A method for automatically detecting the roundness of a plate roller

Abstract

The present application relates to the field of plate roller roundness detection, and discloses a kind of plate roller roundness automatic detection method, comprising: driving gravure printing roller uniform speed rotation, synchronous acquisition driving motor's instantaneous torque time sequence signal or instantaneous current time sequence signal as source end time sequence signal, the radial displacement time sequence signal of gravure printing roller surface as effect end time sequence signal and the real-time angular position time sequence signal of gravure printing roller rotation;According to real-time angular position time sequence signal, source end time sequence signal and effect end time sequence signal are respectively mapped to circumferential angle domain;The reference fluctuation interval of source end angle domain signal and the reference displacement interval of effect end angle domain signal are obtained, source end abnormal angle span and effect end abnormal angle span are obtained;Source end abnormal angle span and effect end abnormal angle span are compared in overlap on circumferential angle coordinate, whether there is true roundness defect according to comparison result is judged and defect positioning interval is recorded, and the maximum deviation amplitude of effect end angle domain signal in defect positioning interval is output.

Description

Technical Field

[0001] This invention relates to the field of printing roller roundness detection, and more specifically to an automatic method for detecting printing roller roundness. Background Technology

[0002] Gravure printing rollers are the core component of gravure printing presses, and their roundness accuracy directly affects the registration accuracy and printing quality of the printed materials. In the manufacturing and maintenance of gravure printing rollers, roundness inspection before shipment is a crucial process to ensure product quality.

[0003] Currently, the roundness detection of gravure printing rollers mainly uses laser displacement sensors or eddy current displacement sensors to collect radial displacement signals from the roller surface. The roundness is then determined by analyzing the fluctuation amplitude of the radial displacement signal. This method relies solely on a single radial displacement signal as the basis for judgment. During manufacturing, transportation, or storage, the roller surface is prone to the adhesion of anti-rust oil films, polishing wax, or fine particles. These deposits can also cause signal fluctuations in the displacement sensor similar to those caused by geometric protrusions. A single sensor cannot distinguish whether the fluctuations in the radial displacement signal originate from the roller's own geometric deviations or surface deposits, easily misjudging surface deposits as roundness defects, leading to qualified products being mistakenly identified as unqualified. Furthermore, instantaneous disturbances in the drive system can also cause abnormal fluctuations in the displacement signal, further increasing the possibility of misjudgment.

[0004] To address the aforementioned problems, this invention provides an intelligent detection method for the roundness of gravure printing rollers based on the angular domain overlap comparison of source and effect signals. Summary of the Invention

[0005] The purpose of this invention is to provide an automatic method for detecting the roundness of printing rollers, thereby solving the above-mentioned technical problems.

[0006] The objective of this invention can be achieved through the following technical solutions: A method for intelligent detection of the roundness of gravure printing rollers includes the following steps: S1. Install the gravure printing roller to be tested on the rotary drive device of the detection station and drive the gravure printing roller to rotate at a uniform speed. Simultaneously collect the instantaneous torque timing signal or instantaneous current timing signal of the gravure printing roller drive motor as the source timing signal, the radial displacement timing signal of the surface of the gravure printing roller as the effect timing signal, and the real-time angular position timing signal of the rotation of the gravure printing roller. S2. Based on the real-time angular position timing signal, the source-end timing signal and the effect-end timing signal are mapped and converted to the circumferential angle domain respectively to obtain the source-end angle domain signal and the effect-end angle domain signal. The source-end angle domain signal and the effect-end angle domain signal are then filtered respectively. S3. Obtain the reference fluctuation range of the source end angle domain signal and the reference displacement range of the effect end angle domain signal. Take the continuous angle range in the source end angle domain signal where the amplitude exceeds the reference fluctuation range as the source end abnormal angle span, and take the continuous angle range in the effect end angle domain signal where the amplitude exceeds the reference displacement range as the effect end abnormal angle span. S4. Overlap and compare the abnormal angle span of the source end and the abnormal angle span of the effect end on the circumferential angle coordinate. Based on the comparison results, determine whether there is a true roundness defect and record the defect location range. S5. If it exists, output the maximum deviation amplitude of the angle domain signal at the effective end within the defect location interval and the defect location interval and the maximum deviation amplitude.

[0007] As a preferred approach, the methods for obtaining the benchmark fluctuation range and benchmark displacement range include: During the stable running period when the gravure printing roller rotates at a constant speed for one revolution, the mean and standard deviation of the source end angle domain signal and the effect end angle domain signal are calculated over the entire circumference angle range. The standard deviation obtained by adding or subtracting the mean is used as the boundary of their respective reference intervals.

[0008] As a preferred option, if the overlap angle span between the abnormal angle span at the source end and the abnormal angle span at the effect end is less than a preset judgment angle threshold, then the first overlap ratio of the overlap angle span to the abnormal angle span at the source end and the second overlap ratio of the overlap angle span to the abnormal angle span at the effect end are calculated respectively. If both the first overlap ratio and the second overlap ratio exceed the preset ratio threshold, then it is determined that there is a true roundness defect in the overlap angle interval and it is recorded as the defect location interval.

[0009] As a preferred embodiment, if the first overlap ratio exceeds a preset ratio threshold while the second overlap ratio does not exceed the preset ratio threshold, then the first average abnormal amplitude of the source end angle domain signal within the abnormal angle span of the source end and the second average abnormal amplitude of the effective end angle domain signal within the overlapping angle interval are obtained, and the amplitude ratio of the first average abnormal amplitude to the second average abnormal amplitude is calculated. The corrected effective end angle length is obtained by weighting the angle length of the abnormal angle span of the effective end based on the amplitude ratio, and the second overlap ratio is recalculated based on the corrected effective end angle length. If the recalculated second overlap ratio exceeds the preset ratio threshold, it is determined that there is a true roundness defect in the overlap angle range.

[0010] As a preferred option, when the angle length of the abnormal angle span at the effective end is corrected by weighting according to the amplitude ratio, the weighting coefficient is proportional to the amplitude ratio, and the corrected angle length at the effective end is equal to the original angle length of the abnormal angle span at the effective end multiplied by the weighting coefficient.

[0011] As a preferred embodiment, if the first overlap ratio exceeds a preset ratio threshold while the second overlap ratio does not exceed the preset ratio threshold, and the amplitude ratio of the first average abnormal amplitude to the second average abnormal amplitude exceeds a preset amplitude ratio upper limit, then it is determined that there is no true roundness defect.

[0012] As a preferred embodiment, if the first overlap ratio does not exceed the preset ratio threshold while the second overlap ratio exceeds the preset ratio threshold, then the amplitude ratio of the third average abnormal amplitude to the fourth average abnormal amplitude is calculated based on the third average abnormal amplitude of the effective end angle domain signal within the effective end abnormal angle span and the fourth average abnormal amplitude of the source end angle domain signal within the overlapping angle interval. The source end angle length is obtained by weighting and correcting the angle length of the abnormal angle span based on the amplitude ratio, and the first overlap ratio is recalculated based on the corrected source end angle length. If the recalculated first overlap ratio exceeds the preset ratio threshold, it is determined that there is a true roundness defect in the overlap angle range.

[0013] As a preferred option, if the first overlap ratio does not exceed the preset ratio threshold while the second overlap ratio exceeds the preset ratio threshold, and the ratio of the third average abnormal amplitude to the fourth average abnormal amplitude exceeds the preset amplitude ratio upper limit, then it is determined that there is no true roundness defect.

[0014] As a preferred option, if neither the first overlap ratio nor the second overlap ratio exceeds a preset ratio threshold, then the source end abnormal angle span and the effect end abnormal angle span are each extended by a preset extension angle in the angle increasing direction and the angle decreasing direction, respectively, and then the overlap comparison is performed again; if the extended overlap angle span reaches or exceeds the preset judgment angle threshold, then it is determined that there is a true roundness defect in the extended overlap angle range.

[0015] As a preferred embodiment, the preset extension angle is inversely proportional to the current rotation speed of the gravure printing roller; the higher the current rotation speed, the smaller the preset extension angle.

[0016] The beneficial effects of this invention are: (1) This invention synchronously acquires the instantaneous torque timing signal or instantaneous current timing signal of the drive motor as the source signal and the radial displacement timing signal of the roller surface as the effect signal, and maps the two signals to the circumferential angle domain for overlap comparison. When there are geometric protrusions on the roller surface, the load of the drive motor will increase instantaneously when the protrusion rotates to the detection position, and the source signal and the effect signal will generate abnormalities simultaneously in the same angle range; while when there are only oil films or particles on the roller surface, only the effect signal generates abnormalities and the source signal does not respond. Based on this, it is possible to distinguish between true geometric defects and surface deposits, and reduce the probability of qualified products being misjudged as unqualified products.

[0017] (2) The present invention maps the source signal and effect signal acquired in the time domain to the angle domain for processing, so that the signal corresponds to the circumferential angle position of the printing roller. When the printing roller is detected at low speed, the rotation speed may fluctuate slightly due to the influence of the drive system. The width of the abnormal interval in the time domain signal will change with the rotation speed. After mapping to the angle domain, the width of the abnormal interval depends only on the circumferential angle range where the defect is located and is not affected by the rotation speed fluctuation. Thus, the consistency of the defect location interval output at different detection times for the same physical defect is guaranteed.

[0018] (3) This invention determines true roundness defects by comparing the abnormal angle span of the source end with the abnormal angle span of the effect end, rather than by directly using the simple judgment method of whether the signal amplitude exceeds the standard. The single signal amplitude exceeding the standard may be due to multiple factors. When two signals from different physical sources simultaneously exhibit abnormalities in the same angle range and reach the overlap requirement, the probability of geometric defects in that angle range is high. This judgment method based on signal overlap helps to reduce the risk of misjudgment in principle. Attached Figure Description

[0019] The invention will now be further described with reference to the accompanying drawings.

[0020] Figure 1 This is a diagram illustrating the method steps of the present invention. Detailed Implementation

[0021] 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.

[0022] See Figure 1 As shown, a method for intelligent detection of the roundness of gravure printing rollers is described. The specific implementation of signal acquisition and angle domain mapping in this method is as follows: The inspection station is located in the final inspection area for gravure printing rollers, and it is independent of the pressure roller system of the gravure printing machine. The gravure printing roller to be tested is mounted on a V-shaped support at the inspection station and is driven by a servo motor to rotate uniformly around its own axis at a speed ranging from 5 r / min to 20 r / min. During the inspection, the surface of the gravure printing roller is not contacted by any pressure rollers or other components, and it is in a state of free rotation without load.

[0023] A torque sensor is connected in series between the servo motor output shaft and the drive end of the gravure printing roller. The torque sensor acquires the instantaneous torque value of the servo motor in real time, with a sampling frequency set to 1000Hz, forming an instantaneous torque timing signal. A laser triangulation displacement sensor is fixedly installed on the outer surface of the gravure printing roller. This sensor is mounted on a rigid detection base independent of the gravure printing roller support structure, with its measurement axis perpendicular to the rotation axis of the gravure printing roller. The sensor acquires the radial displacement value of the roller surface in real time, with a sampling frequency set to 1000Hz, forming a radial displacement timing signal. An incremental rotary encoder is installed at the output shaft end of the servo motor. The encoder pulse count is 3600 pulses / revolution, rotating synchronously with the servo motor output shaft and outputting a real-time angular position timing signal. After acquisition begins, the data processing unit simultaneously performs timestamp alignment processing on the above three signals.

[0024] The data processing unit first preprocesses the instantaneous torque time-series signal and the radial displacement time-series signal, including removing the DC component and eliminating trend terms. Then, based on the real-time angular position time-series signal, the instantaneous torque time-series signal and the radial displacement time-series signal are mapped to the circumferential angle domain, respectively. Specifically, an angle sequence from 0° to 360° is established, with a step size of 0.1°. For each angle value in the angle sequence, the corresponding time is searched in the real-time angular position time-series signal. If there is no matching sampling point in the real-time angular position time-series signal, linear interpolation is used to obtain the corresponding time. After obtaining the time, the corresponding signal amplitude is searched in both the instantaneous torque time-series signal and the radial displacement time-series signal, again using linear interpolation. After mapping and transformation, the instantaneous torque time-series signal is converted into a source-end angle domain signal with angle as the abscissa, and the radial displacement time-series signal is converted into an effect-end angle domain signal with angle as the abscissa. The source-end angle domain signal and the effect-end angle domain signal have 3600 sampling points on the circumferential angle coordinates, corresponding to an angle interval of 0.1°.

[0025] After the mapping transformation is completed, angle domain filtering is performed on the source angle domain signal and the effect angle domain signal respectively. A finite-length unit impulse response low-pass filter is used to filter out signal components with wavelengths less than 1° in the angle domain.

[0026] The mapping transformation establishes a correspondence between the signal and the circumferential angular position of the gravure printing roller. During the low-speed rotation detection process of the gravure printing roller, the rotation speed may fluctuate slightly due to the influence of the drive system. The width of the abnormal interval in the time domain signal will change with the rotation speed. After mapping to the angular domain, the width of the abnormal interval depends only on the angular range occupied by the defect on the circumference and is not affected by the rotation speed fluctuation, thus ensuring the consistency of the defect location interval output at different detection times for the same physical defect.

[0027] The methods for obtaining the reference fluctuation range of the source-end angle domain signal and the reference displacement range of the effect-end angle domain signal, as well as the extraction process of the source-end abnormal angle span and the effect-end abnormal angle span, are as follows: During the stable operating period of the gravure printing roller rotating at a uniform speed for one revolution (i.e., within the range of 0° to 360°), 3600 sampling points of the source-end angle domain signal are taken. The arithmetic mean of these 3600 sampling points is calculated as the mean of the source-end signal. The square root of the sum of the squares of the differences between each sampling point and the mean, divided by the number of sampling points minus 1, is taken as the standard deviation of the source-end signal. The upper limit of the baseline fluctuation range of the source-end angle domain signal is the mean plus 3 times the standard deviation, and the lower limit is the mean minus 3 times the standard deviation. Similarly, 3600 sampling points of the effective-end angle domain signal are taken, and the mean and standard deviation of the effective-end signal are calculated. The upper limit of the baseline displacement range of the effective-end angle domain signal is the mean plus 3 times the standard deviation, and the lower limit is the mean minus 3 times the standard deviation.

[0028] Starting from 0°, the source-end angle domain signal is traversed point by point, determining whether the amplitude of each sampling point is greater than the upper limit or less than the lower limit of the reference fluctuation range. When the amplitude of a sampling point exceeds the reference fluctuation range, the sampling point is marked as an anomaly. Consecutive anomalies constitute a continuous angle interval, and the starting and ending angles of this interval are recorded as a source-end anomaly angle span. During the extraction process, if the interval between two anomaly angle intervals is less than the preset fallback tolerance threshold of 0.3°, the two intervals are merged into the same source-end anomaly angle span. The starting angle of the merged interval is the starting angle of the first interval, and the ending angle is the ending angle of the second interval. Similarly, continuous angle intervals with amplitudes exceeding the reference displacement range in the effect-end angle domain signal are extracted in the same way and used as effect-end anomaly angle spans. The same fallback tolerance threshold of 0.3° is used during the merging process.

[0029] After the above processing, the source-end angle domain signal and the effect-end angle domain signal each form one or more discrete anomalous angle spans. Each anomalous angle span includes three attributes: starting angle, ending angle, and angle length. The anomalous angle span characterizes the specific location and range of abnormal fluctuations in the signal along the circumferential angle, providing input data for subsequent overlap comparison. The reference interval is based on the statistical distribution of the full-circumference signal and can adaptively reflect the signal fluctuation level of different gravure printing rollers under normal conditions, eliminating the need to calibrate a reference value separately for each gravure printing roller.

[0030] The specific method for comparing the abnormal angular span at the source end and the abnormal angular span at the effect end on the circumferential angular coordinate system, and the rules for determining true roundness defects are as follows: For each angle interval in the source-end abnormal angle span set and each angle interval in the effect-end abnormal angle span set, determine whether the two angle intervals overlap on the circumferential angle coordinates. The condition for two angle intervals to overlap is: the starting angle of the first angle interval is less than or equal to the ending angle of the second angle interval, and the starting angle of the second angle interval is less than or equal to the ending angle of the first angle interval. When overlap exists, the starting angle of the overlapping angle interval is the larger of the two starting angles, the ending angle is the smaller of the two ending angles, and the overlapping angle span is equal to the ending angle minus the starting angle.

[0031] If the abnormal angle span at the source end and the abnormal angle span at the effect end overlap, and the overlap angle span reaches or exceeds a preset judgment angle threshold of 1°, then the overlapping angle interval is determined to have a true roundness defect, and the overlapping angle interval is recorded as the defect location interval. If the abnormal angle span at the source end does not exist but the abnormal angle span at the effect end does, then it is determined to be surface attachment. If the abnormal angle span at the effect end does not exist but the abnormal angle span at the source end does, then it is determined to be a drive system disturbance.

[0032] True geometric defects in gravure printing rollers physically manifest as localized bulges or depressions on the roller surface. When this defective area rotates to the detection position of the radial displacement sensor, the distance measured by the sensor deviates, resulting in an abnormal signal at the effective end. Simultaneously, as the defective area passes the detection position, it causes a momentary increase or decrease in the load on the drive motor, generating an abnormal signal at the source end. Since both abnormalities originate from the same physical defect, they overlap in the angular domain. Surface deposits only alter the measurement distance of the radial displacement sensor and do not affect the load on the drive motor; therefore, only the effective end signal is abnormal, while the source end signal remains unresponsive. Drive system disturbances only change the motor's output torque and do not cause a real change in the roller surface geometry; therefore, only the source end signal is abnormal, while the effective end signal remains unresponsive. Based on this, true roundness defects can be distinguished from surface deposits and drive system disturbances.

[0033] If both the source-end and effect-end abnormal angle spans have multiple discrete angle intervals, then each discrete angle interval is compared pairwise. If the source-end abnormal angle span set contains m intervals and the effect-end abnormal angle span set contains n intervals, then m×n pairs of angle intervals are compared. Each pair of overlapping angle intervals is determined to correspond to a true roundness defect, and their respective defect location intervals are recorded.

[0034] When the abnormal angle spans at the source and effect ends overlap but the overlapping angle span is less than the judgment angle threshold, the overlap ratio is used as a supplementary criterion. The first overlap ratio is the proportion of the overlapping angle span to the abnormal angle span at the source end, and the second overlap ratio is the proportion of the overlapping angle span to the abnormal angle span at the effect end. If both the first and second overlap ratios exceed the preset ratio threshold of 0.6, a true roundness defect is determined to exist in the overlapping angle interval and recorded as the defect location interval. The overlap ratio criterion ensures that true defects with slightly smaller overlapping angles than the threshold due to minor fluctuations in the signal boundary can still be detected.

[0035] When the first overlap ratio exceeds the ratio threshold but the second overlap ratio does not, the first average abnormal amplitude of the source-end angle domain signal within the abnormal angle span and the second average abnormal amplitude of the effective-end angle domain signal within the overlapping angle interval are extracted. The amplitude ratio of the first average abnormal amplitude to the second average abnormal amplitude is calculated. The angle length of the effective-end abnormal angle span is weighted and corrected based on the amplitude ratio. The second overlap ratio is recalculated using the corrected effective-end angle length. If the recalculated second overlap ratio exceeds the ratio threshold, a true roundness defect is determined to exist. If the amplitude ratio of the first average abnormal amplitude to the second average abnormal amplitude exceeds the upper limit of 5, a true roundness defect is determined not to exist. When the first overlap ratio does not exceed the threshold but the second overlap ratio does, a symmetrical processing method is used. This processing method can compensate for the angle length deviation caused by the difference in sensitivity between the source-end signal and the effective-end signal, making the overlap ratio determination more accurate.

[0036] When neither the first nor the second overlap ratio exceeds the ratio threshold, the source-end abnormal angle span and the effect-end abnormal angle span are each extended by one angle in the angle-increasing and angle-decreasing directions, respectively. The extension angle is set to 0.2°. The overlap is then re-compared using the extended angle interval. If the extended overlap angle span reaches or exceeds the judgment angle threshold of 1°, a true roundness defect is determined, and the defect location interval is recorded. The extension angle is inversely proportional to the current rotational speed of the gravure printing roller: 0.3° at 5 r / min, 0.2° at 10 r / min, and 0.1° at 20 r / min. This interval extension compensates for the problem of missed detection of near-overlap situations due to insufficient positioning accuracy at the signal abnormality boundary.

[0037] The processing method for defect location areas, the output format of detection results, and the implementation method for multi-axial section detection are as follows: For the defect location interval determined to be a true roundness defect, the maximum deviation amplitude of the effective end angle domain signal within this defect location interval is extracted. The maximum deviation amplitude is extracted as follows: within the range from the starting angle to the ending angle of the defect location interval, all sampling points of the effective end angle domain signal are traversed, and the deviation of the amplitude of each sampling point relative to the upper or lower limit of the reference displacement interval is calculated. The maximum value among all deviations is taken as the maximum deviation amplitude. If the sampling point amplitude is greater than the upper limit of the reference displacement interval, the deviation is the sampling point amplitude minus the upper limit value; if the sampling point amplitude is less than the lower limit of the reference displacement interval, the deviation is the lower limit value minus the sampling point amplitude; if the sampling point amplitude is within the reference displacement interval, the deviation is 0.

[0038] The data processing unit outputs the defect location range and maximum deviation amplitude to the display screen. The defect location range is displayed as a start angle and an end angle in degrees, and the maximum deviation amplitude is displayed in μm. In cases with multiple defect location ranges, they are output in ascending order of their start angles, with each range labeled with its corresponding maximum deviation amplitude. The data processing unit also compares the maximum deviation amplitude with a preset roundness tolerance. If the maximum deviation amplitude exceeds the roundness tolerance, a judgment of unqualified roundness of the gravure printing roller is output; otherwise, a judgment of qualified roundness is output. The roundness tolerance is obtained based on the design precision level of the gravure printing roller; in this embodiment, it is 5 μm.

[0039] When the axial length of the gravure printing roller is large, three laser triangulation displacement sensors are arranged at equal intervals along the axial direction of the roller, installed at 1 / 4, 1 / 2, and 3 / 4 positions along the axial length. The measurement axis of each sensor is perpendicular to the rotation axis of the roller, and the mounting bases are independent and fixed to a rigid detection base. Each sensor acquires the radial displacement timing signal at its respective axial section. The source signal is either a momentary torque timing signal or a momentary current timing signal from the drive motor. After mapping and conversion to the angular domain, the source-end angular domain signal is one signal, and the effective-end angular domain signals at each axial section are separate signals. The abnormal angle spans at the source and effect ends of each axial section are overlapped and compared to determine whether a true roundness defect exists at each axial section. The output of the detection results includes the defect location range, maximum deviation amplitude, and axial position of each axial section. The axial position is expressed as the distance from one end of the gravure printing roller, in mm. Multi-section inspection can assess the roundness of the gravure printing roller along its entire length, avoiding local defects that may be missed by single-section inspection.

[0040] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. An intelligent detection method for the roundness of a gravure printing roller, characterized in that, Includes the following steps: S1. Install the gravure printing roller to be tested on the rotary drive device of the detection station and drive the gravure printing roller to rotate at a uniform speed. Simultaneously collect the instantaneous torque timing signal or instantaneous current timing signal of the gravure printing roller drive motor as the source timing signal, the radial displacement timing signal of the surface of the gravure printing roller as the effect timing signal, and the real-time angular position timing signal of the rotation of the gravure printing roller. S2. Based on the real-time angular position timing signal, the source-end timing signal and the effect-end timing signal are mapped and converted to the circumferential angle domain respectively to obtain the source-end angle domain signal and the effect-end angle domain signal. The source-end angle domain signal and the effect-end angle domain signal are then filtered respectively. S3. Obtain the reference fluctuation range of the source end angle domain signal and the reference displacement range of the effect end angle domain signal. Take the continuous angle range in the source end angle domain signal where the amplitude exceeds the reference fluctuation range as the source end abnormal angle span, and take the continuous angle range in the effect end angle domain signal where the amplitude exceeds the reference displacement range as the effect end abnormal angle span. S4. Overlap and compare the abnormal angle span of the source end and the abnormal angle span of the effect end on the circumferential angle coordinate. Based on the comparison results, determine whether there is a true roundness defect and record the defect location range. S5. If it exists, output the maximum deviation amplitude of the angle domain signal at the effective end within the defect location interval and the defect location interval and the maximum deviation amplitude.

2. The method for detecting the roundness of a gravure printing roller according to claim 1, wherein The methods for obtaining the benchmark fluctuation range and benchmark displacement range include: During the stable running period when the gravure printing roller rotates at a constant speed for one revolution, the mean and standard deviation of the source end angle domain signal and the effect end angle domain signal are calculated over the entire circumference angle range. The standard deviation obtained by adding or subtracting the mean is used as the boundary of their respective reference intervals.

3. The intelligent detection method for the roundness of a gravure printing roller according to claim 1, characterized in that, If the overlap angle span between the abnormal angle span at the source end and the abnormal angle span at the effect end is less than a preset judgment angle threshold, then the first overlap ratio of the overlap angle span to the abnormal angle span at the source end and the second overlap ratio of the overlap angle span to the abnormal angle span at the effect end are calculated respectively. If both the first overlap ratio and the second overlap ratio exceed the preset ratio threshold, then it is determined that there is a true roundness defect in the overlap angle interval and it is recorded as the defect location interval.

4. The intelligent detection method for the roundness of a gravure printing roller according to claim 3, characterized in that, If the first overlap ratio exceeds the preset ratio threshold while the second overlap ratio does not exceed the preset ratio threshold, then the first average abnormal amplitude of the source end angle domain signal within the abnormal angle span of the source end and the second average abnormal amplitude of the effective end angle domain signal within the overlapping angle interval are obtained, and the amplitude ratio of the first average abnormal amplitude to the second average abnormal amplitude is calculated. The corrected effective end angle length is obtained by weighting the angle length of the abnormal angle span of the effective end based on the amplitude ratio, and the second overlap ratio is recalculated based on the corrected effective end angle length. If the recalculated second overlap ratio exceeds the preset ratio threshold, it is determined that there is a true roundness defect in the overlap angle range.

5. The intelligent detection method for the roundness of a gravure printing roller according to claim 4, characterized in that, When the angle length of the abnormal angle span at the effective end is corrected by weighting the amplitude ratio, the weighting coefficient is proportional to the amplitude ratio. The corrected angle length at the effective end is equal to the original angle length of the abnormal angle span at the effective end multiplied by the weighting coefficient.

6. The intelligent detection method for the roundness of a gravure printing roller according to claim 4, characterized in that, If the first overlap ratio exceeds a preset ratio threshold while the second overlap ratio does not exceed the preset ratio threshold, and the amplitude ratio of the first average abnormal amplitude to the second average abnormal amplitude exceeds a preset amplitude ratio upper limit, then it is determined that there is no true roundness defect.

7. The intelligent detection method for the roundness of a gravure printing roller according to claim 3, characterized in that, If the first overlap ratio does not exceed the preset ratio threshold while the second overlap ratio exceeds the preset ratio threshold, then the ratio of the third average abnormal amplitude to the fourth average abnormal amplitude is calculated based on the third average abnormal amplitude of the effective end angle domain signal within the effective end abnormal angle span and the fourth average abnormal amplitude of the source end angle domain signal within the overlapping angle interval. The source end angle length is obtained by weighting and correcting the angle length of the abnormal angle span based on the amplitude ratio, and the first overlap ratio is recalculated based on the corrected source end angle length. If the recalculated first overlap ratio exceeds the preset ratio threshold, it is determined that there is a true roundness defect in the overlap angle range.

8. The intelligent detection method for the roundness of a gravure printing roller according to claim 7, characterized in that, If the first overlap ratio does not exceed the preset ratio threshold, but the second overlap ratio exceeds the preset ratio threshold, and the ratio of the third average abnormal amplitude to the fourth average abnormal amplitude exceeds the preset amplitude ratio upper limit, then it is determined that there is no true roundness defect.

9. The intelligent detection method for the roundness of a gravure printing roller according to claim 3, characterized in that, If neither the first overlap ratio nor the second overlap ratio exceeds the preset ratio threshold, then the source end abnormal angle span and the effect end abnormal angle span are each extended by a preset extension angle in the angle increasing direction and the angle decreasing direction, respectively, and then the overlap comparison is performed again; if the extended overlap angle span reaches or exceeds the preset judgment angle threshold, then it is determined that there is a true roundness defect in the extended overlap angle interval.

10. The intelligent detection method for the roundness of a gravure printing roller according to claim 9, characterized in that, The preset extension angle is inversely proportional to the current rotation speed of the gravure printing roller; the higher the current rotation speed, the smaller the preset extension angle.

Images