data and the historical printed product characteristic data corresponding to the same original image and the historical printed product as a data group, wherein the data in the data group are mutually related, and the data group is named after the original image; obtaining n data groups within the historical period, and constructing the n data groups into a database; S21, combining historical printed product feature data and historical printing device parameters and historical environment parameters corresponding to the historical printed product feature data into a first sample set, and numerically numbering the historical printing device parameters; S22, constructing the historical printed material feature data and historical environmental parameters of each sample into a feature vector as the input of the machine learning model; S23, use keras to define the input layer and output layer and build a machine learning framework, with the feature vector as the input layer and the historical printing equipment parameter number as the output layer; S24, dividing the first sample set into a training set and a test set, constructing a machine learning model, training the machine learning model to obtain the machine learning model, testing the machine learning model using the test set, and outputting a machine learning model that meets a preset accuracy, wherein the machine learning model is one of a logistic regression model, a naive Bayes model, or a support vector machine model; The calculation formula for the preset accuracy is: Among them, Q is the total number of samples in the first sample set, Q C The minimum accuracy is preset for the number of samples that are correctly predicted. When the accuracy reaches the minimum accuracy, the training of the machine learning model can be stopped.

7. The printing production information management method according to claim 1, characterized in that: The step of collecting current environmental parameters and obtaining historical printed product feature data to input into a machine learning model to obtain corresponding current device parameters includes: S30, searching the database for the original image corresponding to the to-be-printed product by name to obtain a data group corresponding to the to-be-printed product; S31, obtaining historical printed product characteristic data corresponding to the product to be printed according to the data group; S32. Input historical printed feature data and current environmental parameters into the machine learning model, and the machine learning model outputs the current device parameter number.

8. The printing production information management method according to claim 1, characterized in that: The step of preliminarily adjusting the printing device according to the current device parameters to obtain the current printed product, obtaining the feature data difference between the current printed product and the historical printed product, and adjusting the device parameters according to the feature data difference until the feature data difference is less than the maximum feature data difference includes: S40, applying the current device parameters to the printing device, and performing preliminary regulation on the printing device; S41, the printing device prints the original image using the current device parameters to obtain a current printed product; S42, photographing and preprocessing the current printed product by a camera device to obtain feature data of the current printed product, and obtaining a feature data difference between the feature data of the current printed product and feature data of historical printed products; S43, inputting the feature data difference and the current environmental parameters into the machine learning model, and the machine learning model obtains the difference device parameters; S44, adding the difference device parameter to the current device parameter and adjusting the printing device until the current printed product feature data is less than the maximum feature data difference.

9. A printing production information management system, used to implement the printing production information management method according to any one of claims 1 to 8, characterized in that: include, A data collection module collects the original image and the historical equipment parameters and historical environment parameters of the historical printed product corresponding to the original image as the first historical printing data, and obtains the historical printed product characteristic data by pre-processing the historical printed product; A database construction module is used to construct a database according to the first historical printing data and the historical printed product feature data, and a machine learning model is constructed and trained through the database; The real-time acquisition output module collects current environmental parameters and obtains historical printed product feature data to input into the machine learning model to obtain the corresponding current device parameters; The control and optimization module performs preliminary control on the printing equipment according to the current equipment parameters, obtains the current printed product, obtains the feature data difference between the current printed product and the historical printed product, and controls the equipment parameters according to the feature data difference until the feature data difference is less than the maximum feature data difference.

Citation Information

Patent Citations

  • Method for detecting quality of printed matter

    CN105844621A

  • Information processing system, information processing method, and learning device

    CN114193929A

  • Information processing system, learning device, and information processing method

    CN114261202A

  • Automatic control method and system for jet printing of multi-product production line

    CN115179667A

  • Ink droplet printing method, device and system and storage medium

    CN115674922A