An automated control method and system for a gravure color printing machine

By acquiring printing tension, calibration timestamps, and image phase sequences, exponentially weighted moving average and least squares prediction correction are performed. Combined with a fuzzy PID rule base to adjust motor parameters, the problem of low color registration accuracy in gravure printing is solved, and printing stability and color registration accuracy are improved.

CN123104223APending Publication Date: 2026-11-17GUANGDONG SHANZHANG MASCH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611319981.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-28
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

In gravure printing, the dynamic changes in printing speed and tension affect the registration accuracy, resulting in color difference and pattern distortion. Existing technologies cannot effectively coordinate speed adjustment, tension stability and registration correction, resulting in low registration accuracy.

Method used

By acquiring printing tension, recent calibration timestamp, marked image, and color group phase sequence, exponentially weighted moving average and least squares prediction correction are performed. Combined with a fuzzy PID rule base, motor parameters are dynamically adjusted to identify abnormal periodic fluctuations and perform misalignment analysis, generating optimized execution parameters to control the gravure printing machine.

Benefits of technology

It achieves adaptive and coordinated control of tension and speed, improves printing stability and color registration accuracy, and overcomes the shortcomings of existing technologies such as response lag and registration correction being independent of tension control.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123104223A_ABST
    Figure CN123104223A_ABST
Patent Text Reader

Abstract

This invention relates to the field of printing control technology and discloses an automated control method and system for a gravure color printing machine. The method includes acquiring a printing tension sequence, a recent calibration timestamp, a marker image sequence, and a color group phase sequence; performing an exponentially weighted moving average calculation on the printing tension sequence to obtain a stability evaluation sequence; adjusting motor parameters according to the stability evaluation sequence to obtain an adjustment speed sequence; performing abnormal periodicity detection on the adjustment speed sequence and performing a secondary calibration interference check on the recent calibration timestamp to obtain fluctuation correction requirements; performing misalignment analysis according to the fluctuation correction requirements to obtain a preliminary correction signal; and performing execution parameter matching according to the preliminary correction signal to obtain optimized execution parameters. This method can improve color registration accuracy.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of printing control technology, and in particular to an automated control method and system for a gravure color printing machine. Background Technology

[0002] Currently, in gravure printing, the dynamic changes in printing speed and tension directly affect registration accuracy, while tension imbalance interferes with ink supply uniformity, resulting in color differences or pattern distortion. Industrial control systems need to simultaneously coordinate speed adjustment, tension stabilization, and registration correction; however, complex nonlinear couplings exist between these variables, making it difficult to cope with sudden fluctuations by adjusting any single component alone. Therefore, there is an urgent need for an automated method capable of real-time analysis of tension trends, identification of abnormal speed cycles, and automatic matching of control parameters to achieve adaptive coordination between printing speed and tension.

[0003] In one existing technology, a tension sensor collects real-time tension values ​​during the printing process, which are then sent to a PID controller. The controller outputs a speed compensation value, which is superimposed on the main motor speed loop setpoint. Within each control cycle, the system calculates the tension deviation and its rate of change, adjusting the proportional gain and integral time according to a preset rule table. Registration detection uses an independent photoelectric sensor. When the color mark offset exceeds a fixed threshold, the servo motor adjusts the printing plate cylinder phase with a constant step size. When tension fluctuations exceed a set range, the system limits the speed setpoint to the previous stable value multiplied by a preset proportional coefficient, gradually restoring the speed after the tension stabilizes. In this technology, tension adjustment and registration correction are executed independently. Existing technologies rely on tension threshold comparison and fixed-step speed reduction, and registration correction is independent of the tension control loop, resulting in an inability to adjust the color phase in conjunction with speed fluctuations.

[0004] Therefore, the color matching accuracy of existing technologies is low. Summary of the Invention

[0005] This invention provides an automated control method and system for a gravure color printing machine to improve color registration accuracy.

[0006] In a first aspect, to solve the aforementioned technical problems, the present invention provides an automated control method for a gravure printing machine, comprising: acquiring a printing tension sequence, a recent calibration timestamp, a marked image sequence, and a color group phase sequence; performing an exponentially weighted moving average calculation on the printing tension sequence to obtain a stability evaluation sequence; performing correction on the stability evaluation sequence by analyzing the tension change trend to obtain an evaluation prediction sequence; adjusting motor parameters according to the evaluation prediction sequence to obtain an adjustment speed sequence; performing abnormal periodicity detection by analyzing the speed fluctuation characteristics of the adjustment speed sequence, and performing a secondary calibration interference check on the recent calibration timestamp according to the detection results and a preset calibration interference duration to obtain a fluctuation correction requirement; performing misalignment analysis by calculating the image offset of the marked image sequence and the phase offset of the color group phase sequence according to the fluctuation correction requirement to obtain the misalignment degree, and performing adjustment amount calculation according to the misalignment degree to obtain a preliminary correction signal; calculating the final adjustment cycle and the final adjustment step size according to the preliminary correction signal, and querying a preset execution parameter matching library according to the final adjustment cycle and the final adjustment step size to obtain optimized execution parameters to control the operation of the gravure printing machine.

[0007] Secondly, the present invention provides an automated control system for a gravure color printing machine, comprising: a data acquisition module for acquiring a printing tension sequence, a recent calibration timestamp, a marker image sequence, and a color group phase sequence; a stability evaluation module for performing an exponentially weighted moving average calculation on the printing tension sequence to obtain a stability evaluation sequence; a parameter adjustment module for performing correction on the stability evaluation sequence by analyzing the tension change trend to obtain an evaluation prediction sequence, and adjusting motor parameters according to the evaluation prediction sequence to obtain an adjustment speed sequence; and an anomaly detection module for performing anomaly periodicity detection by analyzing the speed fluctuation characteristics of the adjustment speed sequence, and based on the detection... The test results and the preset calibration interference duration are used to perform a second calibration interference check on the most recent calibration timestamp to obtain the fluctuation correction requirement; the misalignment analysis module is used to perform misalignment analysis by calculating the image offset of the marked image sequence and the phase offset of the color group phase sequence according to the fluctuation correction requirement, to obtain the degree of misalignment, and to perform adjustment calculation according to the degree of misalignment to obtain a preliminary correction signal; the parameter matching module is used to calculate the final adjustment period and the final adjustment step size according to the preliminary correction signal, and to query the preset execution parameter matching library according to the final adjustment period and the final adjustment step size to obtain optimized execution parameters to control the operation of the gravure printing machine.

[0008] Compared with the prior art, the present invention has the following beneficial effects: (1) This invention obtains a stability evaluation value by performing an exponentially weighted moving average on the printing tension sequence, uses least squares prediction correction to compensate for the delay, and dynamically adjusts the motor parameters in conjunction with a fuzzy PID rule base. Since the exponentially weighted moving average sensitively captures the tension trend, the least squares correction eliminates the transmission delay, and the fuzzy PID adaptively adjusts according to the degree of deviation, it solves the response lag caused by fixed thresholds and fixed step sizes, realizes adaptive coordinated control of tension and speed, and improves the operational stability; (2) This invention performs fast Fourier transform spectrum analysis and local extreme point detection on the adjustment speed sequence to identify abnormal periodic fluctuations, and performs secondary verification by combining the most recent calibration timestamp with the preset calibration interference duration. Since spectrum analysis quantifies the frequency characteristics of fluctuations, extreme point detection extracts the fluctuation range and amplitude, and calibration interference duration comparison eliminates the calibration interference of the sensor itself, it solves the problem of not being able to distinguish between real fluctuations and measurement interference, and improves the accuracy of identifying fluctuation correction requirements; (3) This invention calculates the degree of misalignment by measuring the offset of the marked image and the phase offset of the color group, generates a preliminary correction signal, determines the adjustment cycle based on the comprehensive stability index and the real-time equipment load rate, obtains optimized parameters by matching the execution parameter library, and corrects the color superposition deviation through closed-loop iteration. Since the execution parameter matching library adaptively outputs optimized control parameters based on the system stability state, equipment load conditions, and real-time response delay, and the superposition deviation verification ensures that the correction effect can be quantitatively evaluated, it overcomes the shortcomings of registration correction being independent of tension control and lacking closed-loop verification, thus improving the color registration accuracy. Attached Figure Description

[0009] Figure 1 This is a schematic diagram of the automated control method for a gravure color printing machine provided in the first embodiment of the present invention; Figure 2 This is a schematic diagram of the automated control system structure of the gravure color printing machine provided in the second embodiment of the present invention. Detailed Implementation

[0010] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0011] Reference Figure 1 The first embodiment of the present invention provides an automated control method for a gravure color printing machine, comprising the following steps: S11, acquire the printing tension sequence, the most recent calibration timestamp, the marked image sequence, and the color group phase sequence; S12, perform an exponentially weighted moving average calculation on the printing tension sequence to obtain a stability evaluation sequence; S13, the stability assessment sequence is corrected by analyzing the tension change trend to obtain the assessment prediction sequence, and the motor parameters are adjusted according to the assessment prediction sequence to obtain the adjustment speed sequence; S14, abnormal periodicity is detected by analyzing the speed fluctuation characteristics of the adjusted speed sequence, and a second calibration interference check is performed on the most recent calibration timestamp based on the detection results and the preset calibration interference duration to obtain the fluctuation correction requirement; S15, based on the fluctuation correction requirement, misalignment analysis is performed by calculating the image offset of the marked image sequence and the phase offset of the color group phase sequence to obtain the degree of misalignment, and adjustment amount calculation is performed based on the degree of misalignment to obtain a preliminary correction signal; S16, calculate the final adjustment cycle and final adjustment step size based on the preliminary correction signal, and query the preset execution parameter matching library based on the final adjustment cycle and final adjustment step size to obtain optimized execution parameters to control the operation of the gravure printing machine.

[0012] In step S11, the printing tension sequence, the most recent calibration timestamp, the marker image sequence, and the color group phase sequence are obtained.

[0013] Specifically, printing tension values ​​are acquired from the tension sensor of the printing equipment. The sampling frequency of the printing tension sequence is predetermined as follows: The tension signal is acquired during a complete production cycle of the printing equipment under normal production conditions. A Fast Fourier Transform (FFT) is performed on this signal to obtain its spectral distribution. The frequency corresponding to the point where the amplitude attenuates to one percent of the maximum amplitude is taken as the highest effective frequency of the signal, and the sampling frequency is set to twice this highest effective frequency. If the calculated sampling frequency is lower than the minimum sampling frequency supported by the equipment sensor hardware, the minimum sampling frequency of the sensor is taken as the preset value. If the calculated sampling frequency is higher than the highest sampling frequency that the equipment controller can handle, the highest sampling frequency of the controller is taken as the preset value. The tension value at each sampling moment is acquired in real time according to this preset sampling frequency and arranged in chronological order to form a printing tension sequence. Each tension value is associated with its sampling timestamp.

[0014] The system time node at the completion of the most recent automatic calibration operation is read from the encoder register of the printing equipment as the most recent calibration timestamp. This timestamp is an absolute time value used to subsequently determine whether speed fluctuations are caused by calibration interference. Registration mark patterns are acquired using a high-speed industrial camera. The camera's acquisition frame rate is determined independently based on the motion spectrum characteristics of the mark images, in the same manner as the tension sampling frequency. Images are continuously captured at the preset frame rate, with each frame representing a two-dimensional pixel matrix. The acquisition time of each frame is associated with the pixel matrix and arranged chronologically to form a mark image sequence. Phase angle values ​​are read from the encoders installed on each color group printing cylinder according to a preset sampling period. This sampling period is determined in the same way as the tension sampling frequency, based on the highest effective frequency of the phase signal. The phase angles at each sampling moment are arranged in color group number order to form a color group phase sequence.

[0015] All the sequences and parameters mentioned above were timestamped with the system clock during acquisition. Since the sampling frequency or frame rate of different data may vary, nearest neighbor matching was used to align each tension value, each image frame, and each phase value with the load rate and response delay closest to the timestamp, forming a dataset with a unified time reference. This data acquisition step provides real-time, time-aligned input for subsequent tension stability assessment, speed fluctuation detection, and color misregistration correction.

[0016] In step S12, the printing tension sequence is subjected to an exponentially weighted moving average calculation to obtain a stability evaluation sequence, including: filtering noise using a Butterworth low-pass filtering algorithm to obtain a filtered tension sequence; dividing the filtered tension sequence into segments according to a preset sliding time window and calculating the variance of the printing tension within each sliding time window to obtain a window variance sequence; and performing a weighted calculation using an exponentially weighted moving average method based on the window variance sequence and a preset attenuation factor to obtain a stability evaluation sequence.

[0017] Specifically, the process of calculating the stability evaluation sequence by performing an exponentially weighted moving average on the printing tension sequence is as follows: The printing tension sequence obtained in step S11 is acquired. This sequence consists of multiple tension values ​​arranged in chronological order. To remove high-frequency noise from the tension signal, a Butterworth low-pass filter algorithm is used to filter the printing tension sequence. The Butterworth low-pass filter is preset to fourth order. The cutoff frequency is determined by acquiring a historical tension sequence from one hour of continuous operation of the printing equipment under fault-free normal conditions, estimating the power spectral density of this sequence, and taking the frequency corresponding to the power spectral density dropping to 50% of the peak value as the cutoff frequency. Each tension value in the printing tension sequence is sequentially input into the filter. The filter calculates based on the weighted sum of the current input value and the previous four input values, as well as the feedback from the previous four output values, and outputs the corresponding filtered tension value. All filtered tension values ​​are arranged in their original chronological order to form a filtered tension sequence.

[0018] The filtered tension sequence is split according to a preset sliding time window. The length of the sliding time window is determined as follows: A continuously running filtered tension sequence of the printing equipment under normal production conditions is collected, and the sequence is split using multiple candidate window lengths, for example, from 0.5 seconds to 5 seconds, with a step of 0.5 seconds. The variance of the tension values ​​within all windows under each candidate window is calculated, and the median of all window variances is taken as the representative variance value of the candidate window. The minimum window length that makes the rate of change of the representative variance value less than 5% for the first time is selected as the preset sliding time window length. The rate of change is obtained by dividing the difference of the representative variance values ​​corresponding to adjacent candidate window lengths by the representative variance value corresponding to the previous candidate window length. The sliding step size is set to half the window length. Starting from the first sampling point of the filtered tension sequence, fixed-length window segments are sequentially extracted according to the sliding step size, and each window segment contains multiple filtered tension values. For each window segment, calculate the arithmetic mean of all filter tension values ​​within that window. Then, calculate the square of the deviation of each tension value from the mean. Sum the squares of all deviations and divide by the number of tension values ​​within the window to obtain the tension variance for that window. Arrange the variance values ​​of all windows in chronological order to form a window variance sequence.

[0019] Based on the window variance sequence and a preset attenuation factor, a stability assessment sequence is obtained through weighted calculation using an exponentially weighted moving average method. The attenuation factor is determined by collecting the window variance sequence of the printing equipment under normal, fault-free operation for multiple production cycles. The absolute value of the difference between two adjacent points in this sequence is taken as the fluctuation amplitude. The distribution of all fluctuation amplitudes is statistically analyzed, and the 95th percentile of the fluctuation amplitude is taken as the reference fluctuation value. The attenuation factor is set such that the output response time of the exponentially weighted moving average (i.e., the number of time windows required for the weights to decay to 37% of the initial value) equals the number of windows corresponding to the reference fluctuation value. Specifically, the first variance value in the window variance sequence is directly used as the first stability assessment value. For each subsequent variance value, the previous stability assessment value is multiplied by the attenuation factor, and this is added to the current window variance value multiplied by one minus the attenuation factor. The sum of these products yields the current stability assessment value. All stability assessment values ​​are arranged in chronological order to form a stability assessment sequence. Each value in this sequence reflects the weighted trend of tension fluctuation up to the current moment; a larger value indicates worse tension stability, which is used for subsequent imbalance judgment and motor parameter adjustment. This step transforms the original tension value into a stability quantification index with trend prediction capabilities, providing a basis for judgment in adaptive control.

[0020] In step S13, the stability assessment sequence is corrected by analyzing the tension change trend to obtain an assessment prediction sequence. The motor parameters are then adjusted according to the assessment prediction sequence to obtain an adjustment speed sequence. This includes: when the stability assessment values ​​in the stability assessment sequence exceed a preset number of consecutive limits, the least squares method is used to perform assessment prediction based on a pre-stored historical assessment sequence to obtain an assessment prediction sequence; when there is an assessment prediction value in the assessment prediction sequence that exceeds the stability assessment range, the drive motor parameters are adjusted according to the assessment prediction sequence using a preset fuzzy PID rule base to obtain motor adjustment parameters; based on the current tension disturbance value and the preset target speed, the printing speed distribution is calculated using a pre-built feedforward feedback model to obtain an adjustment speed sequence, and the control parameters of the drive motor are updated according to the motor adjustment parameters; when there is no assessment prediction value exceeding the stability assessment range in the assessment prediction sequence, or when there is no consecutive preset number of stability assessment values ​​exceeding the preset stability assessment range in the stability assessment sequence, the drive motor parameters are maintained and the detection process continues.

[0021] The construction process of the feedforward-feedback model includes: acquiring historical tension disturbance sequences, historical target velocity sequences, and historical actual velocity sequences; selecting the order using the Akaike Information Criterion to obtain the model order; estimating the coefficients using the least squares method based on the historical tension disturbance sequences, the historical target velocity sequences, the historical actual velocity sequences, and the model order to obtain a feedforward coefficient set and a feedback coefficient set; constructing a feedforward transfer function based on the feedforward coefficient set and the model order, and constructing a feedback transfer function based on the feedback coefficient set and the model order; and connecting the feedforward transfer function and the feedback transfer function in parallel to obtain the feedforward-feedback model.

[0022] Specifically, firstly, the stability assessment sequence generated in step S12 is obtained. This sequence consists of stability assessment values ​​arranged in chronological order. The system presets a stability assessment interval, the upper and lower limits of which are determined through historical data statistics. Specifically, it collects historical stability assessment sequences of the printing equipment running continuously for multiple production cycles under fault-free normal operating conditions, taking the 2.5th percentile of all historical stability assessment values ​​as the lower limit of the interval and the 97.5th percentile as the upper limit. Simultaneously, the system presets a number of consecutive exceedances. This value is determined by analyzing the distribution of event lengths that consecutively exceed the interval in the historical stability assessment sequence. The median of the lengths of all consecutive exceedance events is taken as the preset value; if the calculated median of the consecutive exceedance event lengths is not an integer, it is rounded up. When the stability assessment values ​​of the preset number of consecutive exceedances in the stability assessment sequence all exceed the preset stability assessment interval, the system initiates a correction process. At this point, a pre-stored historical evaluation sequence is read from memory. This sequence records stability evaluation values ​​from the most recent periods. If the total number of sampling periods is less than a set value, all available data is stored. When the total number of sampling periods reaches the set value, for each new stability evaluation value added, the oldest evaluation value in the sequence is deleted, maintaining a fixed sequence length. A least squares method is used to linearly fit the historical evaluation sequence. The average value of each sampling moment and the average value of the corresponding stability evaluation value are calculated. The slope of the fitted line is obtained by summing the products of the differences between each sampling moment and the average value and dividing by the sum of the squares of the differences between the sampling moment and the average value. The intercept of the fitted line is obtained by subtracting the slope multiplied by the average value of the sampling moments from the average value of the stability evaluation values. Based on this fitted line, the current moment is substituted as the independent variable to calculate the predicted evaluation value for the current moment. The same operation is performed on each sampling moment in the historical evaluation sequence to obtain a set of predicted evaluation values. These predicted values ​​are arranged in chronological order to form an evaluation prediction sequence. This prediction sequence eliminates the delay caused by data transmission and processing, reflecting the current actual tension state.

[0023] After obtaining the evaluation prediction sequence, it is compared with the preset stability evaluation interval. When any predicted evaluation value in the evaluation prediction sequence exceeds the interval, the parameters of the drive motor are adjusted according to the evaluation prediction sequence. The adjustment is based on a preset fuzzy PID rule base. The construction process of this rule base is as follows: historical operating data of the printing equipment under different tension deviation levels and deviation change rates are collected. The deviation level is divided into five levels: extremely small, small, medium, large, and extremely large. The deviation change rate is divided into five levels: negative large, negative small, zero, positive small, and positive large. For each combination of deviation level and change rate level, historical operating data of the printing equipment under different combinations of deviation level and change rate are collected. The optimal proportional gain, integral time, and derivative time corresponding to each set of data are recorded. This is the parameter combination that minimizes the time required for the tension to recover to the stable interval and minimizes the overshoot. These parameter combinations are filled into a lookup table of twenty-five rules. In use, the difference between the maximum value exceeding the interval in the evaluation prediction sequence and the upper limit of the interval is calculated as the deviation level. The difference between the two most recent evaluation prediction values ​​is divided by the sampling interval as the deviation change rate. The degree of deviation and the rate of change of deviation are mapped to their respective levels. Based on the level combination, the corresponding proportional gain adjustment, integral time adjustment, and derivative time adjustment are searched from the rule base. These adjustment values ​​are then superimposed on the current motor parameters to obtain the motor adjustment parameters, which include three values: proportional gain, integral time, and derivative time.

[0024] Based on the current tension disturbance value and the preset target speed, the printing speed distribution is calculated using a pre-constructed feedforward feedback model to obtain the adjusted speed sequence. Specifically, in each control cycle, the real-time collected printing tension value is subtracted from the preset target tension value to obtain the current tension disturbance value. The feedforward feedback model is constructed as follows: Data from at least one complete production cycle of the printing equipment under normal operating conditions is collected. This cycle includes various typical operating conditions such as no-load, low speed, medium speed, high speed, and start-stop. Multiple samples are extracted from this data at fixed time intervals, for example, one sample every ten seconds. Each sample contains a continuous and time-aligned tension disturbance sequence, target speed sequence, and actual printing speed sequence, respectively denoted as the historical tension disturbance sequence, historical target sequence, and historical speed adjustment sequence. When extracting historical data samples, the sequence length within each sample is equal, specifically set to the model order plus five to ensure sufficient sample size. If the original data length is insufficient, a sliding window method is used to sequentially extract multiple samples of equal length from the aforementioned continuous operating data, allowing for 50% overlap between adjacent samples. For each sample, the historical tension disturbance sequence, historical target sequence, and historical speed regulation sequence are normalized to zero mean. The arithmetic mean and standard deviation of each sequence are calculated. The standardized sequence is obtained by subtracting the mean from each value in the sequence and then dividing by the standard deviation.

[0025] Each sequence has the same length and contains data from multiple consecutive sampling periods. The model order is selected using the Akaike Information Criterion. Specifically, for each candidate order from one to ten, the model coefficients are estimated using the least squares method, the sum of squared residuals is calculated, and then the Akaike Information Criterion value is calculated as the sequence length multiplied by the logarithm of the sum of squared residuals plus twice the order. The order with the smallest criterion value is selected as the model order. After determining the order, the historical tension perturbation sequence and the historical target sequence are used as the input matrix, and the historical speed regulation sequence is used as the output vector. The coefficient matrix that minimizes the sum of squared fitting errors is solved using the least squares method. This coefficient matrix contains the coefficient sets of the feedforward channel and the feedback channel. The structure of the feedforward transfer function is a polynomial fraction where the numerator polynomial coefficients are determined by the feedforward coefficient set and the denominator polynomial is determined by the model order. The feedback transfer function has the same structure, and its coefficients are determined by the feedback coefficient set. The outputs of the feedforward transfer function and the feedback transfer function are added together to obtain a parallel combined feedforward-feedback model. When using this model, the current tension disturbance value and the target speed are simultaneously input into the model. The model calculates the feedforward output based on the feedforward channel and the feedback output based on the feedback channel. The two are added together to obtain the expected printing speed value at each sampling moment in the next control cycle, and arranged in chronological order to form an adjustment speed sequence. At the same time, the PID control parameters of the drive motor are updated according to the motor adjustment parameters.

[0026] When there are no predicted values ​​exceeding the stability assessment range in the evaluation prediction sequence, or when there are no consecutive events exceeding the preset number of limits in the stability assessment sequence, it indicates that the current tension state is stable, and there is no need to adjust the motor parameters. The current motor parameters are maintained, and the process returns to step S12 to continue monitoring. Through the above steps, a complete closed loop is achieved from tension stability assessment to adaptive adjustment of motor parameters and then to speed distribution planning, providing an adjusted speed benchmark for subsequent speed fluctuation detection and color correction.

[0027] In step S14, abnormal periodicity detection is performed by analyzing the velocity fluctuation characteristics of the adjusted velocity sequence, and a second calibration interference check is performed on the most recent calibration timestamp based on the detection results and a preset calibration interference duration to obtain the fluctuation correction requirement. This includes: performing spectral analysis on the adjusted velocity sequence using a fast Fourier transform algorithm to obtain spectral features, and calculating the standard deviation of the adjusted velocity sequence to obtain velocity fluctuation characteristics; when the amplitude in the spectral features exceeds a preset spectral peak threshold and the velocity fluctuation characteristics exceed a preset velocity fluctuation threshold, continuous local extreme points and the corresponding times in the adjusted velocity sequence are extracted to obtain the fluctuation time interval and fluctuation amplitude; reading The calibration type information associated with the most recent calibration timestamp is used to query a preset calibration interference feature library to determine whether the spectral feature contains the feature frequency corresponding to the calibration type. When the spectral feature contains the feature frequency corresponding to the calibration type, the difference between the start time of the fluctuation time interval and the most recent calibration timestamp is calculated to obtain the fluctuation calibration interval. When the fluctuation calibration interval is less than or equal to a preset calibration interference duration, it is determined to be calibration data interference, and the fluctuation correction requirement is not generated. When the fluctuation calibration interval is greater than the preset calibration interference duration, or the spectral feature does not contain the feature frequency corresponding to the calibration type, a fluctuation correction requirement is generated based on the fluctuation time interval and the fluctuation amplitude.

[0028] Specifically, the adjusted speed sequence obtained in step S13 is acquired. This sequence consists of multiple expected printing speed values ​​arranged in chronological order. A Fast Fourier Transform (FFT) is performed on this sequence. Using the adjusted speed sequence as input, the sequence length is padded with zeros to the nearest power of two using the radix-2 time decimation method. Then, a butterfly operation is performed step-by-step to obtain the complex amplitude of each frequency component. The modulus of each complex amplitude is calculated (square root of the sum of the squared real and imaginary parts) to obtain the spectral characteristics. Simultaneously, the standard deviation of the adjusted speed sequence is calculated. First, the arithmetic mean of all speed values ​​in the sequence is calculated. Then, the square of the deviation of each speed value from the mean is calculated. The sum of all squared deviations is divided by the sequence length minus one, and the square root is taken to obtain the speed fluctuation characteristics.

[0029] A preset peak threshold is established. This threshold is determined by collecting historical adjustment speed sequences from multiple production cycles of the printing equipment under fault-free normal operation. A Fast Fourier Transform is performed on each historical sequence to obtain the spectrum. The maximum amplitude, excluding the DC component, is extracted from each spectrum, and the 95th percentile of all historical maximum amplitudes is taken as the peak threshold. A preset speed fluctuation threshold is also established. This threshold is determined by collecting the standard deviations of each sequence under the same historical data and taking the 95th percentile of all standard deviations as the speed fluctuation threshold. If the amplitude of any frequency component in the spectral characteristics exceeds the peak threshold, and the speed fluctuation characteristic exceeds the speed fluctuation threshold, an abnormal periodic fluctuation is identified, and the process proceeds to the local extremum extraction step. Otherwise, no fluctuation correction requirement is generated, and the process returns to step S12 to continue detection.

[0030] After determining the existence of abnormal periodic fluctuations, continuous local extrema are extracted from the adjusted velocity sequence. Each sampling point in the adjusted velocity sequence is traversed (excluding the first and last points). If the velocity value of the current point is greater than both the previous and next points, the point is marked as a local maximum; if the velocity value of the current point is less than both the previous and next points, the point is marked as a local minimum. The velocity value and its corresponding timestamp are recorded for each local extreme point. Starting from the first local extreme point, segments where local maxima and local minima alternate are found sequentially. The timestamp of the first extreme point in this segment is taken as the start time of the fluctuation, and the timestamp of the last extreme point is taken as the end time of the fluctuation; these two times constitute the fluctuation time interval. Within this segment, the amplitude difference between each adjacent local maximum and local minimum is calculated, and the maximum value among all amplitude differences is taken as the fluctuation amplitude.

[0031] Read the calibration type information associated with the most recent calibration timestamp obtained in step S11. This calibration type information includes zero-point calibration or linearity calibration. Query the preset calibration interference feature library to determine whether the spectral features contain the characteristic frequency corresponding to the calibration type. The construction process of the calibration interference feature library is as follows: During the equipment debugging phase, zero-point calibration and linearity calibration are performed on each color group encoder, and the speed sequence generated during calibration is recorded. Spectral analysis is performed on this sequence to extract the frequency components unique to the calibration signal. The mapping relationship between the calibration type and the characteristic frequency is then stored in the calibration interference feature library.

[0032] The process of constructing the calibration interference feature library is as follows: during the equipment debugging phase, each color group encoder is triggered to perform zero-point calibration and linearity calibration actions; the speed sequence generated during the calibration action is collected, the collected speed sequence is subjected to spectrum analysis, and the frequency components unique to the calibration signal are extracted; the mapping relationship between calibration type and characteristic frequency is stored in the calibration interference feature library.

[0033] If the spectral characteristics contain the characteristic frequency corresponding to the calibration type, the difference between the start time of the fluctuation time interval and the most recent calibration timestamp is calculated to obtain the fluctuation calibration interval. A calibration interference duration is preset. This duration is determined by collecting the duration data of the printing equipment during automatic calibration operations and taking the maximum value of the duration of all calibration operations as the calibration interference duration.

[0034] If the fluctuation calibration interval is less than or equal to the calibration interference duration, the current fluctuation is determined to be a spurious signal interruption or jump caused by encoder calibration, i.e., calibration data interference. No fluctuation correction requirement is generated, and the process returns to step S12 to continue detection. If the fluctuation calibration interval is greater than the calibration interference duration, it indicates that the calibration process has already ended, and the current fluctuation is entirely caused by physical factors, thus generating a fluctuation correction requirement. If the spectral characteristics do not contain the characteristic frequency corresponding to this calibration type, it indicates that the current fluctuation is not caused by abnormal calibration data, but is a real physical fluctuation, and a fluctuation correction requirement is directly generated.

[0035] In step S15, based on the fluctuation correction requirement, misalignment analysis is performed by calculating the image offset of the marked image sequence and the phase offset of the color group phase sequence to obtain the degree of misalignment. An adjustment amount calculation is then performed based on the degree of misalignment to obtain a preliminary correction signal. This includes: according to the fluctuation correction requirement, the marked image sequence is converted to grayscale to obtain a grayscale image sequence; based on the grayscale image sequence, edge features are extracted using the Canny edge detection algorithm to obtain an edge feature map; the pixel position deviation between the edge feature map and a preset standard edge template is calculated to obtain the pixel width deviation, and the pixel width is adjusted accordingly. The image offset is obtained by multiplying the degree deviation by the preset camera calibration parameters; the difference between the phases of adjacent color groups in the color group phase sequence is calculated to obtain the color group phase difference, and the color group phase difference is multiplied by the preset roller circumference and divided by twice pi to obtain the phase offset; the maximum value of the image offset and the phase offset is taken as the degree of misalignment. When the degree of misalignment is not greater than the preset degree of misalignment threshold, no fluctuation correction is required and the detection continues; when the degree of misalignment is greater than the degree of misalignment threshold, the phase adjustment amount and control coefficient are calculated according to the degree of misalignment through the pre-constructed PID control rules to obtain the preliminary correction signal.

[0036] Specifically, based on the degree of misalignment, a preliminary correction signal is obtained by calculating the phase adjustment amount and control coefficient using pre-constructed PID control rules. This includes: calculating the adjustment amount using a PID control algorithm based on the degree of misalignment to obtain a proportional adjustment amount, an integral adjustment amount, and a derivative adjustment amount; adding the proportional adjustment amount, the integral adjustment amount, and the derivative adjustment amount to obtain a phase adjustment amount; and encoding the phase adjustment amount into an instruction to obtain a preliminary correction signal.

[0037] Specifically, the fluctuation correction requirement is obtained from step S14, which includes the fluctuation frequency and time interval to be monitored. The labeled image sequence and color group phase sequence are obtained from step S11. Each frame of the labeled image sequence is converted to grayscale. The red, green, and blue channel values ​​of each pixel in each frame are extracted. The red channel value is multiplied by a preset red weighting coefficient (e.g., 0.299), the green channel value by a preset green weighting coefficient (e.g., 0.587), and the blue channel value by a preset blue weighting coefficient (e.g., 0.114). The three products are added together to obtain the grayscale value of the pixel. The preset weighting coefficients are determined as follows: Multiple color images of a standard grayscale card are acquired under normal lighting conditions on a printing device. Each image contains multiple color blocks with known actual grayscale values. For each color block, its actual grayscale value and the average values ​​of the red, green, and blue channels in the image are recorded. The three-channel average values ​​of each color block are used as independent variables, and the actual grayscale value is used as the dependent variable. Linear regression fitting is performed using the least squares method to find the three weighting coefficients that minimize the sum of squared fitting errors. After fitting, the three weighting coefficients are normalized so that their sum equals one, resulting in the final preset weighting coefficients.

[0038] Edge feature extraction is performed on grayscale image sequences using the Canny edge detection algorithm. First, each frame of the grayscale image is Gaussian smoothed using a 5x5 Gaussian kernel. The weight of each position within the kernel is calculated using a Gaussian function based on the distance from that position to the kernel center. The Gaussian kernel is then convolved with the grayscale image to obtain a smoothed image. Next, the gradient magnitude and direction of each pixel in the smoothed image are calculated. The Sobel operator is used to convolve the smoothed image horizontally and vertically to obtain the horizontal and vertical gradients. The gradient magnitude is equal to the square root of the sum of the squares of the horizontal and vertical gradients, and the gradient direction is equal to the arctangent of the vertical gradient divided by the horizontal gradient. Finally, non-maximum suppression is applied to the gradient magnitude. For each pixel, the gradient magnitudes of the two preceding and following pixels are compared along their gradient direction. If the current pixel's magnitude is not the maximum value, it is set to zero. Finally, a dual-threshold edge connection is performed, with a preset high threshold and a low threshold. The high threshold is determined by acquiring multiple properly registered labeled images and calculating the 95th percentile of the gradient magnitude for each image. The low threshold is half of the high threshold. Pixels with gradient magnitudes greater than the high threshold are marked as strong edges, while pixels with gradient magnitudes greater than the low threshold but less than the high threshold are marked as weak edges. Weak edges connected to strong edges are also marked as edges, resulting in the edge feature map for each frame. The edge feature maps of all frames are arranged in chronological order to form an edge feature map sequence.

[0039] A preset standard edge template is read from memory. This template is a baseline edge feature map generated during the initial device debugging process by capturing an image with registration marks in ideal positions and following the same edge detection procedure. The pixel position deviation is calculated between the current frame's edge feature map and the standard edge template. Preset reference point coordinates (e.g., coordinates of a marked cross intersection) are extracted from the standard edge template. The corresponding actual point coordinates are found in the edge feature map through template matching. The differences between the actual point coordinates and the reference point coordinates in the horizontal and vertical directions are calculated. The square root of the sum of the squares of these two differences is taken as the pixel width deviation. This pixel width deviation is multiplied by preset camera calibration parameters to obtain the image offset. The camera calibration parameters are determined by photographing a standard calibration board at a known actual distance, calculating the pixel distance between two marked points on the calibration board, and dividing the actual distance by the pixel distance to obtain the actual physical size of each pixel, which is used as the camera calibration parameter.

[0040] Simultaneously, the phase offset is calculated from the color group phase sequence. The color group phase sequence contains multiple color groups, such as the real-time phase angle values ​​of the first, second, third, and fourth color groups. The phase difference between adjacent color groups is calculated by subtracting the phase value of the first color group from the phase value of the second color group to obtain the first phase difference, subtracting the phase value of the second color group from the phase value of the third color group to obtain the second phase difference, and so on. Each phase difference is expressed in radians. The preset roller circumference is determined based on the mechanical parameters of the printing equipment, and the average value of the roller circumference for each color group is taken. Each phase difference is multiplied by the roller circumference and then divided by twice pi to obtain the actual offset corresponding to that adjacent color group, which is recorded as the phase offset. For multiple adjacent color groups, multiple phase offsets are obtained, and the maximum value is taken as the final phase offset, or the average value is taken as the representative value.

[0041] The maximum value between image offset and phase offset is taken as the degree of misalignment. A preset misalignment threshold is determined by acquiring multiple sets of marked image sequences and color group phase sequences of the printing equipment under qualified printing production conditions, calculating the misalignment value of each set of data, and taking the 95th percentile of all misalignment values ​​as the misalignment threshold. When the misalignment degree is not greater than the misalignment threshold, it indicates that the current color registration accuracy meets the requirements, no fluctuation correction is needed, the current fluctuation correction requirement is cleared, and the process returns to step S12 to continue detection.

[0042] When the misalignment exceeds a threshold, a phase adjustment is calculated based on the misalignment level to obtain a preliminary correction signal. First, the adjustment is calculated using a PID control algorithm based on the misalignment level, with the misalignment level as the deviation input. The initial values ​​of the proportional gain, integral gain, and derivative gain are determined as follows: Historical data is collected from the printing equipment during normal operation without faults and without color registration errors. The PID controller is placed in pure proportional mode, and the proportional gain is gradually increased until the system exhibits constant-amplitude oscillations. The critical proportional gain and oscillation period at this point are recorded. Based on the critical proportional gain and oscillation period, the initial values ​​of the proportional gain, integral gain, and derivative gain are calculated using the Ziegler-Nichols tuning rule. The proportional adjustment equals the proportional gain multiplied by the current deviation. The integral adjustment equals the integral gain multiplied by the sum of all deviations accumulated from the beginning to the current moment. The derivative adjustment equals the derivative gain multiplied by the difference between the current deviation and the previous deviation. The proportional, integral, and derivative adjustments are added together to obtain the phase adjustment.

[0043] The phase adjustment amount is encoded according to a preset instruction format to generate a preliminary correction signal. This step realizes the conversion from image and phase data to color correction instructions, providing an adjustment basis for closed-loop control.

[0044] In step S16, the final adjustment period and final adjustment step size are calculated based on the preliminary correction signal, and the optimized execution parameters are obtained by querying a preset execution parameter matching library based on the final adjustment period and final adjustment step size. This includes: obtaining the real-time device load rate; calculating the comprehensive stability index at the current moment based on the stability evaluation sequence, wherein the comprehensive stability index is the normalized deviation between the current stability evaluation value and the historical benchmark value; responding to the preliminary correction signal, querying a preset dynamic adjustment rule library based on the comprehensive stability index and the real-time device load rate to obtain the final adjustment period; mapping the amplification coefficient through a preset amplification mapping table based on the preliminary correction signal to obtain the step size amplification coefficient, and multiplying the step size amplification coefficient by a preset initial step size to obtain the final adjustment step size; and matching parameters through a preset execution parameter matching library based on the final adjustment period and final adjustment step size to obtain the optimized execution parameters.

[0045] Specifically, the current motor load rate is read from the main driver register of the printing equipment. This value is the ratio of the actual output torque to the rated torque of the motor, and is used as the real-time equipment load rate. Simultaneously, the latest stability assessment sequence is obtained from the output buffer of step S12. This sequence reflects the current tension fluctuation trend of the system. When calculating the comprehensive stability index, the mean μ and standard deviation σ of the historical stability assessment sequence are first calculated from historical normal operation data as a benchmark. The formula for calculating the comprehensive stability index at the current moment is: ;in This is the current assessment value in the latest stability assessment sequence. The current stability composite index is a dimensionless Z-score value, used to represent how many standard deviations it deviates from the normal level. The stability composite index is divided into three intervals, as shown in the table below. The threshold for each interval is determined by quantile statistics of the Z-score values ​​of historical stability assessment sequences. The 95th quantile of the historical Z-score sequence is denoted as Z. 95 The 99th percentile is Z. 99 :

[0046] In one implementation, the construction process of the dynamic adjustment rule base involves collecting historical operating data of the printing equipment under different combinations of comprehensive stability indicators and real-time equipment load rates, and recording the control cycle that enables the system to recover stability most quickly under each combination. For all combinations, multiple sets of records belonging to that combination are selected from the historical data, and the time required for the system to recover from fluctuation to stability in each set of records is calculated. The control cycle with the shortest required time is taken as the optimal control cycle for that combination. When constructing the dynamic adjustment rule base, a physical constraint is imposed on the optimal control cycle corresponding to each combination. The control cycle must not be less than three times the time required for the actuator to complete a complete adjustment action under the current load conditions. If the optimal cycle selected from the historical data violates this constraint, then the constraint value is taken as the optimal control cycle for that combination. For example, the dynamic adjustment rule base is shown in the table below:

[0047] In practice, based on the range of the currently calculated comprehensive stability index and the range of the real-time equipment load rate, the optimal control cycle is retrieved from the dynamic adjustment rule base and used as the final adjustment cycle. This cycle comprehensively reflects the current stability state of the system and the control frequency requirements of the equipment load conditions. Under light and medium load conditions, the control cycle is appropriately shortened to enhance response capability when system stability decreases; under heavy load conditions, due to the extended mechanical response time of the actuator, the control cycle needs to be increased accordingly to ensure that each adjustment is fully completed. Simultaneously, the final adjustment cycle has a lower limit protection, which must not be less than three times the minimum physical response time of the actuator to avoid control command stacking and actuator oscillation.

[0048] Based on the phase adjustment value in the preliminary correction signal, the amplification coefficient is mapped using a preset amplification mapping table. The construction process of the amplification mapping table is as follows: Historical operating data of the printing equipment under different phase adjustment values ​​is collected, and the optimal step size amplification coefficient corresponding to each adjustment value is recorded, thus obtaining the amplification multiple that allows for rapid convergence and no overshoot in color correction. The absolute value of the phase adjustment value is divided into several intervals, such as 0 to 0.01 radians, 0.01 to 0.05 radians, 0.05 to 0.1 radians, and above 0.1 radians. The median of the historical optimal amplification coefficient within each interval is taken as the mapping coefficient for that interval. During use, the corresponding step size amplification coefficient is read from the amplification mapping table based on the interval where the absolute value of the current phase adjustment value falls. Simultaneously, the preset initial step size is read from memory. Under no-load operation and without color correction deviation, a set of incremental phase adjustment commands is sent through the actuator. The actual displacement of the color correction mark after each adjustment is measured to find the minimum phase adjustment value that enables reliable movement of the actuator. This minimum phase adjustment value is multiplied by a safety factor to obtain the initial step size. Multiply the step size increase factor by the preset initial step size to obtain the final adjustment step size.

[0049] The system acquires real-time equipment load rate and real-time response delay. The motor load rate, calculated as the ratio of actual output torque to rated torque, is read from the main controller of the printing equipment. The real-time response delay is calculated from the time elapsed between the issuance of the current control command and the start of the actuator's action, measured by a high-precision timer. A pre-defined execution parameter matching library, a multi-dimensional lookup table, is used. The final adjustment period, final adjustment step size, real-time equipment load rate, and real-time response delay are used as input indices to search for the corresponding proportional gain and integral time constant in the execution parameter matching library. These two values ​​are then combined to form the optimized execution parameters.

[0050] The data for constructing the parameter matching library comes from historical operating logs and actual control experiment data of printing equipment under different operating conditions. Six months of operating records were exported from the equipment's historical database, and operating condition segments containing typical color registration deviations, tension fluctuations, and speed changes were selected. These raw data were preprocessed to remove abnormal data points caused by sensor malfunctions or human error. The system dynamic characteristics corresponding to each valid operating condition segment were calculated, including average equipment load rate, average response delay, maximum tension fluctuation amplitude, and the rate of change of color registration deviation. These processed system dynamic characteristics were used as input features, and the proportional gain and integral time constant obtained through manual debugging under the corresponding operating conditions, which minimized system stability and adjustment time, were used as output labels to form the basic dataset required for constructing the matching library.

[0051] The execution parameter matching library uses a four-dimensional array structure for storage, with index dimensions including final adjustment period, final adjustment step size, real-time device load rate, and real-time response latency. The final adjustment period is divided into five intervals: millisecond-level ultra-short period, ten-millisecond-level short period, hundred-millisecond-level medium period, second-level long period, and ten-second-level ultra-long period. The final adjustment step size is divided into three levels: fine-tuning step size, medium step size, and large step size. The real-time device load rate is divided into four intervals: idle interval, light load interval, medium load interval, and heavy load interval. The real-time response latency is divided into three levels: low latency, medium latency, and high latency. At the intersection of interval combinations in each dimension, a set of optimal proportional gain and integral time constants is stored. Parameter selection is determined using a grid search method combined with historical best data. Within the operating range corresponding to each intersection point, all matching historical records are selected from the basic dataset, and the geometric mean of the proportional gain and integral time constant in these records is calculated as the stored value for that intersection point. If a certain intersection point has no corresponding historical record, an interpolation method is used to fill it with the weighted average of the parameter values ​​of its eight adjacent intersection points. The weights are determined based on the reciprocal of the Euclidean distance between each adjacent point in the parameter space.

[0052] During the control process, the currently calculated final adjustment cycle, final adjustment step size, and real-time equipment load rate and real-time response delay read from the equipment sensors are acquired. Based on the values ​​of these four variables, their respective intervals or levels are determined. Using the index values ​​of these four intervals or levels as address pointers, the four-dimensional array of the execution parameter matching library is accessed. The proportional gain and integral time constant stored at this address are read, and these two values ​​are combined as optimized execution parameters, which are output to the controller for parameter adjustment of the actuator. If the current variable value happens to be on the boundary between two intervals, linear interpolation is used to read the parameter values ​​of the two adjacent intervals, and the weight is calculated based on the relative position of the current value within the interval to obtain the final optimized execution parameters. The optimized execution parameters include the proportional gain and integral time constant. The optimized execution parameters are loaded into the color-matching actuator driver of the printing press, and the driver adjusts the control loop parameters of the servo motor according to the proportional gain and integral time constant.

[0053] Reference Figure 2The second embodiment of the present invention provides an automated control system for a gravure color printing machine, comprising: a data acquisition module for acquiring a printing tension sequence, a recent calibration timestamp, a marker image sequence, and a color group phase sequence; a stability evaluation module for performing an exponentially weighted moving average calculation on the printing tension sequence to obtain a stability evaluation sequence; a parameter adjustment module for performing correction on the stability evaluation sequence by analyzing the tension change trend to obtain an evaluation prediction sequence, and adjusting motor parameters according to the evaluation prediction sequence to obtain an adjustment speed sequence; and an anomaly detection module for performing anomaly periodicity detection by analyzing the speed fluctuation characteristics of the adjustment speed sequence, and based on the detection... The test results and the preset calibration interference duration are used to perform a second calibration interference check on the most recent calibration timestamp to obtain the fluctuation correction requirement; the misalignment analysis module is used to perform misalignment analysis by calculating the image offset of the marked image sequence and the phase offset of the color group phase sequence according to the fluctuation correction requirement, to obtain the degree of misalignment, and to perform adjustment calculation according to the degree of misalignment to obtain a preliminary correction signal; the parameter matching module is used to calculate the final adjustment period and the final adjustment step size according to the preliminary correction signal, and to query the preset execution parameter matching library according to the final adjustment period and the final adjustment step size to obtain optimized execution parameters to control the operation of the gravure printing machine.

[0054] It should be noted that the gravure printing machine automation control system provided in this embodiment of the invention is used to execute all the process steps of the gravure printing machine automation control method of the above embodiment. The working principle and beneficial effect of the two are one-to-one, so they will not be described again.

[0055] It should be noted that the system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and 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 modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the system embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0056] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. An automated control method for a gravure color printing machine, characterized in that, include: Acquire the printing tension sequence, the most recent calibration timestamp, the marked image sequence, and the color group phase sequence; An exponentially weighted moving average is calculated on the printing tension sequence to obtain a stability assessment sequence. The stability assessment sequence is corrected by analyzing the tension change trend to obtain an assessment prediction sequence. The motor parameters are adjusted according to the assessment prediction sequence to obtain an adjustment speed sequence. Abnormal periodicity is detected by analyzing the speed fluctuation characteristics of the adjustment speed sequence. A second calibration interference check is performed on the most recent calibration timestamp based on the detection results and a preset calibration interference duration to obtain the fluctuation correction requirement. Based on the fluctuation correction requirements, misalignment analysis is performed by calculating the image offset of the marked image sequence and the phase offset of the color group phase sequence to obtain the degree of misalignment. Adjustment amount calculation is then performed based on the degree of misalignment to obtain a preliminary correction signal. The final adjustment cycle and final adjustment step size are calculated based on the preliminary correction signal. A preset execution parameter matching library is then queried based on the final adjustment cycle and the final adjustment step size to obtain optimized execution parameters, thereby controlling the operation of the gravure printing machine.

2. The automated control method for a gravure color printing machine according to claim 1, characterized in that, The step of performing an exponentially weighted moving average calculation on the printing tension sequence to obtain a stability evaluation sequence includes: filtering noise from the printing tension sequence using a Butterworth low-pass filtering algorithm to obtain a filtered tension sequence; dividing the filtered tension sequence into segments according to a preset sliding time window and calculating the variance of the printing tension within each sliding time window to obtain a window variance sequence; and performing a weighted calculation using an exponentially weighted moving average method based on the window variance sequence and a preset attenuation factor to obtain a stability evaluation sequence.

3. The automated control method for a gravure color printing machine according to claim 1, characterized in that, The process of correcting the stability assessment sequence by analyzing the tension change trend to obtain an assessment prediction sequence, and adjusting the motor parameters according to the assessment prediction sequence to obtain an adjustment speed sequence, includes: when the stability assessment values ​​in the stability assessment sequence exceed a preset number of consecutive preset out-of-limit values, the least squares method is used to perform assessment prediction based on a pre-stored historical assessment sequence to obtain an assessment prediction sequence; when there are assessment prediction values ​​in the assessment prediction sequence that exceed the stability assessment range, the parameters of the drive motor are adjusted according to the assessment prediction sequence using a preset fuzzy PID rule base to obtain motor adjustment parameters; the printing speed distribution is calculated using a pre-built feedforward feedback model based on the current tension disturbance value and the preset target speed to obtain an adjustment speed sequence, and the control parameters of the drive motor are updated according to the motor adjustment parameters; when there are no assessment prediction values ​​in the assessment prediction sequence that exceed the stability assessment range, or when there are no consecutive preset number of stability assessment values ​​that exceed the preset stability assessment range, the parameters of the drive motor are maintained and the detection continues.

4. The automated control method for a gravure color printing machine according to claim 3, characterized in that, The process of constructing the feedforward-feedback model includes: acquiring historical tension disturbance sequences, historical target velocity sequences, and historical actual velocity sequences; selecting the order using the Akaike Information Criterion to obtain the model order; estimating the coefficients using the least squares method based on the historical tension disturbance sequences, the historical target velocity sequences, the historical actual velocity sequences, and the model order to obtain a feedforward coefficient set and a feedback coefficient set; constructing a feedforward transfer function based on the feedforward coefficient set and the model order, and constructing a feedback transfer function based on the feedback coefficient set and the model order; and connecting the feedforward transfer function and the feedback transfer function in parallel to obtain the feedforward-feedback model.

5. The automated control method for a gravure color printing machine according to claim 1, characterized in that, The process involves analyzing the velocity fluctuation characteristics of the adjusted velocity sequence to detect abnormal periodicity, and performing a secondary calibration interference check on the most recent calibration timestamp based on the detection results and a preset calibration interference duration to obtain the fluctuation correction requirement. This includes: performing spectral analysis on the adjusted velocity sequence using a Fast Fourier Transform algorithm to obtain spectral features, and calculating the standard deviation of the adjusted velocity sequence to obtain velocity fluctuation characteristics; when the amplitude in the spectral features exceeds a preset peak threshold and the velocity fluctuation characteristics exceed a preset velocity fluctuation threshold, extracting consecutive local extreme points and their corresponding times from the adjusted velocity sequence to obtain the fluctuation time interval and fluctuation amplitude; and reading the most recent calibration timestamp. The calibration type information associated with the nearest calibration timestamp is queried from a preset calibration interference feature library to determine whether the spectral feature contains the characteristic frequency corresponding to the calibration type. When the spectral feature contains the characteristic frequency corresponding to the calibration type, the difference between the start time of the fluctuation time interval and the nearest calibration timestamp is calculated to obtain the fluctuation calibration interval. When the fluctuation calibration interval is less than or equal to a preset calibration interference duration, it is determined to be calibration data interference, and the fluctuation correction requirement is not generated. When the fluctuation calibration interval is greater than the preset calibration interference duration, or the spectral feature does not contain the characteristic frequency corresponding to the calibration type, a fluctuation correction requirement is generated based on the fluctuation time interval and the fluctuation amplitude.

6. The automated control method for a gravure color printing machine according to claim 1, characterized in that, The process, based on the fluctuation correction requirement, involves performing misalignment analysis by calculating the image offset of the marked image sequence and the phase offset of the color group phase sequence to obtain the degree of misalignment, and then calculating the adjustment amount based on the degree of misalignment to obtain a preliminary correction signal. This includes: converting the marked image sequence to grayscale based on the fluctuation correction requirement to obtain a grayscale image sequence; extracting edge features from the grayscale image sequence using the Canny edge detection algorithm to obtain an edge feature map; calculating the pixel position deviation between the edge feature map and a preset standard edge template to obtain the pixel width deviation; and then setting the pixel width deviation... Multiply by preset camera calibration parameters to obtain image offset; calculate the difference between the phases of adjacent color groups in the color group phase sequence to obtain color group phase difference, and multiply the color group phase difference by the preset roller circumference and divide by twice pi to obtain phase offset; take the maximum value of the image offset and the phase offset as the degree of misalignment. When the degree of misalignment is not greater than the preset degree of misalignment threshold, no fluctuation correction is required and the detection continues; when the degree of misalignment is greater than the degree of misalignment threshold, the phase adjustment amount and control coefficient are calculated according to the degree of misalignment through the pre-constructed PID control rules to obtain a preliminary correction signal.

7. The automated control method for a gravure color printing machine according to claim 6, characterized in that, The step of calculating the phase adjustment amount and control coefficient according to the degree of misalignment using a pre-constructed PID control rule to obtain a preliminary correction signal includes: calculating the adjustment amount according to the degree of misalignment using a PID control algorithm to obtain a proportional adjustment amount, an integral adjustment amount, and a derivative adjustment amount; adding the proportional adjustment amount, the integral adjustment amount, and the derivative adjustment amount to obtain a phase adjustment amount; and encoding the phase adjustment amount into an instruction to obtain a preliminary correction signal.

8. The automated control method for a gravure color printing machine according to claim 1, characterized in that, The process of calculating the final adjustment period and final adjustment step size based on the preliminary correction signal, and querying a preset execution parameter matching library based on the final adjustment period and final adjustment step size to obtain optimized execution parameters includes: obtaining the real-time device load rate; calculating the comprehensive stability index at the current moment based on the stability assessment sequence, wherein the comprehensive stability index is the normalized deviation between the current stability assessment value and the historical benchmark value; responding to the preliminary correction signal, querying a preset dynamic adjustment rule library based on the comprehensive stability index and the real-time device load rate to obtain the final adjustment period; mapping the amplification coefficient through a preset amplification mapping table based on the preliminary correction signal to obtain the step size amplification coefficient, and multiplying the step size amplification coefficient by a preset initial step size to obtain the final adjustment step size; and matching parameters through a preset execution parameter matching library based on the final adjustment period and final adjustment step size to obtain optimized execution parameters to control the operation of the gravure printing machine.

9. An automated control system for a gravure color printing machine, characterized in that, include: The data acquisition module is used to acquire the printing tension sequence, the most recent calibration timestamp, the marker image sequence, and the color group phase sequence; The stability assessment module is used to calculate the exponentially weighted moving average of the printing tension sequence to obtain a stability assessment sequence; the parameter adjustment module is used to correct the stability assessment sequence by analyzing the tension change trend to obtain an assessment prediction sequence, and adjust the motor parameters according to the assessment prediction sequence to obtain an adjustment speed sequence; the anomaly verification module is used to perform anomaly periodic detection by analyzing the speed fluctuation characteristics of the adjustment speed sequence, and perform a secondary calibration interference verification on the most recent calibration timestamp according to the detection results and the preset calibration interference duration to obtain the fluctuation correction requirements; The misalignment analysis module is used to perform misalignment analysis by calculating the image offset of the marked image sequence and the phase offset of the color group phase sequence according to the fluctuation correction requirements, to obtain the degree of misalignment, and to perform adjustment calculation according to the degree of misalignment to obtain a preliminary correction signal. The parameter matching module is used to calculate the final adjustment cycle and the final adjustment step size based on the preliminary correction signal, and to query the preset execution parameter matching library based on the final adjustment cycle and the final adjustment step size to obtain optimized execution parameters to control the operation of the gravure printing machine.