A smart production control system and method for tinplate printing

Abstract

This invention relates to the field of intelligent manufacturing technology, specifically to an intelligent production control system and method for tinplate printing. The system includes a surface temperature sensing module, a thermal deformation calculation module, an alignment image analysis module, a displacement calculation module, and a pulse frequency control module. In this invention, by acquiring continuous voltage waveform sequences from front-end equipment and identifying discrete amplitude characteristics, the system extracts level change information and converts it into surface temperature values, overcoming the drawbacks of manual temperature measurement lag. It compares the ambient temperature with the standard state to extract the state temperature difference offset, integrates the linear expansion reference value of the sheet metal, performs scaling transformation, generates absolute elongation deformation values ​​to compensate for thermal deformation, scans the grayscale pixel array of registration marks, filters the target point set, performs pixel centroid extraction, generates crosshair alignment center coordinates to deduce planar position deviation, performs discrete fractional decomposition of the alignment error scalar, calculates the total pulse volume, determines polarity, reshapes the operating frequency, generates correction commands, and achieves automated and precise registration control.

Description

Technical Field

[0001] This invention relates to the field of intelligent manufacturing technology, and in particular to an intelligent production control system and method for tinplate printing. Background Technology

[0002] The field of intelligent manufacturing technology encompasses the entire lifecycle management process from production planning and material flow to on-site equipment operation. Its core aspects include collecting physical quantities from underlying sensors, using programmable logic controllers to operate actuators such as motors and cylinders, and coordinating production orders, process formulas, and quality parameters to construct an industrial production system. Traditional intelligent production control systems for tinplate printing refer to systems that manage specific technical aspects of the metal packaging printing process, such as order scheduling, ink mixing, oven heating, and equipment operation status monitoring. Traditionally, this involves manually filling out paper production work orders to assign tasks. Operators manually move the tinplate to the printing press's feeding table, manually compare the ink to a standard color chart, mix the ink, and perform sample testing with a scraper. During the baking process, workers manually rotate temperature control knobs to set the current of the heating resistance wire in the oven. Meanwhile, inspectors, equipped with infrared thermometers and portable vibration pens, periodically record the surface temperature and vibration frequency of the spindle bearings near the coating and printing machines. Finally, production progress and raw material consumption data are entered into an electronic spreadsheet file on a desktop computer by a statistician at the end of each shift.

[0003] Traditional tinplate production control relies on manual document transfer, task assignment, and material handling. The production scheduling cycle is lengthy and susceptible to human fatigue. Manual comparison of color cards and ink mixing ratios by hand can easily lead to color deviations. The heating process in the drying room relies on manual adjustment of the current, which cannot achieve dynamic and precise temperature control. The use of portable instruments to collect equipment operating parameters at regular intervals causes gaps in monitoring data and serious delays in feedback. The centralized entry of production data afterward blocks real-time linkage and in-depth traceability of on-site conditions. The overall operating mode results in extremely poor process consistency and serious waste of resources. Summary of the Invention

[0004] To address the shortcomings of existing tinplate printing production control technologies, which rely on manual document delivery, task assignment, and material handling, resulting in lengthy production scheduling cycles susceptible to human fatigue, and the high risk of color deviation due to manual comparison of color charts and ink mixing ratio testing, the invention provides an intelligent tinplate printing production control system and method. The technical solution is as follows: On the one hand, a smart production control system for tinplate printing is provided, which includes: The surface temperature sensing module acquires the continuous voltage waveform sequence transmitted by the front-end device, identifies the characteristics of discrete amplitude data, inputs it into the support vector regression model to extract level change information, and converts it into surface temperature values. The heat deformation calculation module obtains the preset standard state ambient temperature value, compares the surface temperature value with the standard state ambient temperature value to extract the state temperature difference offset, and performs scaling transformation operation on the state temperature difference offset by integrating the inherent linear expansion reference value of the printed iron plate and the plate reference length constant to generate the absolute elongation deformation value. The alignment image parsing module acquires a two-dimensional grayscale image pixel array of the registration mark edge region, scans the two-dimensional grayscale image pixel array to filter the set of target pixel coordinate points within a predetermined grayscale range, performs centroid extraction operation, and generates cross alignment center coordinate values. The displacement calculation module compares the cross alignment center coordinates with the pre-stored ideal reference alignment center coordinates to deduce the planar position deviation, and combines the absolute elongation deformation value with the planar position deviation to construct an alignment error scalar. The pulse frequency control module obtains the single-step servo displacement constant, performs a fractional discretization operation on the alignment error scalar, obtains the total amount of compensation execution pulses, determines the polarity sign orientation, reshapes the basic operating frequency value, and generates a dynamic alignment servo correction command.

[0005] As a further aspect of the present invention, the surface temperature values ​​include peak heating point readings, steady-state heating average values, and regional thermal radiance; the absolute elongation deformation values ​​include longitudinal tensile length, transverse extension width, and micro buckling increment; the cross alignment center coordinate values ​​include horizontal axis positioning anchor points, vertical axis intersection poles, and origin spatial vectors; the alignment error scalars include geometric offset amplitude, angular deflection margin, and overall misalignment spacing; and the dynamic registration servo correction command includes motor drive timing, duty cycle adjustment signal, and phase compensation period.

[0006] As a further aspect of the present invention, the surface temperature sensing module includes: The waveform feature perception submodule acquires the continuous voltage waveform sequence transmitted by the front-end device, extracts the voltage amplitude data items corresponding to each discrete time node in the continuous voltage waveform sequence according to the preset frequency reference, arranges the voltage amplitude data items into a one-dimensional vector matrix, calculates the amplitude difference between adjacent elements in the one-dimensional vector matrix, extracts the vector elements corresponding to those that cross the preset fluctuation boundary threshold, and generates discrete amplitude feature vectors. The level change operation submodule substitutes the discrete amplitude feature vector into the configuration mapping matrix to perform an inner product operation, calculates the distance measure value between the discrete amplitude feature vector and each support vector in the mapping matrix, extracts the abnormal component data corresponding to the distance measurement value that crosses the preset support boundary spacing benchmark, performs weighted summation calculation on the abnormal component data to extract the corresponding deviation index information item, and obtains the level change deviation rate. The temperature value construction submodule collects the basic temperature reference value at the reference environment node, multiplies the level change deviation rate by a preset temperature coefficient to convert it into a temperature compensation value, sums it with the basic temperature reference value, calculates the deviation compensation metric value corresponding to the basic temperature reference value, adds the deviation compensation metric value to the basic temperature reference value to obtain the merged summation state metric item, removes out-of-bounds abnormal numerical items, and obtains the surface temperature value.

[0007] As a further aspect of the present invention, the thermal deformation calculation module includes: The surface temperature difference calculation submodule collects the preset standard state ambient temperature value, calls the surface temperature value, extracts the difference item between the standard state ambient temperature value and the surface temperature value, inputs the difference item into the comparison operation node to perform the subtraction operation, calculates the distance measurement value where the standard state ambient temperature value deviates from the surface temperature value, performs an absolute value conversion operation on the distance measurement value, extracts the unsigned difference item, and establishes the state temperature difference offset. The scaling transformation submodule collects the inherent linear expansion reference value of the printed iron sheet, assigns the linear expansion reference value to the corresponding coefficient term position within the state temperature difference offset, performs a product operation on the state temperature difference offset and the linear expansion reference value to obtain the scaling transformation product term, determines whether the scaling transformation product term exceeds the preset elongation extreme value threshold and related out-of-bounds parts, removes the related out-of-bounds parts, and obtains the absolute elongation deformation value.

[0008] As a further aspect of the present invention, the alignment image parsing module includes: The pixel array acquisition submodule acquires a two-dimensional grayscale image pixel array in the registration mark edge region, reads the spatial position associated grayscale measurement terms within the two-dimensional grayscale image pixel array, extracts the difference terms between the grayscale measurement terms and the preset background reference value, determines invalid data units corresponding to the difference terms exceeding the preset noise tolerance threshold, removes invalid data units and retains the remaining associated grayscale measurement terms, and generates the target grayscale matrix. The target coordinate filtering submodule extracts the grayscale measurement items carried by each data node in the target grayscale matrix, collects a predetermined grayscale range threshold, performs a comparison operation between the grayscale measurement items and the predetermined grayscale range threshold, filters the associated data nodes within the predetermined grayscale range threshold, extracts the row and column identifier sequences corresponding to the associated data nodes, merges the row and column identifier sequences corresponding to the associated data nodes, and obtains the target pixel coordinate point set. The centroid coordinate calculation submodule extracts the horizontal and vertical identifier sequences within the target pixel coordinate point set, calculates the sum of the horizontal identifier sequences as the horizontal cumulative term, calculates the sum of the vertical identifier sequences as the vertical cumulative term, reads the total number of nodes as the divisor reference, performs division operations between the horizontal and vertical cumulative terms and the divisor reference respectively to extract the mean term, merges the mean terms into spatial coordinate components, and establishes the cross alignment center coordinate values.

[0009] As a further aspect of the present invention, the upper noise tolerance value and the lower noise tolerance value included in the preset noise tolerance threshold are extracted. The difference item is numerically compared with the upper noise tolerance value and the lower noise tolerance value. The grayscale measurement item associated with the difference item being greater than the upper noise tolerance value is extracted as the high-end noise measurement item, and the grayscale measurement item associated with the difference item being less than the lower noise tolerance value is extracted as the low-end noise measurement item. The high-end noise measurement item and the low-end noise measurement item are merged as invalid data units.

[0010] As a further aspect of the present invention, the displacement calculation module includes: The planar deviation deduction submodule collects the predetermined ideal reference coordinate values, substitutes the cross alignment center coordinate values ​​and the ideal reference coordinate values ​​into the difference node to perform a subtraction operation, calculates the one-dimensional displacement difference term, performs a sum of squares and square root operation on the one-dimensional displacement difference term to extract the Euclidean distance metric term, compares the Euclidean distance metric term with the preset allowable limit reference value and removes out-of-limit distortion data terms, and generates the planar position deviation amount. The error comparison submodule, for the planar position deviation, combines the absolute elongation deformation value to extract the directional component item carried by the planar position deviation and the scaling scale item contained in the absolute elongation deformation value. The directional component item and the scaling scale item are input into the fusion node to perform a product operation, calculate the spatial error combination metric, compare the spatial error combination metric with the preset extreme error benchmark, remove divergent outlier data, and obtain the alignment error scalar.

[0011] As a further aspect of the present invention, the cross alignment center coordinate values ​​are extracted to include the cross alignment center horizontal value and the cross alignment center vertical value, and the ideal reference coordinate values ​​are extracted to include the ideal reference horizontal value and the ideal reference vertical value. The cross alignment center horizontal value and the ideal reference horizontal value are subtracted to obtain a horizontal one-dimensional displacement difference term, and the cross alignment center vertical value and the ideal reference vertical value are subtracted to obtain a vertical one-dimensional displacement difference term. The horizontal one-dimensional displacement difference term and the vertical one-dimensional displacement difference term are combined as a one-dimensional displacement difference term.

[0012] As a further aspect of the present invention, the pulse frequency control module includes: The error share splitting submodule obtains the single-step servo running displacement constant, performs a division calculation between the alignment error scalar and the single-step servo running displacement constant, extracts the quotient data item, extracts the integer component associated with the quotient data item as the share discrete splitting value, compares the share discrete splitting value with the preset pulse saturation benchmark threshold to remove out-of-bounds excess data, retains the non-out-of-bounds regular splitting value, and obtains the total amount of compensation execution pulses. The polarity direction discrimination submodule extracts the positive and negative algebraic sign terms carried by the total amount of compensation execution pulses, performs matching and verification between the positive and negative algebraic sign terms and the predetermined positive operation reference mark, extracts the positive displacement state quantity when the positive and negative algebraic sign terms match the positive operation reference mark, and extracts the reverse displacement state quantity when the positive and negative algebraic sign terms deviate from the positive operation reference mark, and establishes the polarity sign guidance parameter. The frequency reshaping submodule monitors the base operating frequency value, reads the direction factor in the polarity sign guide parameter and assigns it to the corresponding phase angle bit of the base operating frequency value, performs a comparison operation between the base operating frequency value and the predetermined upper limit benchmark, cuts out the overclocking band data in the base operating frequency value that exceeds the upper limit benchmark, assembles the remaining regular base operating frequency value and direction factor and converts them into a control layer machine code sequence, and generates a dynamic registration servo correction instruction.

[0013] On the other hand, a method for intelligent production control of tinplate printing, which is executed based on the aforementioned intelligent production control system for tinplate printing, includes the following steps: S1: Acquire the continuous voltage waveform sequence transmitted by the front-end device, identify the characteristics of discrete amplitude data, input the data into the support vector regression model to extract level change information, and convert it into surface temperature values; S2: Obtain the preset standard state ambient temperature value, compare the surface temperature value with the standard state ambient temperature value to extract the state temperature difference offset, and perform scaling transformation operation on the state temperature difference offset by integrating the inherent linear expansion reference amount of the printed iron plate and the plate reference length constant to generate the absolute elongation deformation value. S3: Acquire a two-dimensional grayscale image pixel array of the registration mark edge region, scan the two-dimensional grayscale image pixel array to filter the target pixel coordinate point set within the predetermined grayscale range, perform centroid extraction operation, and generate cross alignment center coordinate values; S4: Compare the cross alignment center coordinates with the pre-stored ideal reference alignment center coordinates to deduce the planar position deviation, and combine the absolute elongation deformation value with the planar position deviation to construct the alignment error scalar. S5: Obtain the single-step servo running displacement constant, perform a fractional discretization operation on the alignment error scalar, obtain the total amount of compensation execution pulses, determine the polarity sign orientation, reshape the basic running frequency value, and generate a dynamic alignment servo correction command.

[0014] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following: By acquiring continuous voltage waveform sequences from front-end devices and identifying discrete amplitude characteristics, the information on sudden level changes is extracted and converted into surface temperature values. This overcomes the drawbacks of manual temperature measurement lag. By comparing with the standard ambient temperature, the temperature difference offset is extracted. The linear expansion reference value of the board is fused, and scaling transformation is performed to generate absolute elongation deformation values ​​to compensate for thermal deformation. The grayscale pixel array of registration marks is scanned, the target point set is selected, and pixel centroid extraction is performed to generate cross alignment center coordinates to deduce the planar position deviation. The deformation and position deviation are merged, and the alignment error scalar is constructed for discrete fractional decomposition. The total pulse amount is calculated to determine polarity and reshape the operating frequency. Correction commands are generated to achieve automated and precise registration control. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a schematic diagram of the system provided by the present invention; Figure 2 This is a schematic diagram of the system framework of the present invention; Figure 3 This is a flowchart of the surface temperature sensing module in this invention; Figure 4 This is a flowchart of the thermal deformation calculation module in this invention; Figure 5 This is a flowchart of the alignment image parsing module in this invention; Figure 6 This is a flowchart of the displacement calculation module in this invention; Figure 7 This is a flowchart of the pulse frequency control module in this invention; Figure 8 This is a flowchart of the method provided by the present invention. Detailed Implementation

[0017] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0018] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0019] This invention provides an intelligent production control system for tinplate printing, such as... Figure 1-2The diagram shown is of a smart production control system for tinplate printing. This system includes: The surface temperature sensing module acquires the continuous voltage waveform sequence transmitted by the front-end device, identifies the characteristics of discrete amplitude data, inputs the discrete amplitude data characteristics into the support vector regression model to extract level change information, and converts it into surface temperature values. The heat deformation calculation module obtains the preset standard state ambient temperature value based on the surface temperature value, compares the surface temperature value with the standard state ambient temperature value to extract the state temperature difference offset, and performs scaling transformation operation on the state temperature difference offset by integrating the inherent linear expansion reference value of the printed iron plate and the plate reference length constant to generate the absolute elongation deformation value. The alignment image parsing module acquires a two-dimensional grayscale image pixel array of the registration mark edge region, scans the two-dimensional grayscale image pixel array to filter the target pixel coordinate point set within a predetermined grayscale range, performs centroid extraction operation on the target pixel coordinate point set, and generates cross alignment center coordinate values. The displacement calculation module extracts the pre-stored ideal reference alignment center coordinates based on the cross alignment center coordinates, compares the cross alignment center coordinates with the ideal reference alignment center coordinates to deduce the planar position deviation, and merges the absolute elongation deformation value with the planar position deviation to construct the alignment error scalar. The pulse frequency control module obtains the single-step servo operation displacement constant based on the alignment error scalar, performs a fractional discretization operation on the alignment error scalar based on the single-step servo operation displacement constant, obtains the total amount of compensation execution pulses, determines the polarity sign orientation of the total amount of compensation execution pulses, reshapes the basic operating frequency value according to the polarity sign orientation, and generates a dynamic alignment servo correction command. Surface temperature values ​​include peak heating point readings, steady-state average heating value, and regional thermal radiation. Absolute elongation deformation values ​​include longitudinal tensile length, transverse extension width, and micro buckling increment. Cross alignment center coordinate values ​​include horizontal axis positioning anchor point, vertical axis intersection pole, and origin spatial vector. Alignment error scalars include geometric offset amplitude, angular deflection margin, and overall misalignment spacing. Dynamic registration servo correction commands include motor drive timing, duty cycle adjustment signal, and phase compensation period.

[0020] Specifically, such as Figure 2 , 3 As shown, the surface temperature sensing module includes: The waveform feature perception submodule acquires the continuous voltage waveform sequence transmitted by the front-end device, extracts the voltage amplitude data items corresponding to each discrete time node in the continuous voltage waveform sequence according to the preset frequency reference, arranges the voltage amplitude data items into a one-dimensional vector matrix, calculates the amplitude difference between adjacent elements in the one-dimensional vector matrix, extracts the vector elements corresponding to those that cross the preset fluctuation boundary threshold, and generates discrete amplitude feature vectors. The continuous voltage waveform sequence transmitted from the front-end device is collected through a high-precision voltage probe acquisition channel. Through process control, and relying on the concurrent multi-threaded processing architecture of edge computing nodes, low-latency synchronous acquisition of the underlying data is achieved, ensuring high real-time performance of the waveform input. The preset frequency reference is set to 2000 sampling actions per second. The waveform feature perception submodule converts the acquired analog signal into a digital sequence via an analog-to-digital converter and stores it in the system's dynamic random access memory buffer. The waveform feature perception submodule processes the continuous voltage waveform sequence using a moving average filtering mechanism, removing high-frequency glitches that deviate from the local mean by more than 15%. Then, it extracts the voltage amplitude data items corresponding to each discrete time point within the continuous voltage waveform sequence according to the preset frequency reference. The waveform feature perception submodule arranges the extracted voltage amplitude data items in chronological order into a 1D vector matrix, with the data structure represented as a single-row array containing multiple floating-point values. The waveform feature perception submodule executes an adjacent data calculation mechanism. It inputs the voltage amplitude data item of the next time node in the 1D vector matrix into the differential operation unit for subtraction, calculating the amplitude difference between adjacent elements in the 1D vector matrix. The waveform feature perception submodule extracts vector elements that exceed a preset fluctuation boundary threshold. This threshold is determined by collecting sample data from 1000 fault-free operating conditions, recording the adjacent amplitude differences of all samples, taking the maximum absolute value of all fault-free differences, and adding a 10% tolerance margin. Specifically, the preset fluctuation boundary threshold is set to 0.5 mV. When the amplitude difference exceeds 0.5 mV, the waveform feature perception submodule identifies it as an out-of-bounds element. The waveform feature perception submodule then sequentially combines the extracted elements to generate a discrete amplitude feature vector. For example, if the voltage amplitude data items for four consecutive time points are 1.2 mV, 1.5 mV, 1.8 mV, and 2.5 mV, the waveform feature perception submodule subtracts 1.2 mV from 1.5 mV to get 0.3 mV, subtracts 1.5 mV from 1.8 mV to get 0.3 mV, and subtracts 1.8 mV from 2.5 mV to get 0.7 mV. Comparing this to the preset fluctuation boundary threshold of 0.5 mV, the waveform feature perception submodule only extracts 0.7 mV as the out-of-bounds element and generates a discrete amplitude feature vector.

[0021] The level change operation submodule substitutes the discrete amplitude feature vector into the configuration mapping matrix to perform inner product operation, calculates the distance measure value between the discrete amplitude feature vector and each support vector in the mapping matrix, extracts the abnormal component data corresponding to the distance measurement value that crosses the preset support boundary spacing benchmark, performs weighted summation calculation on the abnormal component data to extract the corresponding deviation index information item, and obtains the level change deviation rate. The configuration mapping matrix is ​​a set of pre-calibrated historical fault feature template data stored in read-only memory. The level mutation operation submodule performs corresponding multiplication operations on each numerical component of the discrete amplitude feature vector with the corresponding dimension element of each support vector in the configuration mapping matrix. Then, it sums all the product results to calculate the distance metric between the discrete amplitude feature vector and each support vector in the mapping matrix. The level mutation operation submodule extracts abnormal component data corresponding to distance metric values ​​that cross a preset support boundary distance benchmark. The preset support boundary distance benchmark is set to 15, and this value is iteratively tested in the range of 10 to 20 using a grid search algorithm with a step size of 1. The highest classification accuracy point is locked at 15. The level mutation operation submodule compares the distance metric value with the preset support boundary distance benchmark of 15. When the distance metric value is greater than 15, it is marked as abnormal component data and extracted. The level mutation operation submodule performs weighted summation calculations on the abnormal component data to extract corresponding deviation index information items, multiplying each abnormal component data with its corresponding dimension's preset feature weight coefficient and summing the results. The preset feature weighting coefficients are assigned based on the Pearson correlation coefficients of different features with respect to temperature drift in the historical fault database. Higher correlation coefficients result in larger weighting coefficients, thus obtaining the level change deviation rate. For example, if the abnormal component data contains two elements, 4 and 6, the corresponding feature weighting coefficients are set to 0.3 and 0.7. The level change operation submodule multiplies 4 by 0.3 to obtain 1.2, and multiplies 6 by 0.7 to obtain 4.2. After summing these values, the level change deviation rate is calculated to be 5.4.

[0022] The temperature value construction submodule collects the basic temperature reference value at the reference environment node, multiplies the level change deviation rate by the preset temperature coefficient to convert it into a temperature compensation value, sums it with the basic temperature reference value, calculates the deviation compensation metric value corresponding to the basic temperature reference value, adds the deviation compensation metric value to the basic temperature reference value to obtain the merged summation state metric item, removes out-of-bounds abnormal numerical items, and obtains the surface temperature value. A baseline temperature reference value is acquired at a reference environmental node using a patch thermocouple sensor. This baseline temperature reference value is obtained by averaging five consecutive measurements; for example, the calculated result is 25 degrees Celsius. The temperature value construction submodule multiplies the level change deviation rate by a preset temperature coefficient to convert it into a temperature compensation value. This value is then summed with the baseline temperature reference value. The level change deviation rate is multiplied by a preset temperature conversion factor of 0.5 degrees Celsius per unit deviation to calculate the deviation compensation metric corresponding to the baseline temperature reference value. For example, the level change calculation submodule multiplies the level change deviation rate of 5.4 by the temperature conversion factor of 0.5 to calculate a deviation compensation metric of 2.7 degrees Celsius. The temperature value construction submodule adds the deviation compensation metric to the baseline temperature reference value to obtain a combined summed state metric. Adding 25 degrees Celsius to 2.7 degrees Celsius yields a combined summed state metric of 27.7 degrees Celsius. The temperature value construction submodule determines whether the merged summation state measurement item is between -10 degrees Celsius and the preset high temperature warning threshold of 150 degrees Celsius. If it is within the normal range, it is retained; if it exceeds the range, it performs the process of removing out-of-bounds abnormal numerical items, and finally obtains a surface temperature value of 27.7 degrees Celsius.

[0023] Table 1: Temperature Sensing Parameters at Each Measuring Point

[0024] Table 1 shows the basic reference parameters at different measuring points and the surface temperature values ​​derived from the above calculation logic.

[0025] Specifically, such as Figure 2 , 4 As shown, the thermal deformation calculation module includes: The surface temperature difference calculation submodule collects the preset standard state ambient temperature value, calls the surface temperature value, extracts the difference item between the standard state ambient temperature value and the surface temperature value, inputs the difference item into the comparison operation node to perform the subtraction operation, calculates the distance measurement value where the standard state ambient temperature value deviates from the surface temperature value, performs an absolute value conversion operation on the distance measurement value, extracts the unsigned difference item, and establishes the state temperature difference offset. The ambient temperature is collected by an environmental meteorological monitoring probe under a preset standard condition. This preset standard condition ambient temperature is written into a configuration table based on the historical average of the constant temperature workshop over 30 consecutive days, specifically set to 20 degrees Celsius. The surface temperature difference calculation submodule calls the previously acquired surface temperature value of 27.7 degrees Celsius and extracts the difference between the standard condition ambient temperature value and the surface temperature value. The surface temperature difference calculation submodule inputs the difference into a comparison operation node to perform a subtraction operation. Specifically, the standard condition ambient temperature value is subtracted from the surface temperature value to calculate the distance measurement value where the standard condition ambient temperature value deviates from the surface temperature value. The surface temperature difference calculation submodule performs an absolute value conversion operation on the distance measurement value, determines the sign of the highest algebraic bit of the distance measurement value. If the highest sign bit is 0, it represents a positive number, and the original value remains unchanged. If the highest sign bit is 1, it represents a negative number, and the sign bit is cleared or masked to remove the negative sign, extracting the unsigned difference term. Through the above calculations, the surface temperature difference calculation submodule subtracts 20 degrees Celsius from 27.7 degrees Celsius to obtain 7.7 degrees Celsius. After absolute value conversion, it is still 7.7 degrees Celsius, and the state temperature difference offset is established to be 7.7 degrees Celsius.

[0026] The scaling transformation submodule collects the inherent linear expansion reference value of the printed iron sheet, assigns the linear expansion reference value to the corresponding coefficient term position within the state temperature difference offset, performs a product operation on the state temperature difference offset and the linear expansion reference value, obtains the scaling transformation product term, determines whether the scaling transformation product term exceeds the preset elongation extreme value threshold and related out-of-bounds parts, removes the related out-of-bounds parts, and obtains the absolute elongation deformation value. The inherent linear expansion reference value of the printed steel sheet is collected from the equipment basic material configuration database. This reference value is entered into the system from the material supplier's factory property test report, specifically set as an expansion ratio of 0.000012 units of length per degree Celsius. The scaling transformation submodule assigns the linear expansion reference value to the corresponding coefficient term within the state temperature difference offset. It then inputs the state temperature difference offset and the linear expansion reference value into the multiplication register of the central processing unit for product operation, multiplying the state temperature difference offset (7.7 degrees Celsius) with the linear expansion reference value (0.000012). The scaling transformation submodule reads the initial reference length parameter of the sheet material via the system bus, setting the initial reference length to 1000 mm. It then performs a multiplication operation on the state temperature difference offset, the linear expansion reference value, and the initial reference length to obtain the scaling transformation product term. The specific numerical values ​​are substituted into the calculation results. The scaling transformation submodule multiplies 7.7 degrees Celsius by 0.000012 and 1000 millimeters, resulting in a scaling transformation product term of 0.0924 millimeters. The scaling transformation submodule determines whether the scaling transformation product term exceeds a preset elongation extreme threshold, which is set to 2.0 millimeters based on the maximum tolerance limit of the mechanical guide rail of the printing equipment. The scaling transformation submodule inputs the scaling transformation product term 0.0924 millimeters and the preset elongation extreme threshold 2.0 millimeters into a comparator for numerical comparison. If the scaling transformation product term is greater than the preset elongation extreme threshold, it is determined that an out-of-bounds connection has occurred, and the system cuts off subsequent data streams and removes the out-of-bounds portion; if it does not exceed the limit, the data is directly saved to the result register, and the absolute elongation deformation value is obtained as 0.0924 millimeters.

[0027] Specifically, such as Figure 2 , 5 As shown, the alignment image parsing module includes: The pixel array acquisition submodule acquires a two-dimensional grayscale image pixel array in the registration mark edge region, reads the spatial position associated grayscale measurement terms within the two-dimensional grayscale image pixel array, extracts the difference terms between the grayscale measurement terms and the preset background reference value, determines invalid data units corresponding to the difference terms exceeding the preset noise tolerance threshold, removes invalid data units and retains the remaining associated grayscale measurement terms, and generates the target grayscale matrix. A 2D grayscale image pixel array of the registered edge region is acquired using an industrial-grade charge-coupled device (CCD) camera. Through process control and a hardware-level image caching pipeline, lossless image quality is achieved through real-time image dumping. The acquisition resolution is set to 1000 pixels horizontally by 1000 pixels vertically, generating matrix data containing 1,000,000 data units. The pixel array acquisition submodule traverses and reads the spatially associated grayscale measurement term within the 2D grayscale image pixel array. This grayscale measurement term is quantized into an 8-bit unsigned integer value between 0 and 255. The pixel array acquisition submodule extracts the difference term between the grayscale measurement term and a preset background reference value. The preset background reference value is obtained by extracting a 50x50 unmarked blank pixel region at the image edge, summing the grayscale measurement terms of 2500 sample pixels within this region, and dividing by 2500 to obtain the average value. For example, the calculated preset background reference value is 240. The pixel array acquisition submodule subtracts the grayscale measurement term of each pixel node from the preset background baseline value of 240, and inputs the subtraction result into the absolute value conversion module to obtain a non-negative value, thus obtaining the difference term value. The pixel array acquisition submodule determines invalid data units corresponding to difference terms exceeding a preset noise tolerance threshold. The preset noise tolerance threshold is set to 15, which is obtained by comparing the grayscale range of reflective noise on the substrate under varying illumination through a dispersion experiment. When the difference term corresponding to a pixel node is less than or equal to the preset noise tolerance threshold of 15, the pixel array acquisition submodule classifies it as an invalid data unit belonging to the background color, discards the invalid data unit, and retains the associated grayscale measurement terms with a difference term greater than 15 to generate the target grayscale matrix.

[0028] The target coordinate filtering submodule extracts the grayscale measurement items carried by each data node in the target grayscale matrix, collects the threshold of the predetermined grayscale range, performs a comparison operation between the grayscale measurement items and the predetermined grayscale range threshold, filters the associated data nodes within the predetermined grayscale range threshold, extracts the row and column identifier sequences corresponding to the associated data nodes, merges the row and column identifier sequences corresponding to the associated data nodes, and obtains the set of target pixel coordinate points. The grayscale measurement items carried by each data node in the target grayscale matrix are extracted. A predetermined grayscale range threshold is collected, which is set between a lower limit of 30 and an upper limit of 120. This range corresponds to the set of solid-state reflective characteristic values ​​of the ink mark of this model under a specific light source. The target coordinate filtering submodule compares each retained grayscale measurement item with the predetermined grayscale range threshold to determine whether each retained associated grayscale measurement item simultaneously meets the logical condition of being greater than or equal to 30 and less than or equal to 120. The target coordinate filtering submodule filters associated data nodes within the predetermined grayscale range threshold, discarding data units that deviate from this range by setting them to zero, and extracts the row and column identifier sequences corresponding to the filtered associated data nodes. The target coordinate filtering submodule extracts the horizontal coordinate index of each qualified associated data node in its original 2D grayscale image pixel array as the column identifier and the vertical coordinate index as the row identifier, and merges and stores the corresponding row and column identifier sequences of the associated data nodes into a dynamic linked list structure to obtain the target pixel coordinate point set. For example, if there are 3 matching data nodes, whose corresponding coordinate positions are the 50th pixel horizontally and the 100th pixel vertically, the 52nd pixel horizontally and the 102nd pixel vertically, and the 48th pixel horizontally and the 98th pixel vertically, the target coordinate filtering submodule merges them into a summary set of these 3 sets of coordinate data.

[0029] The centroid coordinate calculation submodule extracts the horizontal and vertical identifier sequences within the target pixel coordinate point set, calculates the sum of the horizontal identifier sequences as the horizontal cumulative term, calculates the sum of the vertical identifier sequences as the vertical cumulative term, reads the total number of nodes as the divisor reference, performs division operations between the horizontal and vertical cumulative terms and the divisor reference respectively to extract the mean term, merges the mean terms into spatial coordinate components, and establishes the cross alignment center coordinate values. The horizontal and vertical coordinate values ​​within the dynamic linked list are allocated to two independent calculation channels. The centroid coordinate calculation submodule calculates the sum of the horizontal identifier sequences as the horizontal cumulative term. In the example above, the horizontal coordinates 50 and 52, along with 48, are input into the adder for summation, resulting in a horizontal cumulative term of 150. The centroid coordinate calculation submodule calculates the sum of the vertical identifier sequences as the vertical cumulative term. In the example above, the vertical coordinates 100 and 102, along with 98, are input into the adder for summation, resulting in a vertical cumulative term of 300. The centroid coordinate calculation submodule reads the total number of valid data nodes as the divisor benchmark. In this example, the total number of nodes is 3, so the divisor benchmark is set to 3. The centroid coordinate calculation submodule performs division operations on the horizontal and vertical cumulative terms with the divisor benchmark to extract the mean term. Dividing the horizontal cumulative term 150 by the divisor benchmark 3 yields a horizontal mean term of 50, and dividing the vertical cumulative term 300 by the divisor benchmark 3 yields a vertical mean term of 100. The centroid coordinate operation submodule merges the horizontal and vertical mean terms as spatial coordinate components, and establishes the cross alignment center coordinate value as the 50th column horizontally and the 100th row vertically.

[0030] Table 2: Parameters for Calculating Alignment Image Coordinates

[0031] As shown in Table 2, the numerical evolution of the horizontal and vertical identifier coordinates of each sample point sequence within the target pixel coordinate point set during the process of being sent into the accumulator register is illustrated.

[0032] Specifically, such as Figure 2 , 6 As shown, the displacement calculation module includes: The planar deviation deduction submodule collects the predetermined ideal reference coordinate values, substitutes the cross alignment center coordinate values ​​and the ideal reference coordinate values ​​into the difference node to perform a subtraction operation, calculates the one-dimensional displacement difference term, performs a sum of squares and square root operation on the one-dimensional displacement difference term to extract the Euclidean distance metric term, compares the Euclidean distance metric term with the preset allowable limit reference value and removes out-of-limit distortion data terms, and generates the planar position deviation amount. The module collects predetermined ideal reference coordinate values ​​through the built-in layout blueprint parameter storage area. These values ​​represent the theoretically correct, error-free positions where the registration marks should exist under the standard design file. For example, the horizontal coordinate of these predetermined ideal reference coordinate values ​​is set to column 55, and the vertical coordinate to row 100. The plane deviation calculation submodule substitutes the established crosshair alignment center coordinate values ​​with the ideal reference coordinate values ​​into the difference node and performs a subtraction operation. The specific calculation is performed in two steps: first, subtracting the horizontal coordinate of the ideal reference coordinate value from the horizontal coordinate of the crosshair alignment center coordinate value; second, subtracting the vertical coordinate of the ideal reference coordinate value from the vertical coordinate of the crosshair alignment center coordinate value. The plane deviation calculation submodule then subtracts 55 from the horizontal coordinate of the calculated crosshair alignment center coordinate value (50) to obtain -5, and subtracts 100 from the vertical coordinate (100) to obtain 0, generating the corresponding 1D displacement difference terms for the horizontal and vertical directions. The planar deviation estimation submodule performs a squaring operation on the 1D displacement difference term to extract the Euclidean distance metric. It multiplies the lateral displacement difference term (-5) by itself to obtain 25, and multiplies the longitudinal displacement difference term (0) by itself to obtain 0. The squares of these two values ​​are then added together in a summation unit to obtain a total of 25. Subsequently, the square root of the sum 25 is calculated using Newton's iteration method, yielding an Euclidean distance metric of 5 pixels. The planar deviation estimation submodule compares the Euclidean distance metric with a preset allowable limit benchmark value and removes out-of-limit distortion data. Through process control, the preset allowable limit benchmark value is set to 20 pixels based on the highest image overlap requirement of printing quality standards. Since the calculated Euclidean distance metric of 5 pixels is less than the preset allowable limit benchmark value of 20 pixels, the planar deviation estimation submodule classifies this data as a regular deviation and retains it, subsequently converting it into the actual physical length. Since the calibration parameters of the camera lens record that the actual physical distance corresponding to each pixel is 0.05 mm, the planar deviation inference submodule multiplies the Euclidean distance metric 5 by the pixel conversion ratio of 0.05 mm to generate a planar position deviation of 0.25 mm.

[0033] The error comparison submodule, for the planar position deviation, combines the absolute elongation deformation value to extract the directional component item carried by the planar position deviation and the scaling scale item contained in the absolute elongation deformation value. The directional component item and the scaling scale item are input into the fusion node to perform a product operation, calculate the spatial error combination metric, compare the spatial error combination metric with the preset extreme error benchmark, remove divergent outlier data, and obtain the alignment error scalar. The planar position deviation carrying the directional component and the scaling factor contained in the absolute elongation deformation value are extracted. Since the subtraction result of the aforementioned horizontal coordinates is negative, indicating that the actual position is biased to the left and in the opposite direction, this negative sign attribute is extracted by the error comparison submodule and defined as the directional component. The error comparison submodule inputs the directional component and the scaling factor into the fusion node for addition. This addition operation is interpreted by the system as the addition compensation logic of vectors in the same direction. Since thermal expansion deformation and physical position deviation have a coaxial superposition effect under this condition, the error comparison submodule sends the planar position deviation of 0.25 mm and the absolute elongation deformation value of 0.0924 mm into the floating-point adder for addition, obtaining the spatial error combination metric of 0.3424 mm. The error comparison submodule compares the spatial error combination metric with the preset extreme error benchmark, which is set to 1.5 mm according to the correction stroke limit of the mechanical guide rail. The error comparison submodule determines that the spatial error combination metric of 0.3424 mm does not exceed the limit of 1.5 mm, and then discards the divergent outlier data as normal working conditions, retains the fusion calculation results, and obtains the final alignment error scalar of 0.3424 mm.

[0034] Specifically, such as Figure 2 , 7 As shown, the pulse frequency control module includes: The error share splitting submodule obtains the single-step servo running displacement constant, performs a division calculation between the alignment error scalar and the single-step servo running displacement constant, extracts the quotient data item, extracts the integer component associated with the quotient data item as the share discrete splitting value, compares the share discrete splitting value with the preset pulse saturation benchmark threshold to remove out-of-bounds and excessive data, retains the normal splitting value that does not exceed the boundary, and obtains the total amount of compensation execution pulses. The single-step servo displacement constant is obtained by retrieving the driver device configuration registry. This constant represents the linear propulsion distance of the mechanical screw generated by the stepper servo motor for each square wave electrical pulse signal received. The specific value is read from the motor nameplate parameter file and set to 0.002 mm. The error share splitting submodule loads the previously determined alignment error scalar of 0.3424 mm and the single-step servo displacement constant of 0.002 mm into the division operator to perform division calculation and extract the quotient data item. The error share splitting submodule divides 0.3424 mm by 0.002 mm, and the quotient data item is calculated to be 171.2. The error share splitting submodule extracts the integer component associated with the quotient data item as the share discrete splitting value. By calling the underlying data truncation function, the quotient data item 171.2 is rounded down to discard the decimal part, forcibly erasing all values ​​after the decimal point, and the share discrete splitting value is obtained as 171. The advantage of this operational logic is that it ensures that the control cycle units received by the hardware execution elements are all indivisible complete instruction shares through data truncation, preventing motor jitter caused by pulse fragmentation. The error share splitting submodule compares the discrete splitting value of the share with the preset pulse saturation benchmark threshold to eliminate out-of-bounds excessive data. The preset pulse saturation benchmark threshold is determined by the upper limit of the input buffer width of the motor controller, and the upper limit of receiving a single instruction is explicitly set to 1000 pulses. Since the discrete splitting value of 171 is much lower than the preset pulse saturation benchmark threshold of 1000, the error share splitting submodule marks the data status as safe, retains the normal splitting value that has not exceeded the limit, and obtains a total compensation execution pulse of 171 independent pulse cycles.

[0035] The polarity direction determination submodule extracts the positive and negative algebraic sign terms carried by the total amount of compensation execution pulses, performs matching and verification between the positive and negative algebraic sign terms and the predetermined positive operation reference mark. When the positive and negative algebraic sign terms match the positive operation reference mark, the positive displacement state quantity is extracted. When the positive and negative algebraic sign terms deviate from the positive operation reference mark, the reverse displacement state quantity is extracted, and a polarity sign guidance parameter is established. Because the difference in the horizontal coordinate of the offset position is negative in the previous error derivation, the corresponding position compensation needs to be corrected towards the positive axis. Here, the polarity direction discrimination submodule reads the direction register association information and generates a positive or negative algebraic sign term. The polarity direction discrimination submodule matches the positive or negative algebraic sign term with a predetermined forward operation reference flag, which is defined as binary status word 1 in the system library. When the positive or negative algebraic sign term is positive and matches the forward operation reference flag 1, the polarity direction discrimination submodule extracts the forward displacement state quantity and sets the direction pin level attribute of the hardware control chip to high level. When the positive or negative algebraic sign term is negative and deviates from the forward operation reference flag, the polarity direction discrimination submodule extracts the reverse displacement state quantity and sets the direction pin level attribute of the hardware control chip to low level. The current comparison result is that the parameters match, and the polarity direction discrimination submodule establishes the polarity sign guiding parameter as a high-level constant.

[0036] The frequency reshaping submodule monitors the basic operating frequency value, reads the direction factor in the polarity sign guide parameter and assigns it to the corresponding phase angle bit of the basic operating frequency value, performs a comparison operation between the basic operating frequency value and the predetermined upper limit benchmark, cuts out the overclocking band data that exceeds the upper limit benchmark in the basic operating frequency value, assembles the remaining regular basic operating frequency value and direction factor and converts them into a control layer machine code sequence, and generates a dynamic registration servo correction instruction. The system clock divider monitors the base operating frequency. Initially, the base operating frequency is set to a reference transmission frequency of 200 pulses per second, based on the equipment's cold start configuration. The operating frequency reshaping submodule reads the direction factor from the polarity sign guide parameter and assigns it to the corresponding phase angle bit of the base operating frequency value. It then performs a bit-field concatenation operation between the high-level direction indicator data bit and the frequency value data bit. The operating frequency reshaping submodule compares the base operating frequency value with a predetermined upper frequency limit reference, which is set to 500 pulses per second based on the motor torque attenuation characteristics. Since the base operating frequency value of 200 is lower than the predetermined upper frequency limit reference of 500, the operating frequency reshaping submodule determines that it is in a safe output range and cuts off the overclocking band data within the base operating frequency value that exceeds the upper frequency limit reference. Under these normal parameter conditions, forced blocking and cutting are not required. The frequency reshaping submodule assembles the remaining conventional basic operating frequency values ​​and direction factors and converts them into control layer machine code sequences. It then converts the information set containing 200 pulses per second, a total of 171 pulses, and a high-level direction indicator into hexadecimal machine instructions according to the underlying communication protocol. Finally, it generates dynamic alignment servo correction instructions and sends them to the bottom servo actuator firmware terminal.

[0037] Table 3: Servo Execution Control Parameter Table

[0038] As shown in Table 3, the core hardware action parameters that the frequency reshaping submodule needs to set and send to the underlying servo motor controller after completing the calculation process are listed in detail.

[0039] Please see Figure 8 The intelligent production control method for tinplate printing is implemented based on the aforementioned intelligent production control system for tinplate printing, and includes the following steps: S1: Acquire the continuous voltage waveform sequence transmitted by the front-end device, identify the characteristics of discrete amplitude data, input the data into the support vector regression model to extract level change information, and convert it into surface temperature values; S2: Obtain the preset standard ambient temperature value, compare the surface temperature value with the standard ambient temperature value to extract the temperature difference offset, and perform scaling transformation on the temperature difference offset by integrating the inherent linear expansion reference value of the printed iron plate and the plate reference length constant to generate the absolute elongation deformation value. S3: Acquire a two-dimensional grayscale image pixel array of the registration mark edge region, scan the two-dimensional grayscale image pixel array to filter the set of target pixel coordinate points within the predetermined grayscale range, perform centroid extraction operation, and generate cross alignment center coordinate values. S4: Compare the cross alignment center coordinates with the pre-stored ideal reference alignment center coordinates to deduce the planar position deviation, and combine the absolute elongation deformation value with the planar position deviation to construct the alignment error scalar. S5: Obtain the single-step servo displacement constant, perform a fractional discretization operation on the alignment error scalar, obtain the total amount of compensation execution pulses, determine the polarity sign orientation, reshape the basic operating frequency value, and generate dynamic alignment servo correction commands.

[0040] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of protection of the described technical solutions.

Claims

1. A smart production control system for tinplate printing, characterized in that, The system includes: The surface temperature sensing module acquires the continuous voltage waveform sequence transmitted by the front-end device, identifies the characteristics of discrete amplitude data, inputs it into the support vector regression model to extract level change information, and converts it into surface temperature values. The heat deformation calculation module obtains the preset standard state ambient temperature value, compares the surface temperature value with the standard state ambient temperature value to extract the state temperature difference offset, and performs scaling transformation operation on the state temperature difference offset by integrating the inherent linear expansion reference value of the printed iron plate and the plate reference length constant to generate the absolute elongation deformation value. The alignment image parsing module acquires a two-dimensional grayscale image pixel array of the registration mark edge region, scans the two-dimensional grayscale image pixel array to filter the set of target pixel coordinate points within a predetermined grayscale range, performs centroid extraction operation, and generates cross alignment center coordinate values. The displacement calculation module compares the cross alignment center coordinates with the pre-stored ideal reference alignment center coordinates to deduce the planar position deviation, and combines the absolute elongation deformation value with the planar position deviation to construct an alignment error scalar. The pulse frequency control module obtains the single-step servo displacement constant, performs a fractional discretization operation on the alignment error scalar, obtains the total amount of compensation execution pulses, determines the polarity sign orientation, reshapes the basic operating frequency value, and generates a dynamic alignment servo correction command.

2. The intelligent production control system for tinplate printing according to claim 1, characterized in that, The surface temperature values ​​include peak heating point readings, steady-state average heating value, and regional thermal radiance. The absolute elongation deformation values ​​include longitudinal tensile length, transverse extension width, and micro buckling increment. The cross alignment center coordinate values ​​include horizontal axis positioning anchor point, vertical axis intersection pole, and origin spatial vector. The alignment error scalar values ​​include geometric offset amplitude, angular deflection margin, and overall misalignment spacing. The dynamic alignment servo correction command includes motor drive timing, duty cycle adjustment signal, and phase compensation period.

3. The intelligent production control system for tinplate printing according to claim 1, characterized in that, The surface temperature sensing module includes: The waveform feature perception submodule acquires the continuous voltage waveform sequence transmitted by the front-end device, extracts the voltage amplitude data items corresponding to each discrete time node in the continuous voltage waveform sequence according to the preset frequency reference, arranges the voltage amplitude data items into a one-dimensional vector matrix, calculates the amplitude difference between adjacent elements in the one-dimensional vector matrix, extracts the vector elements corresponding to those that cross the preset fluctuation boundary threshold, and generates discrete amplitude feature vectors. The level change operation submodule substitutes the discrete amplitude feature vector into the configuration mapping matrix to perform an inner product operation, calculates the distance measure value between the discrete amplitude feature vector and each support vector in the mapping matrix, extracts the abnormal component data corresponding to the distance measurement value that crosses the preset support boundary spacing benchmark, performs weighted summation calculation on the abnormal component data to extract the corresponding deviation index information item, and obtains the level change deviation rate. The temperature value construction submodule collects the basic temperature reference value at the reference environment node, multiplies the level change deviation rate by a preset temperature coefficient to convert it into a temperature compensation value, sums it with the basic temperature reference value, calculates the deviation compensation metric value corresponding to the basic temperature reference value, adds the deviation compensation metric value to the basic temperature reference value to obtain the merged summation state metric item, removes out-of-bounds abnormal numerical items, and obtains the surface temperature value.

4. The intelligent production control system for tinplate printing according to claim 3, characterized in that, The thermal deformation calculation module includes: The surface temperature difference calculation submodule collects the preset standard state ambient temperature value, calls the surface temperature value, extracts the difference item between the standard state ambient temperature value and the surface temperature value, inputs the difference item into the comparison operation node to perform the subtraction operation, calculates the distance measurement value where the standard state ambient temperature value deviates from the surface temperature value, performs an absolute value conversion operation on the distance measurement value, extracts the unsigned difference item, and establishes the state temperature difference offset. The scaling transformation submodule collects the inherent linear expansion reference value of the printed iron sheet, assigns the linear expansion reference value to the corresponding coefficient term position within the state temperature difference offset, performs a product operation on the state temperature difference offset and the linear expansion reference value to obtain the scaling transformation product term, determines whether the scaling transformation product term exceeds the preset elongation extreme value threshold and related out-of-bounds parts, removes the related out-of-bounds parts, and obtains the absolute elongation deformation value.

5. The intelligent production control system for tinplate printing according to claim 4, characterized in that, The alignment image parsing module includes: The pixel array acquisition submodule acquires a two-dimensional grayscale image pixel array in the registration mark edge region, reads the spatial position associated grayscale measurement terms within the two-dimensional grayscale image pixel array, extracts the difference terms between the grayscale measurement terms and the preset background reference value, determines invalid data units corresponding to the difference terms exceeding the preset noise tolerance threshold, removes invalid data units and retains the remaining associated grayscale measurement terms, and generates the target grayscale matrix. The target coordinate filtering submodule extracts the grayscale measurement items carried by each data node in the target grayscale matrix, collects a predetermined grayscale range threshold, performs a comparison operation between the grayscale measurement items and the predetermined grayscale range threshold, filters the associated data nodes within the predetermined grayscale range threshold, extracts the row and column identifier sequences corresponding to the associated data nodes, merges the row and column identifier sequences corresponding to the associated data nodes, and obtains the target pixel coordinate point set. The centroid coordinate calculation submodule extracts the horizontal and vertical identifier sequences within the target pixel coordinate point set, calculates the sum of the horizontal identifier sequences as the horizontal cumulative term, calculates the sum of the vertical identifier sequences as the vertical cumulative term, reads the total number of nodes as the divisor reference, performs division operations between the horizontal and vertical cumulative terms and the divisor reference respectively to extract the mean term, merges the mean terms into spatial coordinate components, and establishes the cross alignment center coordinate values.

6. The intelligent production control system for tinplate printing according to claim 5, characterized in that, Extract the upper and lower noise tolerance values ​​from the preset noise tolerance threshold. Perform a numerical comparison between the difference item and the upper and lower noise tolerance values. Extract the grayscale measurement items associated with the difference item being greater than the upper noise tolerance value as high-end noise measurement items. Extract the grayscale measurement items associated with the difference item being less than the lower noise tolerance value as low-end noise measurement items. Merge the high-end and low-end noise measurement items as invalid data units.

7. The intelligent production control system for tinplate printing according to claim 5, characterized in that, The displacement calculation module includes: The planar deviation deduction submodule collects the predetermined ideal reference coordinate values, substitutes the cross alignment center coordinate values ​​and the ideal reference coordinate values ​​into the difference node to perform a subtraction operation, calculates the one-dimensional displacement difference term, performs a sum of squares and square root operation on the one-dimensional displacement difference term to extract the Euclidean distance metric term, compares the Euclidean distance metric term with the preset allowable limit reference value and removes out-of-limit distortion data terms, and generates the planar position deviation amount. The error comparison submodule, for the planar position deviation, combines the absolute elongation deformation value to extract the directional component item carried by the planar position deviation and the scaling scale item contained in the absolute elongation deformation value. The directional component item and the scaling scale item are input into the fusion node to perform a product operation, calculate the spatial error combination metric, compare the spatial error combination metric with the preset extreme error benchmark, remove divergent outlier data, and obtain the alignment error scalar.

8. The intelligent production control system for tinplate printing according to claim 7, characterized in that, Extract the horizontal and vertical values ​​of the cross alignment center from the cross alignment center coordinate values. Extract the horizontal and vertical values ​​of the ideal reference coordinate values ​​from the ideal reference coordinate values. Subtract the horizontal values ​​of the cross alignment center from the horizontal values ​​of the ideal reference to obtain the horizontal one-dimensional displacement difference term. Subtract the vertical values ​​of the cross alignment center from the vertical values ​​of the ideal reference to obtain the vertical one-dimensional displacement difference term. Combine the horizontal one-dimensional displacement difference term and the vertical one-dimensional displacement difference term into a one-dimensional displacement difference term.

9. The intelligent production control system for tinplate printing according to claim 7, characterized in that, The pulse frequency control module includes: The error share splitting submodule obtains the single-step servo running displacement constant, performs a division calculation between the alignment error scalar and the single-step servo running displacement constant, extracts the quotient data item, extracts the integer component associated with the quotient data item as the share discrete splitting value, compares the share discrete splitting value with the preset pulse saturation benchmark threshold to remove out-of-bounds excess data, retains the non-out-of-bounds regular splitting value, and obtains the total amount of compensation execution pulses. The polarity direction determination submodule extracts the positive and negative algebraic sign terms carried by the total amount of compensation execution pulses, performs matching and verification between the positive and negative algebraic sign terms and the predetermined positive operation reference mark, extracts the positive displacement state quantity when the positive and negative algebraic sign terms match the positive operation reference mark, and extracts the reverse displacement state quantity when the positive and negative algebraic sign terms deviate from the positive operation reference mark, and establishes the polarity sign guidance parameter. The frequency reshaping submodule monitors the base operating frequency value, reads the direction factor in the polarity sign guide parameter and assigns it to the corresponding phase angle bit of the base operating frequency value, performs a comparison operation between the base operating frequency value and the predetermined upper limit benchmark, cuts out the overclocking band data in the base operating frequency value that exceeds the upper limit benchmark, assembles the remaining regular base operating frequency value and direction factor and converts them into a control layer machine code sequence, and generates a dynamic registration servo correction instruction.

10. A method for intelligent production control of tinplate printing, characterized in that, The intelligent production control system for tinplate printing is executed according to any one of claims 1-9. Includes the following steps: S1: Acquire the continuous voltage waveform sequence transmitted by the front-end device, identify the characteristics of discrete amplitude data, input the data into the support vector regression model to extract level change information, and convert it into surface temperature values; S2: Obtain the preset standard state ambient temperature value, compare the surface temperature value with the standard state ambient temperature value to extract the state temperature difference offset, and perform scaling transformation operation on the state temperature difference offset by integrating the inherent linear expansion reference amount of the printed iron plate and the plate reference length constant to generate the absolute elongation deformation value. S3: Acquire a two-dimensional grayscale image pixel array of the registration mark edge region, scan the two-dimensional grayscale image pixel array to filter the target pixel coordinate point set within the predetermined grayscale range, perform centroid extraction operation, and generate cross alignment center coordinate values; S4: Compare the cross alignment center coordinates with the pre-stored ideal reference alignment center coordinates to deduce the planar position deviation, and combine the absolute elongation deformation value with the planar position deviation to construct the alignment error scalar. S5: Obtain the single-step servo running displacement constant, perform a fractional discretization operation on the alignment error scalar, obtain the total amount of compensation execution pulses, determine the polarity sign orientation, reshape the basic running frequency value, and generate a dynamic alignment servo correction command.

Images