Printing control method, device, system and equipment of flexo printing machine and program product

Abstract

The embodiment of the invention discloses a printing control method, device, system and equipment of a flexo printing machine and a program product, and belongs to the technical field of printing. The method comprises the steps that spectral data of a printing area in a printed matter and operation data of the flexo printing machine are obtained, the printed matter is obtained by printing a standard printing sample through the flexo printing machine, and the operation data of the flexo printing machine are obtained; the operation data comprises process parameters and environment parameters of the flexo printing machine; the spectral data and the operation data are input into a chromatic aberration prediction model, a predicted chromatic aberration value output by the chromatic aberration prediction model is obtained, and the predicted chromatic aberration value is used for representing the deviation between a predicted chromatic value and a target chromatic value of a printed matter obtained through printing of the flexographic printing machine in the future time period; and according to the predicted color difference value and a preset color difference threshold value, current control parameters of the flexographic printing machine are adjusted. According to the embodiment of the invention, the method achieves the conversion from passive detection to active prediction, and remarkably improves the stability and consistency of the printing color of the flexographic printing machine.

Description

Technical Field

[0001] This application belongs to the field of printing technology, and in particular relates to a printing control method, device, system, equipment and program product for a flexographic printing machine. Background Technology

[0002] During the printing process, flexographic printing machines are affected by various factors such as the shape and color of the substrate and the control parameters of the flexographic printing machine. As a result, there will be a color difference between the printed material and the printed sample, which will affect the aesthetics of the printed material.

[0003] Furthermore, the relevant technology usually involves discovering the color difference between the printed material and the printed sample during the use of the flexographic printing machine, and then adjusting the control parameters of the flexographic printing machine based on experience. This method of adjustment is lagging behind, and the production loss caused by the printing color deviation of the flexographic printing machine during the adjustment period is relatively high. Summary of the Invention

[0004] This application provides a printing control method, apparatus, system, equipment, and program product for a flexographic printing machine, which realizes feedforward closed-loop control of the printing quality of the flexographic printing machine and effectively overcomes the control lag problem caused by post-adjustment in the prior art.

[0005] In a first aspect, embodiments of this application provide a printing control method for a flexographic printing machine, comprising:

[0006] Acquire spectral data of the printed area in the printed material and operating data of the flexographic printing machine, wherein the printed material is obtained by printing a standard printed sample by the flexographic printing machine, and the operating data includes the process parameters and environmental parameters of the flexographic printing machine.

[0007] The spectral data and the operating data are input into the color difference prediction model to obtain the predicted color difference value output by the color difference prediction model. The predicted color difference value is used to characterize the deviation between the predicted color value and the target color value of the printed matter printed by the flexographic printing press in a future time period.

[0008] The current control parameters of the flexographic printing machine are adjusted based on the predicted color difference value and the preset color difference threshold.

[0009] In some embodiments, the color difference prediction model includes an input layer, a one-dimensional convolutional neural network sub-model, a fully connected network sub-model, a feature fusion processing layer, and an output layer, wherein the one-dimensional convolutional neural network sub-model and the fully connected network sub-model are respectively connected to the feature fusion processing layer;

[0010] The color difference prediction model processes the spectral data into a corresponding first feature vector based on the one-dimensional convolutional neural network sub-model; normalizes the running data based on the fully connected network sub-model to obtain a normalized second feature vector; preprocesses the first feature vector and the second feature vector based on the feature fusion processing layer to obtain a fused feature vector, determines the predicted color difference value based on the fused feature vector; and outputs the predicted color difference value based on the output layer, wherein the preprocessing includes spatiotemporal alignment processing and fusion processing.

[0011] In some embodiments, the feature fusion processing layer preprocesses the first feature vector and the second feature vector to obtain the fused feature vector in the following manner:

[0012] Based on the timestamp of the collected running data, the second feature vector and the first feature vector are spatiotemporally aligned to obtain the aligned first feature vector and the aligned second feature vector.

[0013] An attention mechanism is used to fuse the aligned first feature vector and the aligned second feature vector, and the attention weights of each dimension in the aligned first feature vector and the aligned second feature vector are calculated.

[0014] Based on the attention weights of each dimension in the aligned first feature vector and the aligned second feature vector, the aligned first feature vector and the aligned second feature vector are weighted and summed to obtain the fused feature vector.

[0015] In some embodiments, before acquiring spectral data of the printed area in the printed material and operating data of the flexographic printing press, the method further includes:

[0016] Multiple sets of training samples are obtained, each set of training samples includes sample spectral data and sample running data, as well as the corresponding sample color difference value;

[0017] The multiple sets of training samples are divided into training datasets and validation datasets;

[0018] The training dataset is input into a pre-built initial network model. With the goal of minimizing a preset composite loss function, the parameters in the initial network model are updated to obtain an optimized network model. The initial network model includes an input layer, a one-dimensional convolutional neural network sub-model, a fully connected network sub-model, a feature fusion layer, and an output layer.

[0019] Input the validation dataset into the optimized network model and check whether the optimized network model has converged;

[0020] If the optimized network model has converged, it is determined that the optimized network model has been trained successfully, and the color difference prediction model is obtained.

[0021] In some embodiments, obtaining multiple sets of training samples includes:

[0022] Multiple sets of different sample operation data are set, wherein the sample operation data includes sample process parameters and sample environment parameters. The sample process parameters include at least one of different ink key opening, printing speed, and current printing pressure, and the sample environment parameters include at least one of different ambient temperature and ambient humidity.

[0023] For each set of sample operation data, the flexographic printing press is controlled to print the standard printing sample according to the sample operation data of that set, so as to obtain the corresponding sample print.

[0024] The spectral data of the printed area in the printed sample were collected using a spectrometer.

[0025] The color difference between the printed sample and the standard printed sample is measured using a colorimeter, and the resulting color difference is used as the label data corresponding to the sample running data.

[0026] The sample spectral data, sample operation data, and sample color difference values ​​of each group are combined to form a set of training samples.

[0027] In some embodiments, the method further includes:

[0028] Obtain the actual color difference value between the printed material and the standard printed sample;

[0029] The model parameters of the color difference prediction model are adjusted based on the predicted color difference value and the actual color difference value.

[0030] In some embodiments, adjusting the model parameters of the color difference prediction model based on the predicted color difference value and the actual color difference value includes:

[0031] The error value between the predicted color difference value and the actual color difference value is calculated using a preset composite loss function.

[0032] The error value is passed into the color difference prediction model through the backpropagation algorithm, and the convolution kernel parameters, pooling layer parameters, weight matrix and bias terms of the one-dimensional convolutional neural network sub-model, as well as the fully connected network sub-model, are updated sequentially in the color difference prediction model.

[0033] After each round of parameter updates, the spectral data and running data of the current batch are used as test samples and input into the updated color difference prediction model to verify whether the error between the predicted color difference value and the actual color difference value output by the updated color difference prediction model is less than the preset error threshold.

[0034] If the error value is less than the preset error threshold, then the adjustment of the model parameters of the color difference prediction model shall be stopped.

[0035] If the error value is greater than or equal to the preset error threshold, the calculation of the error value between the predicted color difference value and the actual color difference value and subsequent steps are repeated until the error value between the predicted color difference value and the actual color difference value output by the updated color difference prediction model is less than the preset error threshold.

[0036] In some embodiments, the composite loss function is L = α·MSE(ΔE) + β·KL(Spectral), where MSE(ΔE) is the numerical deviation between the predicted color difference value and the actual color difference value, KL(Spectral) is the spectral distribution difference between the predicted color difference value and the actual color difference value, and α and β are weighting coefficients that satisfy α + β = 1.

[0037] In some embodiments, adjusting the current control parameters of the flexographic printing machine based on the predicted color difference value and a preset color difference threshold includes:

[0038] If the predicted color difference value is greater than the preset color difference threshold, the predicted color difference value and the current control parameters of the flexographic printing machine are input into the parameter offset prediction model to obtain the control parameter offset output by the parameter offset prediction model with the preset color difference value range as the target.

[0039] The control parameter offset is sent to the flexographic printing press so that the flexographic printing press adjusts the current control parameters and prints the standard printing sample according to the adjusted control parameters.

[0040] In some embodiments, sending the control parameter offset to the flexographic printing press to cause the flexographic printing press to adjust the current control parameter includes at least one of the following:

[0041] If the current control parameter is printing pressure and the control parameter offset is pressure offset, then the pressure offset is sent to the controller of the flexographic printing press, so that the controller drives the servo motor of the pressure adjustment mechanism of the flexographic printing press according to the pressure offset, and adjusts the current printing pressure applied by the pressure adjustment mechanism to the printing plate cylinder of the flexographic printing press.

[0042] If the current control parameter is the ink key opening, and the control parameter offset is the ink key opening offset, then the ink key opening offset is sent to the controller of the flexographic printing press, so that the controller drives the ink key actuator of the flexographic printing press to adjust the current ink key opening of the flexographic printing press.

[0043] If the current control parameter is the printing speed and the control parameter offset is the speed offset, then the speed offset is sent to the controller of the flexographic printing machine so that the controller adjusts the current printing speed of the flexographic printing machine.

[0044] In some embodiments, acquiring spectral data of the printed area in the printed material and operating data of the flexographic printing machine includes:

[0045] Acquire spectral data of the printed area in the printed material collected by a spectrometer, wherein the printed area includes a color bar area and / or a pattern area;

[0046] The flexographic printing machine receives operating data from sensors via an industrial bus, wherein the sensors include at least one of the following: a pressure sensor for acquiring current printing pressure, an ink key position sensor for acquiring ink key opening, a speed sensor for acquiring printing speed, and a temperature and humidity sensor for acquiring environmental parameters.

[0047] Secondly, embodiments of this application provide a printing control device for a flexographic printing machine, comprising:

[0048] The acquisition unit acquires spectral data of the printing area in the printed material and operating data of the flexographic printing machine, wherein the printed material is obtained by printing a standard printed sample by the flexographic printing machine, and the operating data includes the process parameters and environmental parameters of the flexographic printing machine.

[0049] The prediction unit inputs the spectral data and the operating data into the color difference prediction model to obtain the predicted color difference value output by the color difference prediction model. The predicted color difference value is used to characterize the deviation between the predicted colorimetric value and the target colorimetric value of the printed matter printed by the flexographic printing press in a future time period.

[0050] The adjustment unit adjusts the current control parameters of the flexographic printing machine based on the predicted color difference value and the preset color difference threshold.

[0051] Thirdly, embodiments of this application provide a printing control system for a flexographic printing machine, including a terminal device and a flexographic printing machine, wherein the terminal device is communicatively connected to the flexographic printing machine, and the terminal device is configured to execute the printing control method for the flexographic printing machine described in any one of the first aspects above.

[0052] In some embodiments, the terminal device is configured to, when the predicted color difference value is greater than the preset color difference threshold, input the predicted color difference value and the current control parameters of the flexographic printing machine into a parameter offset prediction model to obtain a control parameter offset output by the parameter offset prediction model with a preset color difference value range as the target; and send the control parameter offset to the flexographic printing machine.

[0053] The flexographic printing machine is configured to receive a control parameter offset sent by the terminal device, adjust the current control parameters based on the control parameter offset, and print a standard printing sample according to the adjusted control parameters.

[0054] In some embodiments, the system further includes sensors.

[0055] The sensor is communicatively connected to the terminal device. The sensor is configured to collect environmental parameters of the operating environment of the flexographic printing machine and send the environmental parameters to the terminal device.

[0056] Fourthly, embodiments of this application provide a terminal device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the printing control method of the flexographic printing machine according to any one of the first aspects described above.

[0057] Fifthly, embodiments of this application provide a computer program product, including a computer program, which, when run, causes the printing control method of the flexographic printing press described in any one of the first aspects to be executed.

[0058] Compared to related technologies, the printing control method, apparatus, system, equipment, and program products for flexographic printing machines provided in this application acquire spectral data of the printing area in the printed material and the operating data of the flexographic printing machine. The printed material is obtained by printing a standard printed sample using the flexographic printing machine. The operating data includes the process parameters and environmental parameters of the flexographic printing machine. The spectral data and operating data are input into a color difference prediction model to obtain a predicted color difference value output by the model. This predicted color difference value characterizes the deviation between the predicted chromaticity value and the target chromaticity value of the printed material to be printed by the flexographic printing machine in a future time period. Based on the predicted color difference value and a preset color difference threshold, the current control parameters of the flexographic printing machine are adjusted. This achieves feedforward closed-loop control of the printing quality of the flexographic printing machine, effectively overcoming the control lag problem caused by post-adjustment in the prior art, significantly shortening the control response time, reducing production losses caused by printing color deviations in the flexographic printing machine, and thus improving the stability of the printing quality of the flexographic printing machine. Attached Figure Description

[0059] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0060] Figure 1 This is a flowchart of a printing control method for a flexographic printing machine according to an embodiment of this application;

[0061] Figure 2 This is a schematic diagram of the structure of a color difference prediction model according to an embodiment of this application;

[0062] Figure 3 This is a schematic diagram of the printing control system of a flexographic printing machine according to an embodiment of this application;

[0063] Figure 4 This is a schematic diagram of the printing control system of a flexographic printing machine according to another embodiment of this application;

[0064] Figure 5 This is a schematic diagram of the printing control device of a flexographic printing machine according to an embodiment of this application;

[0065] Figure 6 This is a schematic diagram of the structure of a terminal device according to an embodiment of this application. Detailed Implementation

[0066] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0067] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0068] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0069] As used in this application specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if detected [the described condition or event]" may be interpreted, depending on the context, as meaning "once determined," "in response to determination," "once detected [the described condition or event]," or "in response to detection [the described condition or event]."

[0070] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0071] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0072] Flexographic printing presses (also known as flexographic printing machines) use highly fluid inks. The ink is transferred to the image areas of the printing plate by the ink fountain roller and anilox roller, where it is inked. Then, the pressure roller applies printing pressure, transferring the ink from the printing plate to the substrate. Finally, the printing process is completed through a drying process. Its main advantages are its simple machine structure and ease of operation.

[0073] During the printing process, color differences will exist between the printed material and the original electronic image due to various factors such as the shape and color of the substrate and the current control parameters of the flexographic printing press. This affects the aesthetics of the printed material. Currently, color differences are often checked manually by visual inspection or by operators using handheld instruments to inspect and adjust them. Furthermore, when the flexographic printing press is not in operation (e.g., stopped), the current control parameters are manually adjusted based on the color difference of the printed material. However, such adjustments rely heavily on the operator's experience. Therefore, these adjustments are inaccurate and inefficient, and cannot respond promptly to color difference changes that occur during the printing process.

[0074] Currently, no effective solution has been proposed to address the problem that adjustments to the current control parameters of flexographic printing presses are difficult to make in a timely and accurate manner in response to color difference changes that occur during the printing process.

[0075] In view of this, this application provides a printing control method for a flexographic printing machine, applicable to flexographic production scenarios with high requirements for color consistency, such as food packaging, label printing, and flexible packaging. It aims to solve the technical problems of relying heavily on post-processing color difference detection and high control delays in the printing control of related flexographic printing machines. By acquiring the spectral data of the printing area in the printed material and the operating data of the flexographic printing machine, it effectively reduces the false alarm rate of metamerism, improves the accuracy of color difference prediction, and ensures the color consistency of the printed material under different light sources. The spectral data and operating data are input into a pre-trained color difference prediction model to obtain the predicted color difference value for future periods. Based on the comparison between the predicted color difference value and a preset color difference threshold, the current control parameters of the flexographic printing machine are dynamically adjusted, realizing feedforward closed-loop control of the printing quality of the flexographic printing machine. This effectively overcomes the control lag problem caused by post-processing adjustments in the prior art, significantly shortens the control response time, reduces production losses caused by printing color deviations, and improves the stability of the printing quality of the flexographic printing machine.

[0076] Details of one or more embodiments of this application are set forth in the following drawings and description to make other features, objects and advantages of this application more readily apparent.

[0077] The following will combine Figure 1 A printing control method for a flexographic printing machine according to one embodiment of this application will be described; please refer to [link to relevant documentation]. Figure 1 , Figure 1 This is a flowchart of a printing control method for a flexographic printing machine according to an embodiment of this application, such as... Figure 1 As shown, the method includes:

[0078] Step S101: Obtain spectral data of the printing area in the printed material and operating data of the flexographic printing machine.

[0079] The printed materials are obtained by printing standard printed samples using a flexographic printing machine, and the operating data includes the process parameters and environmental parameters of the flexographic printing machine.

[0080] In some embodiments, the printed material is generated by a flexographic printing press through continuous printing of a preset standard printed sample.

[0081] In some embodiments, after the flexographic printing press prints a standard printed sample, a spectrometer deployed next to the take-up end of the flexographic printing press can be used to collect spectral data (e.g., local spectral data, local spectral reflectance, local spectral reflectance curve, etc.). The measurement probe of the spectrometer is facing the transmission path of the printed material, and the vertical distance between the center of the probe and the surface of the printed material is controlled within the range of 5cm-10cm to avoid collisions caused by being too close or measurement errors caused by being too far away.

[0082] In some embodiments, the operating data of the flexographic printing machine is obtained through corresponding sensors for process parameters and environmental parameters, and all process and environmental parameters collected by the sensors are transmitted to the processor of the terminal device via an industrial bus. For example, for the current printing pressure in the process parameters, a pressure sensor can be installed next to the contact area between the printing plate cylinder and the impression cylinder to collect the pressure applied to the printing plate cylinder by the pressure regulating mechanism in real time; for the ink key opening in the process parameters, a position sensor can be installed on the actuator of each ink key to calculate the ink key opening by detecting the displacement of the ink key baffle; for the printing speed in the process parameters, a speed sensor can be installed on the output shaft of the main motor of the flexographic printing machine to obtain the printing speed by converting the speed to the cylinder circumference.

[0083] Step S102: Input the spectral data and running data into the color difference prediction model to obtain the predicted color difference value output by the color difference prediction model.

[0084] Among them, the predicted color difference value is used to characterize the deviation between the predicted color value and the target color value of the printed matter obtained by the flexographic printing press in the future period.

[0085] In some embodiments, the color difference prediction model is a pre-trained dual-branch neural network model, which is used to fuse spectral data and running data to obtain a fused feature vector, so as to output the predicted color difference value of the printed matter in the future time period. The predicted color difference value is used to characterize the deviation between the predicted color value and the target color value of the printed matter printed by the flexographic printing press in the future time period (e.g., 50ms-100ms, the specific duration is determined by the printing speed, the higher the speed, the shorter the prediction duration).

[0086] In some embodiments, after the spectral data and the flexographic printing machine's operating data are input into the color difference prediction model through the input layer, the data are processed by a one-dimensional convolutional neural network (1D-CNN) sub-model and a fully connected network sub-model to obtain a first feature vector and a second feature vector, respectively. The color difference prediction model then fuses the first and second feature vectors to obtain a fused feature vector, and calculates and outputs the predicted color difference value based on the fused feature vector. The terminal device transmits the predicted color difference value to the flexographic printing machine's controller in real time and stores it in a local database for subsequent parameter adjustments and model optimization.

[0087] In some embodiments, to ensure the validity of the prediction results, every 10 sets of prediction data, the processor of the terminal device can also randomly select one set of actual color difference values ​​of the current printed material and compare it with the corresponding predicted color difference values. If the deviation between the two is detected to exceed a preset error threshold (e.g., 0.1), a temporary calibration or fine-tuning of the color difference prediction model is triggered to ensure that the prediction accuracy of the color difference prediction model actually used is always maintained within ±0.2.

[0088] Step S103: Adjust the current control parameters of the flexographic printing machine according to the predicted color difference value and the preset color difference threshold.

[0089] In some embodiments, the preset color difference threshold can be set according to the quality level of the printed material, and the control parameters of the flexographic printing machine can be precisely adjusted through feedforward control logic to avoid the expansion of printing color deviation.

[0090] As an illustration, the preset color difference threshold is determined based on the application scenario of the printed material: For high-precision printed materials (such as brand logos on food packaging and cosmetic labels), the preset color difference threshold is set to ΔE = 0.8, that is, when the predicted color difference value exceeds 0.8, the current control parameters of the flexographic printing machine need to be adjusted; for medium-precision printed materials (such as packaging for ordinary daily necessities), the preset color difference threshold is set to ΔE = 1.2; for low-precision printed materials (such as corrugated cardboard boxes), the preset color difference threshold is set to ΔE = 1.5.

[0091] In some embodiments, the preset color difference threshold is stored in the parameter configuration module of the controller and can be modified according to actual production needs through the human-machine interface. The processor of the terminal device compares the predicted color difference value output by the color difference prediction model with the preset color difference threshold in real time: if the predicted color difference value is less than or equal to the preset color difference threshold, it means that the printing color deviation in the future period is within an acceptable range, and the flexographic printing press is controlled to maintain the current control parameters (including the current pressure adjustment, the current ink key opening, and the current printing speed) and continue to operate; if the predicted color difference value is greater than the preset color difference threshold, the current control parameters of the flexographic printing press are adjusted.

[0092] In some embodiments, the processor of the terminal device inputs the predicted color difference value and the current control parameters of the flexographic printing press (i.e., the parameters of the flexographic printing press to be adjusted, including the current pressure, current ink key opening, current printing speed, etc.) into a pre-trained parameter offset prediction model. As an example, and not a limitation, the parameter offset prediction model can be a regression model based on a gradient boosting decision tree, aiming to make the adjusted color difference value fall within a preset color difference value range (0 < ΔE ≤ preset color difference threshold), and outputting corresponding control parameter offsets, including pressure offsets, ink key opening offsets, and speed offsets. Furthermore, the processor of the terminal device sends the control parameter offsets to the controller of the flexographic printing press via an industrial bus to adjust the current control parameters of the flexographic printing press corresponding to the control parameter offsets.

[0093] In some embodiments, acquiring spectral data of the printed area in the printed material and operating data of the flexographic printing press includes:

[0094] Acquire spectral data of the printed area in the printed material collected by a spectrometer, wherein the printed area includes color bar areas and / or pattern areas;

[0095] The flexographic printing machine receives operating data from sensors via an industrial bus. The sensors include at least one of the following: a pressure sensor for acquiring the current printing pressure, an ink key position sensor for acquiring the ink key opening, a speed sensor for acquiring the printing speed, and a temperature and humidity sensor for acquiring environmental parameters.

[0096] In some embodiments, the selection of the printing area can be determined in conjunction with the structure of the printed matter: for example, if the printed matter contains color bar areas (such as color bars set along the height or width direction of the printed matter), then the color bar areas are preferentially selected as the printing areas to be collected; if the printed matter contains pattern areas, that is, the printed matter is a pattern-type printed matter without color bar areas (such as complex packaging patterns), then the key color areas in the pattern (such as brand logos, uniform background areas) are selected as the printing areas to be collected, and each collection area must avoid the registration deviation areas at the edges of the pattern.

[0097] In some embodiments, the spectrometer acquires spectral data of the printing area in a spectral band covering the visible light range of 400nm-700nm, with a sampling interval of 10nm-20nm to balance data accuracy and processing efficiency. The spectrometer transmits the acquired spectral data to the controller of the flexographic printing press in the form of digital signals, and each set of spectral data carries a timestamp accurate to the millisecond level to facilitate subsequent spatiotemporal synchronization processing with the operating data of the flexographic printing press.

[0098] In some embodiments, the operating data of the flexographic printing machine includes process parameters and environmental parameters. These two types of parameters are collected by corresponding sensors, and all operating data collected by the sensors is transmitted to the processor of the terminal device via an industrial bus. For example, regarding the current printing pressure in the process parameters, a pressure sensor can be installed beside the contact area between the printing plate cylinder and the impression cylinder. The measurement range of the pressure sensor is set to 0.1MPa-1.0MPa to collect the pressure applied to the printing plate cylinder by the pressure regulating mechanism in real time. Regarding the ink key opening in the process parameters, a position sensor can be installed on the actuator of each ink key to calculate the ink key opening by detecting the displacement of the ink key baffle (ink key opening = actual displacement / maximum displacement × 100%, for example, the maximum displacement is preset to 25mm). Regarding the printing speed in the process parameters, a speed sensor can be installed on the output shaft of the main motor of the flexographic printing machine to obtain the printing speed by converting the speed to the cylinder circumference (printing speed = motor speed × cylinder circumference / 60). Furthermore, the acquisition frequency of the above process parameters can be set to 10Hz to maintain coordination with the acquisition frequency of the spectral data.

[0099] For example, regarding environmental parameters, a set of temperature and humidity sensors can be deployed at both the infeed and outlet ends of the flexographic printing machine to collect the ambient temperature and humidity of the printing workshop. The temperature measurement range is set to 10℃-40℃, the humidity measurement range is set to 20%RH-80%RH, and the acquisition frequency is set to 5Hz to avoid interference with the validity of the data due to frequent fluctuations in environmental parameters.

[0100] In some embodiments, such as Figure 2 As shown, the color difference prediction model comprises an input layer, a one-dimensional convolutional neural network sub-model, a fully connected network sub-model, a feature fusion processing layer, and an output layer. The one-dimensional convolutional neural network sub-model and the fully connected network sub-model are respectively connected to the feature fusion processing layer. The color difference prediction model processes the spectral data into corresponding first feature vectors based on the one-dimensional convolutional neural network sub-model; it normalizes the running data based on the fully connected network sub-model to obtain a normalized second feature vector; the feature fusion processing layer preprocesses the first and second feature vectors to obtain a fused feature vector, and determines the predicted color difference value based on the fused feature vector; and the output layer outputs the predicted color difference value. The preprocessing includes spatiotemporal alignment and fusion processing.

[0101] In some embodiments, the input layer is used to receive the preprocessed synchronous dataset. As an example and not a limitation, the input dimension of the spectral data is 1×301 (corresponding to the reflectance values ​​of 301 wavelength points with an interval of 10nm-700nm), and the input dimension of the running data is 1×5 (corresponding to 5 parameters: printing pressure, ink key opening, printing speed, ambient temperature, and ambient humidity).

[0102] In some embodiments, by way of example and not limitation, for the high-dimensional temporal characteristics of spectral data, a one-dimensional convolutional neural network (1D-CNN) sub-model can be used as the spectral feature extraction branch. This sub-model can be configured to include three convolutional layers and two pooling layers: the first convolutional layer uses 32 convolutional kernels of size 3×1 with a stride of 1, and a ReLU activation function to perform preliminary feature extraction on the input spectral data; after the first convolutional layer, a max pooling layer (one-dimensional max pooling layer MaxPooling1D, e.g., a pooling kernel size of 2×1 with a stride of 1) is connected to reduce the feature dimension of the spectral data and retain key spectral features; for example, the second convolutional layer uses 64 convolutional kernels of size 3×1 with a stride of 1, and a ReLU activation function to further extract deeper spectral information from the spectral data; after the second convolutional layer, an average pooling layer (global average pooling layer GlobalAverage Pooling, pooling kernel size of 2×1 with a stride of 1) is connected; for example, the third convolutional layer uses 64 convolutional kernels of size 3×1 with a stride of 1. Finally, the output is flattened into a first feature vector of dimension 1×64. This first feature vector contains key color-related information in the spectral data (such as the difference in reflectance peaks and valleys in a specific wavelength range).

[0103] In some embodiments, and not as a limitation, given the numerical characteristics of the running data, the fully connected network sub-model, as the running parameter extraction branch, can be configured to include two fully connected layers. For example, the first fully connected layer contains 128 neurons, and uses the ReLU activation function to normalize the input running data (such as batch normalization), outputting an intermediate vector with a dimension of 1×128; the second fully connected layer contains 64 neurons, and the ReLU activation function converts the intermediate vector into a second feature vector with a dimension of 1×64. This second feature vector reflects the comprehensive influence of running parameters such as printing process parameters and environmental parameters on the printing color.

[0104] In some embodiments, and not as a limitation, since the acquisition frequencies of spectral data and operational data may differ (e.g., spectral data is acquired 10 times per second, operational data is acquired 5 times per second), timestamp matching is needed to eliminate timing discrepancies. The feature fusion layer first compresses the dimension of the first feature vector (using a 1×1 convolution kernel to compress a 1×128 vector into a 1×32 vector), making the dimensions of the compressed first feature vector consistent with those of the second feature vector; if the acquisition time difference between the two types of data exceeds 50 milliseconds, interpolation is performed based on the changing trends of adjacent data to ensure that the fused features correspond to the same printing time.

[0105] An attention mechanism is employed to dynamically allocate the weights of the two feature vectors. For example, the average activation values ​​of the two feature vectors are normalized using a softmax function to obtain attention weights w1 (weight of the first feature vector) and w2 (weight of the second feature vector), satisfying w1 + w2 = 1. It should be understood that in practical use, when spectral data has a more significant impact on color (e.g., ink color fluctuations), the weight of the first feature vector automatically increases; when the impact of operating parameters is more prominent (e.g., sudden changes in environmental humidity), the weight of the second feature vector increases accordingly. Finally, a weighted sum of the first and second feature vectors is obtained to obtain a fused feature vector. This fused feature vector simultaneously contains spectral color features and equipment operating features, achieving deep fusion of spectral features and operating parameter features.

[0106] In some embodiments, and not as a limitation, the output layer of the color difference prediction model may employ a single-neuron structure with a linear activation function. Based on the LAB color space, the aforementioned fused feature vector is calculated to determine the deviation between the predicted chromaticity value and the target chromaticity value of the printed material to be printed by the flexographic printing press in a future time period, thereby obtaining and outputting the predicted color difference value. The target chromaticity value is the standard chromaticity value of a standard printed sample. For example, the actual chromaticity values ​​of multiple regions of the standard printed sample can be pre-measured using a colorimeter, and the average of the actual chromaticity values ​​of the multiple regions can be taken as the target chromaticity value to ensure its reliability.

[0107] Through the structural design of the color difference prediction model described above, the color difference prediction model can make full use of the color details of spectral data and the operating condition information of the operating data to achieve accurate and forward-looking prediction of the printing color difference of the flexographic printing machine, providing a reliable basis for real-time parameter adjustment of the flexographic printing machine.

[0108] In some embodiments, the feature fusion processing layer preprocesses the first feature vector and the second feature vector in the following manner to obtain the fused feature vector:

[0109] Based on the timestamp of the collected running data, the second feature vector and the first feature vector are spatiotemporally aligned to obtain the aligned first feature vector and the aligned second feature vector.

[0110] An attention mechanism is used to fuse the aligned first feature vector and the aligned second feature vector, and the attention weights of each dimension in the aligned first feature vector and the aligned second feature vector are calculated.

[0111] Based on the attention weights of each dimension in the aligned first feature vector and the aligned second feature vector, the aligned first feature vector and the aligned second feature vector are weighted and summed to obtain the fused feature vector.

[0112] In the color difference prediction model of this application, a feature fusion processing layer is used to fuse spectral features and operational parameter features. Through weighted fusion using spatiotemporal alignment and an attention mechanism, it ensures that the two types of features can work together accurately, improving the reliability of the predicted color difference value. The specific processing method of this feature fusion processing layer is described in detail below:

[0113] Because the spectral data and runtime data come from different acquisition devices, the acquisition frequency and timestamps may differ (e.g., spectral data is acquired every 0.1 seconds, and runtime data is acquired every 0.2 seconds). Directly fusing the spectral data and runtime data would lead to feature mismatch due to temporal misalignment. Therefore, the feature fusion processing layer first performs spatiotemporal alignment processing, as follows:

[0114] For example, the original acquisition timestamps, accurate to the millisecond level, are first extracted from the first feature vector (derived from spectral data) and the second feature vector (derived from runtime data). Then, the difference between the two timestamps is calculated. If the difference is within a preset time difference threshold (e.g., 50 milliseconds), the timing of the first and second feature vectors is considered consistent, and no alignment correction is needed. If the difference exceeds the preset time difference threshold, alignment correction is required. For example, the runtime data corresponding to the second feature vector is interpolated and completed using the timestamp of the first feature vector as a reference. Through spatiotemporal alignment processing, it is ensured that the two types of features strictly correspond to the state at the same printing time, avoiding fusion errors caused by data asynchrony.

[0115] In some embodiments, for example, if the timestamp of the first feature vector is t1 and the timestamp of the second feature vector is t2 (e.g., t2 is later than t1 and the difference is 60 milliseconds), then the two most recent valid running data before t2 (timestamps t0 and t2, where t0 is earlier than t1) are extracted. Based on the changing trend of these two running data, the estimated value of the running data at time t1 is calculated, and then the second feature vector is regenerated based on the estimated value, finally obtaining the aligned first feature vector and the aligned second feature vector.

[0116] In some embodiments, after completing the spatiotemporal alignment process, the feature fusion processing layer uses an attention mechanism to fuse the two types of features and dynamically allocates weights according to the contribution of each dimension feature to color difference prediction. The specific steps are as follows: perform dimensional analysis on the aligned first feature vector (128 dimensions) and the aligned second feature vector (32 dimensions) respectively, and calculate the activation intensity of each dimension feature (i.e., the sensitivity of the dimension value to color changes).

[0117] To illustrate, attention weights for each dimension are calculated through normalization based on the activation intensity of each feature dimension. The weight values ​​are between 0 and 1, and the sum of the weights of all dimensions in the same feature vector is 1. For example, if the activation intensity of a certain dimension in the first feature vector accounts for 5% of the total activation intensity of the vector, then its attention weight is 0.05; if the activation intensity of a certain dimension in the second feature vector accounts for 8% of the total activation intensity of the vector, then its attention weight is 0.08.

[0118] Since the first feature vector (e.g., 128-dimensional) and the second feature vector (e.g., 32-dimensional) have different dimensions after alignment, the first feature vector needs to be dimensionally compressed (converted from 128-dimensional to 32-dimensional through feature mapping) to make it consistent with the dimensions of the second feature vector.

[0119] Next, the two types of feature vectors are weighted and summed according to the attention weights of each dimension: for each dimension of the compressed first feature vector, the value of each dimension is multiplied by the corresponding attention weight; for each dimension of the second feature vector, the value of each dimension is also multiplied by the corresponding attention weight; finally, the weighted results of corresponding dimensions in the two feature vectors are added together to obtain a 32-dimensional fused feature vector. For example, if the value of the first dimension of the compressed first feature vector is 0.8 and the attention weight is 0.6; and the value of the first dimension of the second feature vector is 0.5 and the attention weight is 0.4, then the value of the first dimension of the fused feature vector is 0.8 × 0.6 + 0.5 × 0.4 = 0.68.

[0120] The fused feature vector obtained through the above processing not only retains the detailed features directly related to color in the spectral data (such as the reflectivity characteristics of a specific wavelength), but also integrates the key information reflecting the equipment status in the operational data (such as the influence of pressure and speed on ink transfer). At the same time, it highlights the dimensional features that contribute more to color difference prediction through the attention mechanism.

[0121] In some embodiments, after obtaining the fused feature vector, the predicted color difference value is output through the output layer of the color difference prediction model, providing a precise basis for subsequent adjustment of the flexographic printing press control parameters. Compared with a simple feature stitching method, the fusion method in this embodiment achieves deep synergy between spectral data and operational data, enabling the color difference prediction model to more effectively utilize the complementary information of the two types of data. This ensures that the color difference prediction model can adapt to complex dynamic changes during the printing process, significantly improving the accuracy of the flexographic printing press's forward-looking color control.

[0122] In some embodiments, before acquiring spectral data of the printed area in the printed material and operating data of the flexographic printing press, the method further includes:

[0123] Multiple sets of training samples are obtained, each set of training samples includes sample spectral data and sample running data, as well as the corresponding sample color difference value;

[0124] The multiple sets of training samples are divided into training datasets and validation datasets according to a preset ratio;

[0125] The training dataset is input into a pre-built initial network model. With the goal of minimizing the preset composite loss function, the parameters in the initial network model are updated to obtain the optimized network model. The initial network model includes an input layer, a one-dimensional convolutional neural network sub-model, a fully connected network sub-model, a feature fusion layer, and an output layer.

[0126] Input the validation dataset into the optimized network model and check whether the optimized network model has converged.

[0127] If the optimized network model has converged, it is determined that the optimized network model has been successfully trained, and the color difference prediction model is obtained.

[0128] In some embodiments, each training sample set includes sample spectral data, sample operation data (i.e., sample operation parameters), and corresponding sample color difference values ​​(as label data). By collecting sample data covering different operating conditions, it is ensured that the final trained color difference prediction model can adapt to diverse printing scenarios.

[0129] In some embodiments, after obtaining multiple sets of training samples, the multiple sets of training samples are divided into training datasets and validation datasets according to a preset ratio. The training dataset is used for learning and updating model parameters, and the validation dataset is used to evaluate the training effect of the model and determine whether it has converged. For example, the ratio of the training dataset to the validation dataset is 7:3 or 8:2. The partitioning process uses random sampling to ensure that the distribution characteristics of the two datasets are consistent.

[0130] In some embodiments, the structure of the initial network model is consistent with the final color difference prediction model (including an input layer, a one-dimensional convolutional neural network sub-model, a fully connected network sub-model, a feature fusion layer, and an output layer). By inputting the training dataset into the initial network model, with the goal of minimizing a preset composite loss function, the backpropagation algorithm is used to iteratively update various parameters of the model (such as convolutional kernel weights, fully connected layer weights, bias terms, etc.) to obtain the optimized network model.

[0131] In some embodiments, a validation dataset is input into the optimized network model, and the loss function value of the optimized network model on the validation dataset is calculated. By comparing the changes in the loss function value over multiple consecutive iterations (such as the difference between two adjacent iterations), it is determined whether the model has converged: if the loss function value tends to stabilize (the change is less than a preset change threshold, such as 0.001), it is determined that the optimized network model has converged, indicating that the prediction performance of the optimized network model has reached a stable state. At this time, the optimized network model is determined as the trained color difference prediction model, which can be deployed to the terminal device processor or the controller of the flexographic printing machine for predicting color difference values ​​during the printing process of the flexographic printing machine.

[0132] In some embodiments, obtaining multiple sets of training samples includes:

[0133] Set up multiple sets of different sample operation data, including sample process parameters and sample environment parameters. The sample process parameters include at least one of different ink key opening, printing speed, and current printing pressure. The sample environment parameters include at least one of different ambient temperature and ambient humidity.

[0134] For each set of sample operation data, the flexographic printing press is controlled to print the standard printed sample according to the sample operation data of that set of sample to obtain the corresponding sample printed material;

[0135] Spectral data of the printed area in the sample printed material were collected using a spectrometer.

[0136] A colorimeter was used to measure the color difference between the printed sample and the standard printed sample, which was then used as the label data corresponding to the sample running data.

[0137] The sample spectral data, sample operation data, and sample color difference values ​​of each group are combined to form a set of training samples.

[0138] In some embodiments, the quality of the training samples directly affects the prediction accuracy of the model. The sample operation data includes sample process parameters and sample environmental parameters. Orthogonal experimental design can be used to cover the parameter value ranges. For example, for sample process parameters: ink key opening is set to 10%-90% (10% interval), printing speed is set to 50m / min-200m / min (30m / min interval), and current printing pressure is set to 0.1MPa-1.0MPa (0.1MPa interval), including at least one combination of these parameters. Similarly, for sample environmental parameters: ambient temperature is set to 10℃-40℃ (5℃ interval), and ambient humidity is set to 20%RH-80%RH (10%RH interval), including at least one combination of these parameters. Finally, 100-200 representative sets of sample operation data are generated to ensure coverage of the main operating range of the flexographic printing press.

[0139] In some embodiments, for each set of sample operation data, the flexographic printing press is controlled to print standard printed samples according to that set of sample operation data. 5-10 printed materials are printed continuously under each set of sample operation data, and 3 of them with stable quality are selected as sample printed materials (excluding possible transitional deviations in the first and last sheets). A high-precision spectrometer (e.g., wavelength range 400nm-700nm) is used to collect data on the printing area of ​​the sample printed materials. The collection area is consistent with the detection area in actual production (e.g., color bar area or key pattern area). 3-5 points are collected for each sample printed material, and the average value is taken as the spectral data of that sample. A high-precision colorimeter (e.g., measurement accuracy ≤0.05) is used to measure the color difference value between the sample printed materials and the standard printed samples. The measurement area corresponds to the spectral collection area. Each sample printed material is measured 5-8 times, and the average value is taken as the sample color difference value (label data) corresponding to that set of sample operation data. The spectral data, sample operation data (sample process parameters and sample environmental parameters), and sample color difference value of each set of samples are associated and stored to form a complete set of training samples. As an example, all samples are organized in a uniform format (such as CSV format), including information such as data identifier, collection time, and values ​​of each parameter, which facilitates subsequent model training.

[0140] Through the above training process, the color difference prediction model finally trained can fully learn the color difference variation law under different operating parameter characteristics and spectral characteristics. In practical applications, it can achieve a prediction accuracy within ±0.2, providing strong support for the forward-looking color control of flexographic printing presses.

[0141] In some embodiments, the method further includes:

[0142] Obtain the actual color difference value between the printed material and the standard printed sample;

[0143] Adjust the model parameters of the color difference prediction model based on the predicted color difference value and the actual color difference value.

[0144] In some embodiments, during actual printing, the pre-trained color difference prediction model may experience a decrease in prediction accuracy due to factors such as changes in ink properties, equipment wear, and environmental fluctuations. Therefore, this method also includes a step of dynamically adjusting the model based on actual printing data to ensure that the color difference prediction model always maintains high prediction performance. The specific implementation is as follows:

[0145] First, during the flexographic printing process, the actual color difference value between the printed material and a standard printed sample is acquired simultaneously. This serves as the benchmark data for optimizing the color difference prediction model. For example, an online colorimeter can be deployed on the same side as the spectral acquisition device (such as a spectrometer), with the detection area of ​​the online colorimeter completely overlapping with the acquisition area of ​​the spectral acquisition device (such as the same color bar area or pattern area). This ensures that the actual color difference value and the spectral data originate from the same printing area or printing position. Furthermore, the acquisition frequency of the actual color difference value is consistent with the output frequency of the predicted color difference value (e.g., once every 0.1 seconds). Each acquisition involves three consecutive measurements of the same area, and the average value is taken as the actual color difference value at that moment, reducing single measurement errors. Additionally, each actual color difference value is stamped with the same timestamp as the predicted color difference value to ensure a strict temporal correspondence between the two.

[0146] Subsequently, based on the deviation between the predicted and actual color difference values, the parameters of the color difference prediction model are periodically fine-tuned. For example, at each preset interval (such as every 50 meters of printing or every 10 minutes), the mean absolute error (MAE) and maximum error of all predicted and actual color difference values ​​within that preset interval are calculated. For instance, if there are 100 sets of data within a preset interval, the mean absolute error is (Σ|predicted value - actual value|) / 100, and the maximum error is the maximum value of |predicted value - actual value| within that interval.

[0147] As an illustration, when the error exceeds the allowable range (mean absolute error exceeds 0.2 or maximum error exceeds 0.5), the model parameter adjustment process for the color difference prediction model is triggered; if the error is within the allowable range (for example, mean absolute error ≤ 0.2 and maximum error ≤ 0.5), the model parameters for the color difference prediction model will not be adjusted for the time being.

[0148] In some embodiments, the main source of error can be determined by comparing the spectral feature errors of different wavelength bands with the prediction errors of the operating parameters. For example, if the prediction deviation related to the red light band (600nm-650nm) is large, the weights of the corresponding convolutional layers in the one-dimensional convolutional neural network sub-model are adjusted accordingly; if the prediction deviation caused by changes in environmental humidity is significant, the weights of the dimension corresponding to the humidity parameter in the fully connected network sub-model are increased.

[0149] In some embodiments, but not limited to, incremental learning can be used to fine-tune the model parameters based on the latest error data, rather than retraining the entire color difference prediction model. Specifically, a small optimization dataset is formed by combining the predicted color difference values, actual color difference values, and corresponding input data (spectral data, running data) within the preset period. The parameters of the output layer, feature fusion layer, and the last layer of the corresponding sub-model are updated using the backpropagation algorithm. The update magnitude is 1 / 10 of that at the initial training stage to avoid over-adjustment that could lead to a decrease in model stability.

[0150] Furthermore, the adjusted model parameters must meet the value range constraints (e.g., convolution kernel weights limited to ±1) to ensure that the predicted color difference value output by the model is always within a reasonable range (0-5). After adjusting the model parameters, continue to monitor the prediction error in the subsequent printing process. If the average absolute error of two consecutive preset periods is less than 0.2, the adjustment is considered effective; if the error still does not improve, increase the size of the optimization dataset (e.g., expand to the data of the first 3 periods) and repeat the parameter fine-tuning steps.

[0151] Through the aforementioned dynamic adjustment mechanism, the color difference prediction model can adapt to various dynamic changes in the printing process, significantly improving the reliability and consistency of color control in flexographic printing presses: for changes in spectral characteristics caused by batch ink replacement, relevant parameters can be extracted by adjusting spectral features to maintain the model's prediction accuracy; for wear of the pressure mechanism after long-term operation of the flexographic printing press, the model can compensate for the impact of decreased mechanical performance by optimizing the relevant weights of operating parameters; for seasonal changes in the workshop environment (such as high humidity in summer and low temperature in winter), the model can dynamically balance the weights of environmental parameters in feature fusion to ensure prediction stability.

[0152] In some embodiments, adjusting the model parameters of the color difference prediction model based on the predicted color difference value and the actual color difference value includes:

[0153] The error between the predicted color difference value and the actual color difference value is calculated using a preset composite loss function.

[0154] The error value is fed into the color difference prediction model through the backpropagation algorithm, and the convolution kernel parameters, pooling layer parameters, weight matrix and bias terms of the one-dimensional convolutional neural network sub-model, as well as the fully connected network sub-model, are updated in turn.

[0155] After each round of parameter updates, the spectral data and running data of the current batch are used as test samples and input into the updated color difference prediction model to verify whether the error between the predicted color difference value and the actual color difference value output by the updated color difference prediction model is less than the preset error threshold.

[0156] If the error value is less than the preset error threshold, the adjustment of the model parameters of the color difference prediction model will be stopped.

[0157] If the error value is greater than or equal to the preset error threshold, the calculation of the error value between the predicted color difference value and the actual color difference value and subsequent steps are repeated until the error value between the predicted color difference value and the actual color difference value output by the updated color difference prediction model is less than the preset error threshold.

[0158] In this application, the process of adjusting the parameters of the color difference prediction model based on the predicted color difference value and the actual color difference value is a fine-grained optimization process based on real-time error feedback. Closed-loop iteration ensures that the model maintains high-precision prediction capabilities at all times. The specific implementation method is as follows:

[0159] First, the error between the predicted color difference value and the actual color difference value is calculated using a preset composite loss function. This function comprehensively considers multi-dimensional error information to ensure the comprehensiveness of model adjustment.

[0160] In some embodiments, the composite loss function is L = α·MSE(ΔE) + β·KL(Spectral), where MSE(ΔE) is the numerical deviation between the predicted color difference value and the actual color difference value, KL(Spectral) is the spectral distribution difference between the predicted color difference value and the actual color difference value, and α and β are weighting coefficients that satisfy α + β = 1.

[0161] The composite loss function comprises two core components: first, the numerical deviation between the predicted and actual color difference values ​​(such as mean square error and absolute error); and second, the difference in their distribution at the spectral feature level (such as the morphological deviation of the spectral curve). By reasonably allocating weights (e.g., 60% for numerical deviation and 40% for spectral feature deviation), the accuracy of the color difference numerical prediction is ensured while also taking into account the fitting accuracy to the essential color features (spectrum).

[0162] For each set of synchronized predicted color difference values ​​and actual color difference values, the composite loss function is substituted to calculate the single-set error. Then, the average value of all single-set errors within a preset period (such as 10 minutes or 50 sets of data) is taken to obtain the overall error value for that preset period, which serves as the basis for adjusting the model parameters.

[0163] Next, the calculated error values ​​are fed into the color difference prediction model via backpropagation. Model parameters are updated in a specific order to ensure targeted and effective adjustments. Specifically, the parameters of the one-dimensional convolutional neural network sub-model are updated first, including the convolution kernel parameters (affecting the accuracy of spectral feature extraction) and pooling layer parameters (affecting the effectiveness of feature selection). For example, if the prediction deviation in the red band is large, the weights of the convolution kernels capturing red band features are adjusted. Then, the parameters of the fully connected network sub-model are updated, including the weight matrices of each layer (affecting the modeling accuracy of the relationship between operating parameters and color) and bias terms (affecting the baseline values ​​for feature transformation). For example, if changes in environmental humidity cause significant prediction deviations, the adjustment magnitude of the weights corresponding to the humidity parameter is increased.

[0164] In some embodiments, a small-step iteration strategy can be adopted during the model parameter update process, with each parameter update being 1 / 20 to 1 / 10 of the initial training value, to avoid model instability caused by large adjustments. For example, the single adjustment of convolutional kernel parameters is controlled within ±0.01, and the adjustment of fully connected layer weights is controlled within ±0.005.

[0165] Furthermore, after each round of parameter updates, the adjustment effect is immediately verified through test samples to ensure continuous optimization of model performance. For example, the latest spectral data and operational data (such as the most recent 20 sets of data) collected during the current batch of printing are used as test samples. These data were not included in this round of model parameter updates and can objectively reflect the real-time prediction capability of the color difference prediction model. The test samples are input into the updated color difference prediction model, and the error value between the output predicted color difference value and the corresponding actual color difference value is calculated (using the same composite loss function as in step one), and compared with a preset error threshold (such as 0.15).

[0166] Finally, based on the verification results, a decision is made on whether to continue adjustments, forming a closed-loop optimization: If the error value of the test sample is less than the preset error threshold (e.g., 0.15), it indicates that the model parameter adjustment is effective and the current model meets the accuracy requirements. The current round of parameter adjustment is stopped, and the current model parameters are retained for continued use in color difference prediction during the printing process. If the error value of the test sample is greater than or equal to the preset error threshold, the error calculation-backpropagation update-effect verification process is repeated: the loss function is recalculated based on the latest error, and the model parameters are updated again through backpropagation (the adjustment range can be dynamically increased according to the error magnitude; for example, for every 0.1 increase in error, the adjustment range is increased by 1.2 times). The updated latest batch data is used as the test sample for verification until the error value is less than the preset error threshold.

[0167] Through the above-described iterative adjustment of model parameters, the color difference prediction model can quickly adapt to dynamic changes in the printing process (such as ink batch replacement or slight equipment wear), keeping the prediction error stable within ±0.15 for a long time. Compared with the fixed parameter model, the consistency of printed colors is improved by more than 30%, significantly reducing the scrap rate caused by color deviation.

[0168] In some embodiments, adjusting the current control parameters of the flexographic printing machine based on the predicted color difference value and a preset color difference threshold includes:

[0169] If the predicted color difference value is greater than the preset color difference threshold, the predicted color difference value and the current control parameters of the flexographic printing machine are input into the parameter offset prediction model to obtain the control parameter offset output by the parameter offset prediction model with the preset color difference value range as the target.

[0170] The control parameter offset is sent to the flexographic printing press so that the flexographic printing press adjusts the current control parameters and prints the standard printed sample according to the adjusted control parameters.

[0171] In some embodiments, the parameter offset prediction model can output a control parameter offset based on the input color difference value and the current control parameters of the flexographic printing press, with a preset color difference value range as the target. This control parameter offset (e.g., pressure offset, ink key opening offset, speed offset) can serve as an indication for the flexographic printing press to adjust the current control parameters, so that when the flexographic printing press prints according to the adjusted current control parameters, the color difference between the printed material and the standard printed sample falls within the preset color difference value range.

[0172] In some embodiments, parameter shift prediction models can be constructed based on large models, convolutional neural networks, or support vector machines. The prediction accuracy or robustness of various models can be evaluated, and the model with the best evaluation results can be selected for subsequent training and validation steps.

[0173] For example, the steps for building a parameter offset prediction model are as follows:

[0174] Step 1: Build an initial model. For example, you can choose a large model as the initial model.

[0175] Step 2: Obtain multiple sets of training samples.

[0176] In some embodiments, obtaining multiple sets of training samples includes: setting multiple sample control parameters; for each sample control parameter, controlling the flexographic printing press to print a training standard printing sample according to the sample control parameter, and obtaining the sample color difference value between the training printed material printed by the flexographic printing press according to the sample control parameter and the training standard printing sample; determining the sample color difference value and the sample control parameter as sample data; and combining the sample data corresponding to each sample control parameter to obtain multiple sets of training samples.

[0177] In some embodiments, the flexographic printing press may include multiple printing units, each of which may be configured with sample control parameters. Each printing unit prints a training standard printed sample based on the sample control parameters, and the color difference between the printed training print and the training standard printed sample is used as the sample color difference value.

[0178] As an example, a flexographic printing press may include four printing units. Sample control parameters may include sample printing pressure, sample printing speed, sample ink viscosity, sample ink pH value, sample ink temperature, and sample ink drying time. Each printing unit can be controlled to print training standard printed samples according to the sample control parameters corresponding to that printing unit. The color difference value between the printed training print and the training standard printed sample is used as the sample color difference value. Each printing unit can obtain multiple (e.g., 20, 50, or 100) sample color difference values ​​according to the sample control parameters corresponding to that printing.

[0179] In addition, the environmental parameters of the printing press's operating environment can be used as sample environmental parameters, which may include the sample temperature and sample humidity of the printing press's operating environment. Each printing unit prints the training standard printed sample based on the sample control parameters and sample environmental parameters, and the color difference value between the printed training print and the training standard printed sample is used as the sample color difference value.

[0180] Step 3: Divide the multiple sets of training samples into training datasets and validation datasets according to a preset ratio.

[0181] In some embodiments, multiple sets of training samples can be divided into training datasets and validation datasets by randomly sampling them proportionally. Alternatively, multiple sets of training samples can be divided into training datasets, validation datasets, and test datasets. For example, they can be divided according to a preset ratio such as 3:1:1 or 7:2:1.

[0182] The training dataset is used to update the parameters of the initial model to minimize the loss function. The validation dataset is used to fine-tune the optimized network model and select the optimal parameters. The test dataset is used to evaluate the generalization ability and performance of the optimized network model.

[0183] Step 4: Input the training dataset into the initial model and update the parameters of the initial model with the goal of minimizing the preset loss function to obtain the optimized network model.

[0184] In some embodiments, the loss function can be the cross-entropy loss function or the mean squared error loss function.

[0185] Step 5: Input the validation dataset into the optimized network model and check whether the optimized network model has converged.

[0186] Step 6: If the optimized network model has converged, determine that the optimized network model has been trained and obtain the parameter offset prediction model.

[0187] In some embodiments, a validation dataset can be used to determine whether the optimized network model has converged. For example, the validation dataset can be input into the optimized network model to obtain its current loss function value; then, the loss function value of the optimized network model in the previous iteration (with the input data still being the validation dataset) can be obtained; by calculating the difference between the current loss function value and the loss function value in the previous iteration, and comparing this difference with a preset convergence threshold, it can be determined whether the optimized network model has converged.

[0188] For example, if the difference in the loss function value between two iterations of the optimized network model after inputting the validation dataset is less than the preset convergence threshold, then the optimized network model can be determined to have converged.

[0189] In one embodiment, step 6 above includes: if the optimized network model has converged, evaluating the prediction performance of the optimized network model using a test dataset, wherein the test dataset is extracted from multiple training samples; if the prediction performance of the optimized network model meets the preset index, determining that the optimized network model has been trained and obtaining the parameter offset prediction model.

[0190] In some embodiments, the predictive performance of the optimized network model can be evaluated using a test dataset, and the predictive performance of the optimized network model can be comprehensively measured using multiple metrics (such as accuracy, recall, and F1 score). For example, if the accuracy is greater than 0.95, it can be determined that the optimized network model has been successfully trained.

[0191] In some embodiments, during the training of the parameter shift prediction model, data preprocessing techniques can be used to convert different data into a standardized format, and interpolation algorithms and smoothing filtering techniques can be applied to ensure data consistency. During the construction of the parameter shift prediction model, multi-layer neural networks or ensemble learning algorithms can be employed, combined with various feature extraction techniques, to enhance the model's adaptability to diverse input data. Simultaneously, transfer learning can be used to accelerate the training process and improve the model's generalization ability. Furthermore, cross-validation can be implemented during training to evaluate the performance of the parameter shift prediction model under various data inputs, and data augmentation techniques can be used to generate synthetic data to further enhance adaptability. The evaluation mechanism measures model performance using multiple metrics (e.g., accuracy, recall, and F1 score) and provides visualization tools to help analyze the differences in model performance across various data types.

[0192] In one embodiment, when the environmental parameters of the flexographic printing machine's operating environment are used during the training of the parameter offset prediction model, inputting the color difference value and the current control parameters of the flexographic printing machine into the parameter offset prediction model to obtain the control parameter offset output by the parameter offset prediction model with a preset color difference value range may include: obtaining the current environmental parameters, including the current temperature and current humidity of the flexographic printing machine's operating environment; inputting the color difference value, current control parameters, and current environmental parameters into the parameter offset prediction model to obtain the control parameter offset output by the parameter offset prediction model with a preset color difference value range and recommended environmental parameters, including recommended temperature and recommended humidity, which are used to instruct the operator to adjust the environmental parameters of the flexographic printing machine's operating environment.

[0193] In some embodiments, the parameter offset prediction model can output not only the control parameter offset but also recommended environmental parameters. Operators can then adjust the environmental parameters of the flexographic printing machine's operating environment based on these recommended parameters.

[0194] By constructing and training a parameter offset prediction model, when the color difference between the printed material obtained by the flexographic printing press and the standard printed sample is large (greater than the preset color difference threshold), the parameter offset prediction model can predict the control parameter offset. This control parameter offset can be used as a basis for instructing the flexographic printing press to adjust the current control parameters so that when the flexographic printing press prints according to the adjusted current control parameters, the color difference between the printed material and the standard printed sample can fall within the preset color difference range.

[0195] In this way, by monitoring the color difference between the printed material and the standard printed sample in real time, the parameter offset prediction model predicts the control parameter offset when the color difference is large. This allows the flexographic printing press to adjust the current control parameters based on the control parameter offset and print the standard printed sample according to the adjusted current control parameters. This timely response to situations where the color difference is large during the printing process of the flexographic printing press reduces the color difference between the printed material and the standard printed sample, thereby improving the image quality of the printed material.

[0196] In some embodiments, sending a control parameter offset to the flexographic printing press to cause the flexographic printing press to adjust the current control parameters includes at least one of the following:

[0197] If the current control parameter is printing pressure and the control parameter offset is pressure offset, then the pressure offset is sent to the controller of the flexographic printing press, so that the controller drives the servo motor of the pressure adjustment mechanism of the flexographic printing press according to the pressure offset, and adjusts the current printing pressure applied by the pressure adjustment mechanism to the printing plate cylinder of the flexographic printing press.

[0198] If the current control parameter is the ink key opening and the control parameter offset is the ink key opening offset, then the ink key opening offset is sent to the controller of the flexographic printing press so that the controller drives the ink key actuator of the flexographic printing press to adjust the current ink key opening of the flexographic printing press.

[0199] If the current control parameter is the printing speed and the control parameter offset is the speed offset, then the speed offset is sent to the controller of the flexographic printing press so that the controller can adjust the current printing speed of the flexographic printing press.

[0200] In the flexographic printing process, many factors significantly affect the image quality of the printed material. Besides the most direct factors like printing pressure and speed, other factors include the ambient temperature, humidity, ink viscosity, ink temperature, and ink drying time. The parameter offset prediction model can output a corresponding control parameter offset based on the input color difference value and the current control parameters. This allows the flexographic printing machine to adjust the current control parameters according to the offset, enabling it to print materials that meet preset standards (i.e., the color difference value falls within the preset color difference range). Alternatively, it can allow the flexographic printing machine to effectively control the printing effect by adjusting other control parameters when one control parameter fails to meet the requirements.

[0201] In one embodiment, the current control parameter may include the current printing pressure; the control parameter offset may include a pressure offset; the above-mentioned sending the control parameter offset to the flexographic printing machine to adjust the current control parameter may include: sending the pressure offset to the controller of the flexographic printing machine so that the controller drives the servo motor of the pressure regulating mechanism of the flexographic printing machine according to the pressure offset, thereby adjusting the current printing pressure applied by the pressure regulating mechanism. For example, the controller controls the pressure regulating mechanism of the flexographic printing machine according to the pressure offset, thereby adjusting the first printing pressure configured by the pressure regulating mechanism.

[0202] In some embodiments, the current control parameter may further include the ink key opening, and the control parameter offset may further include the ink key opening offset. Sending the control parameter offset to the flexographic printing press so that the flexographic printing press can adjust the current control parameter may further include: sending the ink key opening offset to the controller of the flexographic printing press so that the controller can drive the ink key actuator of the flexographic printing press to adjust the current ink key opening of the flexographic printing press.

[0203] In some embodiments, the current control parameter may further include the current printing speed; the control parameter offset may further include a speed offset; the above-mentioned sending the control parameter offset to the flexographic printing press to make the flexographic printing press adjust the current control parameter may further include: sending the speed offset to the controller of the flexographic printing press to make the controller adjust the current printing speed of the flexographic printing press, for example, controlling the speed adjustment mechanism of the flexographic printing press according to the speed offset, thereby adjusting the current printing speed configured by the speed adjustment mechanism.

[0204] In some embodiments, the current control parameter may further include the current ink viscosity; the control parameter offset may further include the ink volume offset; the above-mentioned sending the control parameter offset to the flexographic printing press to make the flexographic printing press adjust the current control parameter may further include: inputting the ink volume offset into the controller of the flexographic printing press to make the controller control the viscosity adjustment mechanism of the flexographic printing press according to the ink volume offset, thereby adjusting the current ink viscosity configured by the viscosity adjustment mechanism.

[0205] In some embodiments, the current control parameter may further include the current ink pH value; the control parameter offset may further include a pH value offset; the above-mentioned sending the control parameter offset to the flexographic printing press to make the flexographic printing press adjust the current control parameter may further include: inputting the pH value offset into the controller of the flexographic printing press to make the controller control the ink pH value adjustment mechanism of the flexographic printing press according to the pH value offset, thereby adjusting the current ink pH value configured by the ink pH value adjustment mechanism.

[0206] In some embodiments, the current control parameter may further include the current ink temperature; the control parameter offset may further include the ink temperature offset; the above-mentioned sending the control parameter offset to the flexographic printing press to make the flexographic printing press adjust the current control parameter may further include: inputting the ink temperature offset into the controller of the flexographic printing press to make the controller control the ink temperature adjustment mechanism of the flexographic printing press according to the ink temperature offset, thereby adjusting the current ink temperature configured by the ink temperature adjustment mechanism.

[0207] In some embodiments, the current control parameter may further include the current ink drying time; the control parameter offset may further include a drying time offset; the above-mentioned sending the control parameter offset to the flexographic printing press to make the flexographic printing press adjust the current control parameter may further include: inputting the drying time offset into the controller of the flexographic printing press to make the controller control the ink drying adjustment mechanism of the flexographic printing press according to the drying time offset, thereby adjusting the current ink drying time configured by the ink drying adjustment mechanism.

[0208] In the above embodiments, the current printing pressure can be a fixed value, for example, the current printing pressure can be aN. The control parameter offset can be a fixed value or a percentage, for example, the control parameter offset can be +bN or -c%. At this time, the controller can generate a control command based on the pressure offset and send the control command to the pressure regulating mechanism, thereby adjusting the current printing pressure aN configured by the pressure regulating mechanism. For example, the adjusted current printing pressure can be (a+b)N or a*(1-c%)N.

[0209] The adjustment process for the current printing speed, current ink viscosity, current ink pH value, current ink temperature, and current ink drying time can be similar to that for the current printing pressure, and will not be elaborated further in this application.

[0210] In one embodiment, after adjusting the current control parameters by the control parameter offset, the flexographic printing press can be controlled to print the standard printing sample again, and the color difference value between the latest printed material and the standard printing sample can be determined in the above manner. The color difference value is then compared with a preset color difference threshold, and the above process is repeated until a printed material with a color difference value less than the preset color difference threshold is obtained.

[0211] For example, the printing task of a flexographic printing press for a standard printed sample can be divided into multiple batches. After each batch of printing is completed, the color difference between the printed material of the current batch and the standard printed sample is detected. If the color difference value is greater than a preset color difference threshold, a control parameter offset is output through a parameter offset prediction model, so that the flexographic printing press can adjust the current control parameters based on the control parameter offset. In the next batch of printing, the flexographic printing press is controlled to perform the printing task based on the adjusted current control parameters, and the color difference value between the printed material of the next batch and the standard printed sample is determined again in the above manner. The above process is repeated until a printed material with a color difference value less than the preset color difference threshold is obtained.

[0212] The printing control method for a flexographic printing machine provided in this application can be applied to a printing control system for a flexographic printing machine. The exemplary application architecture of the printing control system for a flexographic printing machine provided in this application is described below; see [link to relevant documentation]. Figure 3 , Figure 3 This is a schematic diagram of the printing control system of a flexographic printing machine according to an embodiment of this application. The printing control system 10 of the flexographic printing machine includes a terminal device 100 and a flexographic printing machine 110. The terminal device 100 can establish a communication connection with the flexographic printing machine 110 through a wired network or a wireless network.

[0213] In some embodiments, the terminal device 100 may be configured to acquire spectral data of the printed area in a printed material and operating data of a flexographic printing press, the printed material being obtained by printing a standard printed sample using the flexographic printing press, the operating data including the process parameters and environmental parameters of the flexographic printing press.

[0214] In some embodiments, the terminal device 100 may be configured to input spectral data and operating data into a color difference prediction model to obtain a predicted color difference value output by the color difference prediction model. This predicted color difference value is used to characterize the deviation between the predicted colorimetric value and the target colorimetric value of the printed matter to be printed by the flexographic printing press in a future time period.

[0215] In some embodiments, the terminal device 100 may be configured to adjust the current control parameters of the flexographic printing machine based on the predicted color difference value and a preset color difference threshold.

[0216] In some embodiments, after the flexographic printing press prints a standard printed sample, a spectrometer deployed next to the take-up end of the flexographic printing press can be used to collect spectral data (e.g., local spectral data, local spectral reflectance, local spectral reflectance curve, etc.). The measurement probe of the spectrometer is facing the transmission path of the printed material, and the vertical distance between the center of the probe and the surface of the printed material is controlled within the range of 5cm-10cm to avoid collisions caused by being too close or measurement errors caused by being too far away.

[0217] In some embodiments, such as Figure 4 As shown, the process parameters and environmental parameters of the flexographic printing machine 110 are collected by corresponding sensors 120 to obtain the machine's operating data. Furthermore, all process and environmental parameters collected by the sensors are transmitted to the processor 101 of the terminal device 100 (industrial computer or server) via an industrial bus. For example, the spectrometer 121 in sensor 120 collects spectral data of the printing area in the printed material. In addition, sensor 120 may also include: a pressure sensor 122, an ink key position sensor 123, a speed sensor 124, a temperature sensor 125, and a humidity sensor 126.

[0218] In some embodiments, for the current printing pressure in the process parameters, a pressure sensor 122 can be installed beside the contact area between the printing plate cylinder and the impression cylinder to collect the pressure applied to the printing plate cylinder by the pressure regulating mechanism in real time; for the ink key opening in the process parameters, an ink key position sensor 123 can be installed on the actuator of each ink key to calculate the ink key opening by detecting the displacement of the ink key baffle; for the printing speed in the process parameters, a speed sensor 124 can be installed on the output shaft of the flexographic printing machine's main motor. For example, the speed sensor 124 can be the main motor encoder, and the printing speed can be calculated by the speed and the cylinder circumference. As another example, a temperature sensor 125 is used to collect the temperature of the operating environment of the flexographic printing machine 110, and a humidity sensor 126 is used to collect the humidity of the operating environment of the flexographic printing machine 110.

[0219] In some embodiments, the color difference prediction model is a pre-trained dual-branch neural network model, which is used to fuse spectral data and running data to obtain a fused feature vector, so as to output the predicted color difference value of the printed matter in the future time period. The predicted color difference value is used to characterize the deviation between the predicted color value and the target color value of the printed matter printed by the flexographic printing press in the future time period (e.g., 50ms-100ms, the specific duration is determined by the printing speed, the higher the speed, the shorter the prediction duration).

[0220] In some embodiments, the color difference prediction model 102 and the parameter offset prediction model 103 can be stored in the memory of the terminal device 100, or the terminal device 100 and the color difference prediction model can be mounted together in a server. During the printing process of the flexographic printing press 110, many factors significantly affect the image quality of the printed material. Besides the most direct influence of printing pressure, factors such as the working environment's temperature, humidity, ink viscosity, and ink temperature also play a role. A color difference prediction model (e.g., a large model) can be constructed. The color difference values ​​between printed materials and standard printed samples obtained under various control parameters and environmental parameters are input into the color difference prediction model. The initial network model is trained with the goal of minimizing a preset composite loss function (e.g., cross-entropy loss function or mean squared error loss function) to obtain a fully trained color difference prediction model. This initial network model includes an input layer, a one-dimensional convolutional neural network sub-model, a fully connected network sub-model, a feature fusion layer, and an output layer.

[0221] In other embodiments, the color difference prediction model can be set up in the cloud or on a cloud server, and the terminal device can be configured to use or train the color difference prediction model by calling cloud service interfaces or edge computing nodes.

[0222] In some embodiments, after the spectral data and the flexographic printing machine's operating data are input into the color difference prediction model through the input layer, the data are processed by a one-dimensional convolutional neural network (1D-CNN) sub-model and a fully connected network sub-model to obtain a first feature vector and a second feature vector, respectively. The color difference prediction model then fuses the first and second feature vectors to obtain a fused feature vector, and calculates and outputs the predicted color difference value based on the fused feature vector. The terminal device 100 transmits the predicted color difference value to the flexographic printing machine's controller in real time and stores it in a local database for subsequent parameter adjustments and model optimization.

[0223] In some embodiments, the preset color difference threshold can be set according to the quality level of the printed material, and the control parameters of the flexographic printing machine can be precisely adjusted through feedforward control logic to avoid the expansion of printing color deviation.

[0224] As an illustration, the preset color difference threshold is determined based on the application scenario of the printed material: For high-precision printed materials (such as brand logos on food packaging and cosmetic labels), the preset color difference threshold is set to ΔE = 0.8, that is, when the predicted color difference value exceeds 0.8, the current control parameters of the flexographic printing machine need to be adjusted; for medium-precision printed materials (such as packaging for ordinary daily necessities), the preset color difference threshold is set to ΔE = 1.2; for low-precision printed materials (such as corrugated cardboard boxes), the preset color difference threshold is set to ΔE = 1.5.

[0225] In some embodiments, the preset color difference threshold is stored in the parameter configuration module of the processor of the terminal device, and can be modified according to actual production needs through the human-machine interface. The processor of the terminal device compares the predicted color difference value output by the color difference prediction model with the preset color difference threshold in real time: if the predicted color difference value is less than or equal to the preset color difference threshold, it means that the printing color deviation in the future period is within an acceptable range, and the flexographic printing press is controlled to maintain the current control parameters (including the current pressure adjustment, the current ink key opening, and the current printing speed) and continue to operate; if the predicted color difference value is greater than the preset color difference threshold, the current control parameters of the flexographic printing press are adjusted.

[0226] In some embodiments, the terminal device 100 is further configured to input the color difference value and the current control parameters of the flexographic printing machine 110 into a parameter offset prediction model when the color difference value is greater than a preset color difference threshold, so as to obtain the control parameter offset output by the parameter offset prediction model with the preset color difference value range as the target.

[0227] In some embodiments, the preset color difference threshold can be set based on the historical operating data of the flexographic printing machine 110 and user expectations. For example, the preset color difference threshold can be set according to the user's requirements for the image quality of the printed matter obtained by the flexographic printing machine 110. When the color difference requirement of the printed matter is high, the preset color difference threshold can be set to a smaller value, such as 1, 2, or 3. When the color difference requirement of the printed matter is low, the preset color difference threshold can be set to a larger value, such as any value from 4 to 7. This application embodiment does not limit this.

[0228] In some embodiments, the parameter offset prediction model can be stored in the memory of the terminal device 100, or the terminal device 100 and the parameter offset prediction model can be mounted together on a server. During the printing process of the flexographic printing machine 110, many factors significantly affect the image quality of the printed material. Besides the most direct influence of printing pressure, factors such as the temperature, humidity, ink viscosity, and ink temperature of the working environment also play a role. A parameter offset prediction model (e.g., a large model) can be constructed by inputting the color difference values ​​between printed materials obtained under a large number of different control parameters and environmental parameters and a standard printed sample into the parameter offset prediction model. The model is trained with the goal of minimizing a preset loss function (e.g., cross-entropy loss function or mean squared error loss function) to obtain a fully trained parameter offset prediction model.

[0229] In other embodiments, the parameter offset prediction model can be set up in the cloud or on a cloud server, and the terminal device can be configured to use or train the parameter offset prediction model by calling cloud service interfaces or edge computing nodes.

[0230] In some embodiments, a preset color difference value range can be set for the parameter offset prediction model. For example, the preset color difference value range can be set based on a preset color difference threshold. When the preset color difference threshold is 1, the preset color difference value range can be set to (0, 1). After receiving the color difference value sent by the terminal device 100 and the current control parameters of the flexographic printing machine 110, the parameter offset prediction model can output a control parameter offset with the preset color difference value range as the target. Subsequently, the control parameter offset can be used to adjust the current control parameters so that when the flexographic printing machine 110 prints according to the adjusted current control parameters, the color difference value between the printed material and the standard printed sample falls within the preset color difference value range.

[0231] Subsequently, the terminal device 100 can send the control parameter offset to the flexographic printing machine 110 so that the flexographic printing machine 110 can adjust the current control parameters and print the standard printing sample according to the adjusted current control parameters.

[0232] In other embodiments, the operator can set a target color difference value and input this target color difference value into the parameter offset prediction model via the terminal device 100. After the terminal device 100 inputs the color difference value, the target color difference value, and the current control parameters of the flexographic printing machine 110 into the parameter offset prediction model, the parameter offset prediction model can predict and output a control parameter offset with the target color difference value as the target. Subsequently, this control parameter offset can be used to adjust the current control parameters so that when the flexographic printing machine 110 prints according to the adjusted current control parameters, the color difference value between the printed material and the standard printed sample matches the target color difference value.

[0233] In this way, operators can subjectively adjust the image color of the printed material, thereby adjusting the image quality of the printed material obtained by the flexographic printing machine in real time, which can improve the control and maintenance efficiency of the flexographic printing machine.

[0234] In some embodiments, the flexographic printing machine 110 can be configured to receive control parameter offsets sent by the terminal device 100, adjust the current control parameters based on the control parameter offsets, and print standard printing samples according to the adjusted current control parameters.

[0235] For example, the flexographic printing machine 110 may include a controller 111 (e.g., a programmable logic controller (PLC)), a pressure regulating mechanism 112, a speed regulating mechanism 113, and an ink key regulating mechanism 114; the current control parameters may include the current printing pressure, the current printing speed, and the current ink key opening; the control parameter offsets include pressure offset, speed offset, and ink volume offset.

[0236] In one embodiment, the current control parameters may include, in addition to the current printing pressure, current printing speed, and current ink viscosity, a first ink pH value, a first ink temperature, and a first ink drying time; the control parameter offset may include, in addition to the pressure offset, speed offset, and ink volume offset, pH value offset, ink temperature offset, and drying time offset; correspondingly, the flexographic printing press 110 may also include an ink pH value adjustment mechanism. Figure 3 or Figure 4 (not shown in the image) Ink temperature control mechanism ( Figure 3 or Figure 4 (not shown in the image) and ink drying adjustment mechanism ( Figure 3 or Figure 4 (Not shown in the image).

[0237] In one embodiment, the printing control system 10 of the flexographic printing machine may further include a sensor 120. The sensor 120 establishes a communication connection with the terminal device 100 through a wired network or a wireless network. The sensor 120 is configured to collect environmental parameters of the operating environment of the flexographic printing machine 110 and send the environmental parameters to the terminal device 100.

[0238] As an example, the processor 101 of the terminal device 100 can also be configured to obtain environmental parameters (including current temperature and current humidity) of the operating environment of the flexographic printing machine 110 from the temperature sensor 125 and the humidity sensor 126, and input the color difference value, environmental parameters, and current control parameters into the parameter offset prediction model to obtain the control parameter offset output by the parameter offset prediction model with a preset color difference value range as the target.

[0239] In addition to outputting control parameter offsets, the parameter offset prediction model can also output recommended environmental parameters, including recommended temperature and humidity. These parameters instruct the operator to adjust the environmental parameters of the flexographic printing machine's operating environment. The operator can then adjust the environmental parameters of the flexographic printing machine 110's operating environment based on these recommended parameters (e.g., adjusting the air conditioning temperature, turning the humidifier on / off, etc.). Furthermore, the controller 111 can also send the recommended environmental parameters to the intelligent environmental control system of the flexographic printing machine 110's operating environment, enabling the intelligent environmental control system to adjust the environmental parameters of the flexographic printing machine 110's operating environment based on the recommended parameters.

[0240] It should be noted that, although the above embodiments describe that the current control parameters may include the current printing pressure, current printing speed, current ink viscosity, first ink pH value, first ink temperature, and first ink drying time, and the environmental parameters may include the current temperature and current humidity, the current control parameters and environmental parameters may be increased or decreased according to the usage environment and specific configuration of the flexographic printing machine. For example, the current control parameters may also include the ink layer thickness and solvent ratio of the printing ink, and the environmental parameters may also include ambient light and cleanliness. The embodiments of this application do not limit the specific composition of the current control parameters and environmental parameters.

[0241] Furthermore, the current control parameters described in the above embodiments can be pre-input into the flexographic printing machine 110 by the operator, and the terminal device 100 can directly obtain the current control parameters from the flexographic printing machine 110. Alternatively, various sensors can be installed at different locations within the flexographic printing machine 110 to collect the current control parameters, and the terminal device 100 can be configured to obtain the current control parameters from these sensors. For example, if the current control parameters include the current ink viscosity, a viscosity sensor can be installed in the ink tank to collect the current ink viscosity; if the current control parameters include a first ink temperature, a temperature sensor can be installed in the ink tank to collect the first ink temperature.

[0242] In one embodiment, the printing control system 10 of the flexographic printing machine may further include a color detection device 130, which may be integrated into the terminal device 100 or may be two separate devices from the terminal device 100.

[0243] When the terminal device 100 and the color detection device 130 are two separate devices, the color detection device 130 and the terminal device 100 communicate with each other via a wired or wireless network. The terminal device 100 can control the color detection process of the color detection device 130 through the wired or wireless network, and obtain the actual color value of the target object collected by the color detection device 130. Then, the controller of the terminal device 100 can calculate the actual color value and the target chromaticity value to obtain the actual color difference value.

[0244] In some embodiments, the terminal device 100 provided in this application can be implemented in software. For example, the terminal device 100 can be software in the form of programs and plug-ins, including: an acquisition unit, a prediction unit, and an adjustment unit. These modules are logical, and therefore can be arbitrarily combined or further divided according to the functions they implement. The functions of each module will be described below.

[0245] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0246] Corresponding to the printing control method of the flexographic printing machine in the above embodiment, Figure 5 A schematic diagram of the printing control device of a flexographic printing machine according to an embodiment of this application is shown. For ease of explanation, only the parts related to the embodiment of this application are shown.

[0247] Please see Figure 5 The printing control device 50 of the flexographic printing machine includes:

[0248] The acquisition unit 51 acquires the spectral data of the printing area in the printed material and the operating data of the flexographic printing machine. The printed material is obtained by printing a standard printed sample by the flexographic printing machine, and the operating data includes the process parameters and environmental parameters of the flexographic printing machine.

[0249] Prediction unit 52 inputs spectral data and running data into the color difference prediction model to obtain the predicted color difference value output by the color difference prediction model. The predicted color difference value is used to characterize the deviation between the predicted color value and the target color value of the printed matter printed by the flexographic printing press in the future period.

[0250] Adjustment unit 53 adjusts the current control parameters of the flexographic printing machine based on the predicted color difference value and the preset color difference threshold.

[0251] In some embodiments, the color difference prediction model includes an input layer, a one-dimensional convolutional neural network sub-model, a fully connected network sub-model, a feature fusion processing layer, and an output layer. The one-dimensional convolutional neural network sub-model and the fully connected network sub-model are respectively connected to the feature fusion processing layer. The color difference prediction model processes spectral data into corresponding first feature vectors based on the one-dimensional convolutional neural network sub-model; normalizes the running data based on the fully connected network sub-model to obtain a normalized second feature vector; preprocesses the first feature vector and the second feature vector based on the feature fusion processing layer to obtain a fused feature vector, and determines the predicted color difference value based on the fused feature vector; and outputs the predicted color difference value based on the output layer. The preprocessing includes spatiotemporal alignment processing and fusion processing.

[0252] In some embodiments, the feature fusion processing layer preprocesses the first feature vector and the second feature vector to obtain a fused feature vector in the following manner: Based on the acquisition timestamp of the running data, the second feature vector and the first feature vector are spatiotemporally aligned to obtain aligned first and second feature vectors; an attention mechanism is used to fuse the aligned first and second feature vectors, and the attention weights of each dimension in the aligned first and second feature vectors are calculated; based on the attention weights of each dimension in the aligned first and second feature vectors, a weighted sum is performed on the aligned first and second feature vectors to obtain the fused feature vector.

[0253] In some embodiments, the apparatus is further configured to: acquire multiple sets of training samples, wherein each set of training samples includes sample spectral data and sample running data, as well as corresponding sample color difference values; divide the multiple sets of training samples into training datasets and validation datasets; input the training datasets into a pre-built initial network model, and update the parameters in the initial network model with the goal of minimizing a preset composite loss function to obtain an optimized network model, wherein the initial network model includes an input layer, a one-dimensional convolutional neural network sub-model, a fully connected network sub-model, a feature fusion layer, and an output layer; input the validation datasets into the optimized network model, and detect whether the optimized network model has converged; if the optimized network model has converged, determine that the optimized network model has been trained successfully, and obtain a color difference prediction model.

[0254] In some embodiments, the device acquires multiple sets of training samples, specifically for:

[0255] Multiple sets of different sample operation data are set. The sample operation data includes sample process parameters and sample environment parameters. The sample process parameters include at least one of different ink key opening, printing speed, and current printing pressure. The sample environment parameters include at least one of different ambient temperature and ambient humidity. For each set of sample operation data, the flexographic printing press is controlled to print the standard printed sample according to the sample operation data to obtain the corresponding sample printed material.

[0256] A spectrometer was used to collect the spectral data of the printed area in the sample printed material; a colorimeter was used to measure the color difference between the sample printed material and the standard printed sample, which was used as the label data corresponding to the sample running data; the sample spectral data, sample running data and sample color difference value of each group were combined to form a set of training samples.

[0257] In some embodiments, the apparatus is further configured to: obtain the actual color difference value between the printed material and the standard printed sample; and adjust the model parameters of the color difference prediction model based on the predicted color difference value and the actual color difference value.

[0258] In some embodiments, the device adjusts the model parameters of the color difference prediction model based on the predicted color difference value and the actual color difference value, specifically for:

[0259] The error between the predicted and actual color difference values ​​is calculated using a pre-defined composite loss function. This error value is then passed to the color difference prediction model via backpropagation. The convolution kernel parameters and pooling layer parameters of the one-dimensional convolutional neural network sub-model, as well as the weight matrix and bias terms of the fully connected network sub-model, are updated sequentially. After each round of parameter updates, the current batch of spectral data and running data are used as test samples and input into the updated color difference prediction model to verify whether the error between the predicted and actual color difference values ​​output by the updated model is less than a pre-defined error threshold. If the error value is less than the pre-defined error threshold, the adjustment of the model parameters is stopped. If the error value is greater than or equal to the pre-defined error threshold, the calculation of the error between the predicted and actual color difference values ​​and subsequent steps are repeated until the error between the predicted and actual color difference values ​​output by the updated model is less than the pre-defined error threshold.

[0260] In some embodiments, the adjustment unit is specifically used to: when the predicted color difference value is greater than a preset color difference threshold, input the predicted color difference value and the current control parameters of the flexographic printing machine into the parameter offset prediction model to obtain the control parameter offset output by the parameter offset prediction model with the preset color difference value range as the target; send the control parameter offset to the flexographic printing machine so that the flexographic printing machine adjusts the current control parameters and prints the standard printing sample according to the adjusted control parameters.

[0261] In some embodiments, the adjustment unit sends a control parameter offset to the flexographic printing press to adjust the current control parameter, specifically for at least one of the following: if the current control parameter is printing pressure and the control parameter offset is a pressure offset, the pressure offset is sent to the controller of the flexographic printing press, so that the controller drives the servo motor of the pressure regulating mechanism of the flexographic printing press according to the pressure offset, adjusting the current printing pressure applied by the pressure regulating mechanism to the printing plate cylinder of the flexographic printing press; if the current control parameter is ink key opening and the control parameter offset is an ink key opening offset, the ink key opening offset is sent to the controller of the flexographic printing press, so that the controller drives the ink key actuator of the flexographic printing press to adjust the current ink key opening of the flexographic printing press; if the current control parameter is printing speed and the control parameter offset is a speed offset, the speed offset is sent to the controller of the flexographic printing press, so that the controller adjusts the current printing speed of the flexographic printing press.

[0262] In some embodiments, the acquisition unit is specifically configured to: acquire spectral data of a printed area in a printed material collected by a spectrometer, wherein the printed area includes a color bar area and / or a pattern area; and receive operating data of a flexographic printing machine collected by a sensor via an industrial bus, wherein the sensor includes at least one of the following: a pressure sensor for acquiring the current printing pressure, an ink key position sensor for acquiring the ink key opening, a speed sensor for acquiring the printing speed, and a temperature and humidity sensor for acquiring environmental parameters.

[0263] It should be noted that the information interaction and execution process between the above-mentioned devices / units are based on the same concept as the method embodiments of this application. For details on their specific functions and technical effects, please refer to the method embodiments section, and they will not be repeated here.

[0264] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0265] Figure 6 This is a schematic diagram of the structure of a terminal device according to an embodiment of this application. Figure 6 As shown, the terminal device 100 includes: at least one processor 60 ( Figure 6 (Only one is shown) a processor, a memory 61, and a computer program 62 stored in the memory 61 and executable on at least one processor 60. When the processor 60 executes the computer program 62, it implements the steps in the printing control method embodiments of any of the above-described flexographic printing presses.

[0266] Terminal device 100 can be a computing device such as a desktop computer, laptop, handheld computer, or cloud server. Terminal device 100 may include, but is not limited to, a processor 60 and a memory 61. Those skilled in the art will understand that... Figure 6This is merely an example of terminal device 100 and does not constitute a limitation on terminal device 100. It may include more or fewer components than shown in the figure, or combine certain components, or different components, such as input / output devices, network access devices, etc.

[0267] The processor 60 can be a Central Processing Unit (CPU), but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.

[0268] In some embodiments, memory 61 may be an internal storage unit of terminal device 100, such as a hard disk or memory of terminal device 100. In other embodiments, memory 61 may be an external storage device of terminal device 100, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on terminal device 100. In other embodiments, memory 61 may include both internal and external storage units of terminal device 100. Memory 61 is used to store operating system, applications, boot loader, data, and other programs, such as the program code of computer program 62. Memory 61 may also be used to temporarily store data that has been output or will be output.

[0269] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps described in the above embodiments of the printing control methods for various flexographic printing machines.

[0270] This application provides a computer program product that, when run on a mobile terminal, enables the mobile terminal to execute the steps described in the printing control method embodiments of the various flexographic printing machines.

[0271] This application implements all or part of the processes in the methods of the above embodiments, which can be accomplished by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include at least: any entity or device capable of carrying the computer program code to the printing control device or terminal equipment of a flexographic printing press, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, such as a USB flash drive, a portable hard drive, a magnetic disk, or an optical disk.

[0272] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0273] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0274] In the embodiments provided in this application, it should be understood that the disclosed apparatus / network devices and methods can be implemented in other ways. For example, the apparatus / network device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0275] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs. The above embodiments are only used to illustrate the technical solutions of this application and are not intended to limit it. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. These modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A printing control method for a flexographic printing machine, wherein, include: Acquire spectral data of the printed area in the printed material and operating data of the flexographic printing machine, wherein the printed material is obtained by printing a standard printed sample by the flexographic printing machine, and the operating data includes the process parameters and environmental parameters of the flexographic printing machine. The spectral data and the operating data are input into the color difference prediction model to obtain the predicted color difference value output by the color difference prediction model. The predicted color difference value is used to characterize the deviation between the predicted color value and the target color value of the printed matter printed by the flexographic printing press in a future time period. The current control parameters of the flexographic printing machine are adjusted based on the predicted color difference value and the preset color difference threshold.

2. The printing control method for a flexographic printing machine according to claim 1, wherein, The color difference prediction model includes an input layer, a one-dimensional convolutional neural network sub-model, a fully connected network sub-model, a feature fusion processing layer, and an output layer. The one-dimensional convolutional neural network sub-model and the fully connected network sub-model are respectively connected to the feature fusion processing layer. The color difference prediction model processes the spectral data into a corresponding first feature vector based on the one-dimensional convolutional neural network sub-model; normalizes the running data based on the fully connected network sub-model to obtain a normalized second feature vector; preprocesses the first feature vector and the second feature vector based on the feature fusion processing layer to obtain a fused feature vector, determines the predicted color difference value based on the fused feature vector; and outputs the predicted color difference value based on the output layer, wherein the preprocessing includes spatiotemporal alignment processing and fusion processing.

3. The printing control method for a flexographic printing machine according to claim 2, characterized in that, The feature fusion processing layer preprocesses the first feature vector and the second feature vector in the following manner to obtain the fused feature vector: Based on the timestamp of the collected running data, the second feature vector and the first feature vector are spatiotemporally aligned to obtain the aligned first feature vector and the aligned second feature vector. An attention mechanism is used to fuse the aligned first feature vector and the aligned second feature vector, and the attention weights of each dimension in the aligned first feature vector and the aligned second feature vector are calculated. Based on the attention weights of each dimension in the aligned first feature vector and the aligned second feature vector, the aligned first feature vector and the aligned second feature vector are weighted and summed to obtain the fused feature vector.

4. The printing control method for a flexographic printing machine according to claim 1, wherein, Before acquiring the spectral data of the printed area in the printed material and the operating data of the flexographic printing press, the method further includes: Multiple sets of training samples are obtained, each set of training samples includes sample spectral data and sample running data, as well as the corresponding sample color difference value; The multiple sets of training samples are divided into training datasets and validation datasets; The training dataset is input into a pre-built initial network model. With the goal of minimizing a preset composite loss function, the parameters in the initial network model are updated to obtain an optimized network model. The initial network model includes an input layer, a one-dimensional convolutional neural network sub-model, a fully connected network sub-model, a feature fusion layer, and an output layer. Input the validation dataset into the optimized network model and check whether the optimized network model has converged; If the optimized network model has converged, it is determined that the optimized network model has been trained successfully, and the color difference prediction model is obtained.

5. The printing control method for a flexographic printing machine according to claim 4, characterized in that, The acquisition of multiple sets of training samples includes: Multiple sets of different sample operation data are set, wherein the sample operation data includes sample process parameters and sample environment parameters. The sample process parameters include at least one of different ink key opening, printing speed, and current printing pressure, and the sample environment parameters include at least one of different ambient temperature and ambient humidity. For each set of sample operation data, the flexographic printing press is controlled to print the standard printing sample according to the sample operation data of that set, so as to obtain the corresponding sample print. The spectral data of the printed area in the printed sample were collected using a spectrometer. The color difference between the printed sample and the standard printed sample is measured using a colorimeter, and the resulting color difference is used as the label data corresponding to the sample running data. The sample spectral data, sample operation data, and sample color difference values ​​of each group are combined to form a set of training samples.

6. The printing control method for a flexographic printing press according to any one of claims 1 to 5, wherein, The method further includes: Obtain the actual color difference value between the printed material and the standard printed sample; The model parameters of the color difference prediction model are adjusted based on the predicted color difference value and the actual color difference value.

7. The printing control method for a flexographic printing machine according to claim 6, wherein, The step of adjusting the model parameters of the color difference prediction model based on the predicted color difference value and the actual color difference value includes: The error value between the predicted color difference value and the actual color difference value is calculated using a preset composite loss function. The error value is passed into the color difference prediction model through the backpropagation algorithm, and the convolution kernel parameters, pooling layer parameters, weight matrix and bias terms of the one-dimensional convolutional neural network sub-model, as well as the fully connected network sub-model, are updated sequentially in the color difference prediction model. After each round of parameter updates, the spectral data and running data of the current batch are used as test samples and input into the updated color difference prediction model to verify whether the error between the predicted color difference value and the actual color difference value output by the updated color difference prediction model is less than the preset error threshold. If the error value is less than the preset error threshold, then the adjustment of the model parameters of the color difference prediction model shall be stopped. If the error value is greater than or equal to the preset error threshold, the calculation of the error value between the predicted color difference value and the actual color difference value and subsequent steps are repeated until the error value between the predicted color difference value and the actual color difference value output by the updated color difference prediction model is less than the preset error threshold.

8. The printing control method for a flexographic printing machine according to claim 7, wherein, The composite loss function is L = α·MSE(ΔE) + β·KL(Spectral), where MSE(ΔE) is the numerical deviation between the predicted color difference value and the actual color difference value, KL(Spectral) is the spectral distribution difference between the predicted color difference value and the actual color difference value, and α and β are weighting coefficients that satisfy α + β = 1.

9. The printing control method for a flexographic printing press according to any one of claims 1 to 5, wherein, The step of adjusting the current control parameters of the flexographic printing machine based on the predicted color difference value and the preset color difference threshold includes: If the predicted color difference value is greater than the preset color difference threshold, the predicted color difference value and the current control parameters of the flexographic printing machine are input into the parameter offset prediction model to obtain the control parameter offset output by the parameter offset prediction model with the preset color difference value range as the target. The control parameter offset is sent to the flexographic printing press so that the flexographic printing press adjusts the current control parameters and prints the standard printing sample according to the adjusted control parameters.

10. The printing control method for a flexographic printing machine according to claim 9, wherein, Sending the control parameter offset to the flexographic printing press so that the flexographic printing press adjusts the current control parameter includes at least one of the following: If the current control parameter is printing pressure and the control parameter offset is pressure offset, then the pressure offset is sent to the controller of the flexographic printing press, so that the controller drives the servo motor of the pressure adjustment mechanism of the flexographic printing press according to the pressure offset, and adjusts the current printing pressure applied by the pressure adjustment mechanism to the printing plate cylinder of the flexographic printing press. If the current control parameter is the ink key opening, and the control parameter offset is the ink key opening offset, then the ink key opening offset is sent to the controller of the flexographic printing press, so that the controller drives the ink key actuator of the flexographic printing press to adjust the current ink key opening of the flexographic printing press. If the current control parameter is the printing speed and the control parameter offset is the speed offset, then the speed offset is sent to the controller of the flexographic printing machine so that the controller adjusts the current printing speed of the flexographic printing machine.

11. The printing control method for a flexographic printing machine according to any one of claims 1 to 5, characterized in that, The acquisition of spectral data of the printed area in the printed material and the operating data of the flexographic printing machine includes: Acquire spectral data of the printed area in the printed material collected by a spectrometer, wherein the printed area includes a color bar area and / or a pattern area; The flexographic printing machine receives operating data from sensors via an industrial bus, wherein the sensors include at least one of the following: a pressure sensor for acquiring current printing pressure, an ink key position sensor for acquiring ink key opening, a speed sensor for acquiring printing speed, and a temperature and humidity sensor for acquiring environmental parameters.

12. A printing control device for a flexographic printing machine, wherein, include: The acquisition unit acquires spectral data of the printing area in the printed material and operating data of the flexographic printing machine, wherein the printed material is obtained by printing a standard printed sample by the flexographic printing machine, and the operating data includes the process parameters and environmental parameters of the flexographic printing machine. The prediction unit inputs the spectral data and the operating data into the color difference prediction model to obtain the predicted color difference value output by the color difference prediction model. The predicted color difference value is used to characterize the deviation between the predicted colorimetric value and the target colorimetric value of the printed matter printed by the flexographic printing press in a future time period. The adjustment unit adjusts the current control parameters of the flexographic printing machine based on the predicted color difference value and the preset color difference threshold.

13. A printing control system for a flexographic printing machine, wherein, The device includes a terminal device and a flexographic printing press, the terminal device being communicatively connected to the flexographic printing press, and the terminal device being configured to perform a printing control method for the flexographic printing press as described in any one of claims 1 to 11.

14. The printing control system of the flexographic printing machine according to claim 13, wherein, The terminal device is configured to, when the predicted color difference value is greater than the preset color difference threshold, input the predicted color difference value and the current control parameters of the flexographic printing machine into the parameter offset prediction model to obtain the control parameter offset output by the parameter offset prediction model with the preset color difference value range as the target; and send the control parameter offset to the flexographic printing machine. The flexographic printing machine is configured to receive a control parameter offset sent by the terminal device, adjust the current control parameters based on the control parameter offset, and print a standard printing sample according to the adjusted control parameters.

15. The printing control system of the flexographic printing machine according to claim 13 or 14, wherein, The system also includes sensors. The sensor is communicatively connected to the terminal device. The sensor is configured to collect environmental parameters of the operating environment of the flexographic printing machine and send the environmental parameters to the terminal device.

16. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein, When the processor executes the computer program, it implements the printing control method of the flexographic printing press as described in any one of claims 1 to 11.

17. A computer program product, wherein, Includes a computer program, which, when run, causes the printing control method for a flexographic printing press as described in any one of claims 1 to 11 to be executed.

Images