An in-line high speed sanitary napkin spray decoration system and method

Abstract

The present application relates to the technical field of sanitary product manufacturing, and discloses an online high-speed sanitary napkin spraying decoration system and method. The method comprises collecting a real-time surface image stream of a substrate through an upstream optical sensor array, containing multispectral reflection and microscopic topography data. A virtual reference plane is dynamically fitted based on the median value of the topography data statistics, to adapt to material movement and surface undulation. A spraying control grid is established on the plane, and the reflection data is mapped to each grid point. The theoretical saturation point of pigments is inversely solved according to the reflection characteristics of the grid points, and the initial spraying dose of each point is calculated accordingly. Multi-nozzle coordinated spraying instructions are generated in combination with the conveying speed. The method can adapt to changes in the substrate state in real time, realize accurate positioning of patterns and on-demand supply of doses on a high-speed production line, thereby improving the uniformity and clarity of the decoration layer and saving raw materials.

Description

Technical Field

[0001] This invention relates to the field of sanitary product manufacturing technology, specifically to an online high-speed sanitary napkin spraying decoration system and method. Background Technology

[0002] In the production of disposable hygiene products such as sanitary napkins, patterns or colors are often sprayed onto the surface to enhance the product's aesthetics and recognizability. Existing high-speed production lines typically employ fixed spraying systems with pre-programmed procedures. These systems are programmed based on the assumption of an ideally flat substrate, and their spraying path and dosage parameters are statically preset. However, in actual production, substrates such as nonwoven fabrics experience unavoidable vibrations, stretching, and inherent micro-texture undulations during high-speed transport, causing the actual distance and relative angle between the nozzle and the substrate surface to constantly change. Static spraying standards cannot adapt to these dynamic changes, resulting in blurred patterns, indistinct edges, or uneven thickness of the decorative layer.

[0003] On the other hand, existing technologies mostly employ fixed coating dosages or on / off control based on simple color sensors. The substrate's optical and physical properties, such as whiteness, porosity, and surface wettability, exhibit batch-to-batch and batch-to-batch variations, which fixed dosages cannot accommodate. Areas with high reflectivity may appear pale due to insufficient pigment coverage; areas with high absorbency may experience leakage or clumping due to excessive pigment buildup. This not only affects the consistency of the decorative effect but may also lead to material waste due to over-coating or migration problems during subsequent use due to poor adhesion. Manufacturers need a method that can sense the substrate's condition in real time and intelligently adjust the coating strategy to overcome the challenges of precise decoration on dynamic production lines. Summary of the Invention

[0004] The purpose of this invention is to provide an online high-speed sanitary napkin spraying decoration system and method to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides an online high-speed sanitary napkin spraying decoration method, the method comprising:

[0006] An optical sensor array positioned upstream of the conveyor line is used to acquire a real-time surface image stream of the sanitary napkin substrate, which includes multispectral reflectance data and surface micromorphology data.

[0007] A dynamic reference plane fitting operation is performed on the real-time surface image stream to generate a virtual reference plane of the sanitary napkin substrate in the current motion state. The virtual reference plane is generated based on the statistical median of the surface micromorphology data.

[0008] Multiple spray control grids are defined on the virtual reference plane, and the multispectral reflectance data is mapped to each grid point of the spray control grid to generate a gridded substrate surface model with reflectance features.

[0009] Based on the reflection characteristics of each grid point in the gridded substrate surface model, the theoretical saturation point of pigment adhesion is solved in reverse, and the initial spraying dose corresponding to each grid point is calculated based on the theoretical saturation point.

[0010] The sequence of initial spray dosage is input into the spray timing planner, and combined with the substrate delivery speed, a spray phase and dosage instruction table for multi-nozzle collaborative operation is generated.

[0011] Preferably, the step of performing a dynamic reference plane fitting operation on the real-time surface image stream to generate a virtual reference plane for the sanitary napkin substrate in the current motion state includes:

[0012] From the real-time surface image stream, extract the surface elevation dataset of the sanitary napkin substrate at multiple consecutive sampling times;

[0013] The surface elevation dataset is subjected to outlier filtering and smoothing to obtain a clean elevation dataset. The outlier filtering is based on the statistical outlier degree of the elevation value in the neighborhood.

[0014] Based on the clean elevation dataset, the least squares plane fitting rule is used to calculate the fitting plane equation describing the overall tilting and bending trend of the substrate.

[0015] Subtract the theoretical height values ​​calculated by the fitted plane equation at each point from the elevation values ​​of each point in the clean elevation dataset to obtain the set of relative height residuals;

[0016] Using the median absolute value of the relative height residual set as the reference offset, the fitted plane equation is translated and corrected to generate the final virtual reference plane.

[0017] Preferably, the step of defining multiple spray control grids on the virtual reference plane, mapping the multispectral reflectance data to each grid point of the spray control grids, and generating a gridded substrate surface model with reflectance features includes:

[0018] A Cartesian coordinate system is established with the geometric center of the virtual reference plane as the origin;

[0019] In the Cartesian coordinate system, a uniform grid array is divided along the length and width of the substrate according to a predetermined physical size. Each cell of the grid array is a spray control grid.

[0020] The multispectral reflectance data acquired at each sampling time is associated with one or more grid points in the spraying control grid through coordinate transformation based on its corresponding actual spatial coordinates.

[0021] For each of the spray control grids, all multispectral reflectance data associated with its lifetime are aggregated, and the average reflectance and reflectance variance of the grid points in multiple spectral channels are calculated.

[0022] The average reflectivity and reflectivity variance are used as the reflection characteristics of the grid points and assigned to the gridded substrate surface model to complete the model's attribute construction.

[0023] Preferably, the step of inversely solving for the theoretical saturation point of pigment adhesion based on the reflection characteristics of each grid point in the gridded substrate surface model, and calculating the initial spraying dosage corresponding to each grid point based on the theoretical saturation point, includes:

[0024] An empirical mapping relationship library between reflection characteristics and surface porosity is established. Based on the average reflectance of the grid points, the estimated porosity of the grid point region is obtained by querying the empirical mapping relationship library.

[0025] Based on the estimated porosity and pigment characteristic parameters, the micro-volume that pigment can fill per unit area is calculated, and the micro-volume is defined as the theoretical pigment capacity of the grid point.

[0026] The reflectance variance is introduced as a surface uniformity influencing factor to correct the theoretical pigment capacity for uniformity, thus obtaining the corrected theoretical pigment capacity.

[0027] Set the visual saturation level of the target decorative layer, and convert the visual saturation level into the desired pigment coverage thickness;

[0028] By combining the corrected theoretical pigment capacity with the required pigment coverage thickness, the maximum pigment dosage that does not exceed the capacity under the premise of satisfying the coverage thickness is calculated through iterative approximation calculation. The maximum pigment dosage is the initial spraying dosage.

[0029] Preferably, the step of inputting the sequence of the initial spray dosage into the spray timing planner, and generating a spray phase and dosage instruction table for multi-nozzle collaborative operation in combination with the substrate delivery speed, includes:

[0030] Obtain the constant conveying speed of the production line, as well as the fixed spatial distribution of multiple spray nozzles in the conveying direction;

[0031] The zero point of time is defined as the projection position of the front end of the gridded substrate surface model in the conveying direction reaching the first spray nozzle;

[0032] Based on the conveying speed, calculate the precise time points at which the center point of each of the spraying control grids passes under each of the spraying nozzles in sequence;

[0033] Based on the sequence of initial spraying doses, each spraying control grid is assigned a dose ratio that should be sprayed when passing through different nozzles, the dose ratio being determined based on the atomization characteristics and overlapping areas of each nozzle;

[0034] By combining the precise time point of each grid with the corresponding dose ratio, a time-sorted instruction list is generated for each spray nozzle. The instruction list contains instructions to apply a specific dose at a specific time. The instruction lists of all nozzles constitute the spray phase and dose instruction table.

[0035] Preferably, the step of calculating the fitting plane equation describing the overall tilt and bending trend of the substrate based on the clean elevation dataset and using the least squares plane fitting rule includes:

[0036] Read the three-dimensional spatial coordinates of all sampling points from the clean elevation dataset;

[0037] The fitted plane equation is preset to a standard plane equation form containing three undetermined plane coefficients;

[0038] Construct an objective function that is the sum of squares of the differences between the actual height values ​​of all sampling points and the theoretical height values ​​calculated from the fitted plane equation;

[0039] By taking the partial derivatives of the objective function with respect to the three undetermined plane coefficients, and setting each partial derivative to zero, a system of linear equations with respect to the three undetermined plane coefficients is obtained.

[0040] Solving the system of linear equations yields the specific values ​​of the three undetermined plane coefficients, thereby determining the equation of the fitted plane.

[0041] Preferably, the step of establishing an empirical mapping relationship library between reflection characteristics and surface porosity, and querying the estimated porosity of the grid point region from the empirical mapping relationship library based on the average reflectance of the grid points, includes:

[0042] Pre-collect standard substrate samples with various known porosities;

[0043] Under standard illumination conditions, multispectral reflectance data of each standard substrate sample were acquired, and its average reflectance in each spectral channel was calculated.

[0044] For each of the aforementioned standard substrate samples, its known porosity is correlated with the calculated average reflectance to form a mapping relationship pair;

[0045] Using the mapping pairs of all standard substrate samples, a functional mapping relationship between average reflectance and porosity is obtained through nonlinear regression analysis, and the functional mapping relationship is stored as the empirical mapping relationship library;

[0046] During the query, the average reflectance of any grid point in the gridded substrate surface model is input into the function mapping relationship for calculation, and the estimated porosity of the grid point region is output.

[0047] Preferably, by combining the corrected theoretical pigment capacity with the required pigment coverage thickness, and through iterative approximation calculation, the maximum pigment dosage that does not exceed the capacity while satisfying the coverage thickness is determined. This maximum pigment dosage is the initial spraying dosage, including:

[0048] Set an initial iterative dose value;

[0049] Based on the initial iterative dose value, and according to the physical model of pigment coverage on the surface, the predicted coverage thickness formed after pigment deposition at the current dose is calculated.

[0050] The predicted coverage thickness is compared with the required pigment coverage thickness, and it is determined whether the total volume of pigment under the predicted coverage thickness exceeds the corrected theoretical pigment capacity.

[0051] If the predicted coverage thickness is less than the required pigment coverage thickness, and the total volume of the pigment does not exceed the corrected theoretical pigment capacity, then the iterative dose value is increased by a preset step size.

[0052] If the predicted coverage thickness is greater than or equal to the required pigment coverage thickness, or if the total volume of the pigment has reached the corrected theoretical pigment capacity, then the iteration terminates.

[0053] The iterative dose value obtained in the last iteration that simultaneously satisfies the thickness and capacity constraints is set as the maximum pigment dose, i.e., the initial spraying dose.

[0054] Preferably, the step of taking the projection position of the front end of the meshed substrate surface model reaching the first spray nozzle in the conveying direction as the zero point of time includes:

[0055] Obtain the geometric shape data of the surface model of the gridded substrate and determine its front boundary coordinates in the conveying direction;

[0056] Obtain the precise spatial position coordinates of the first spray nozzle in the conveyor line coordinate system;

[0057] Project the coordinates of the front-end boundary along a plane perpendicular to the conveyor line onto the vertical projection line where the first spray nozzle is located;

[0058] Calculate the straight-line distance required to move in the conveying direction from the front boundary coordinates to coincide with the vertical projection line of the first spray nozzle;

[0059] Calculate the time required to travel the straight-line distance based on the conveying speed;

[0060] The moment when the meshed substrate surface model begins to move and the calculated time has elapsed is set as the zero point of time.

[0061] Preferably, when the processor executes the computer program, it implements the steps of the online high-speed sanitary napkin spraying decoration method as described in any of the above-mentioned methods.

[0062] Compared with the prior art, the beneficial effects of the present invention are:

[0063] Based on dynamic reference plane fitting, the statistical median of real-time acquired surface micro-morphology data is used as the basis for generating a virtual reference plane. This enables the spraying control system to respond in real time to the positional fluctuations and micro-undulations of the substrate during high-speed transport, establishing a dynamic and stable spatial coordinate reference. All subsequent path planning and positioning are based on this dynamically updated plane, rather than a fixed mechanical coordinate system. The nozzle aiming and trajectory maintain a precise spatial correspondence with the actual three-dimensional state of the material surface, thereby overcoming pattern misalignment and blurring caused by substrate vibration or deformation, and achieving high-precision reproduction of the geometric contours of decorative patterns under high-speed conditions.

[0064] Based on the multispectral reflectance characteristics of each grid point in the meshed model, the theoretical saturation point of pigment adhesion in local areas is calculated in reverse, and the initial spraying dosage is calculated accordingly. A physical correlation model between the optical properties of the substrate surface and the required amount of pigment is established. The system identifies the microscopic differences in the light absorption characteristics of the surface through reflectance features and automatically allocates appropriate amounts of pigment to high-reflectance and low-reflectance areas. This realizes the transformation from fixed-dosage spraying to precise quantitative spraying based on real-time surface characteristics. A visually uniform and fully opaque decorative layer can be formed in different areas of the substrate, while avoiding material waste and penetration agglomeration caused by local overspraying, as well as problems such as exposed substrate or uneven color caused by insufficient spraying. Attached Figure Description

[0065] Figure 1 This is a schematic diagram illustrating the working principle of the online high-speed sanitary napkin spraying decoration method described in this invention.

[0066] Figure 2 A flowchart for dynamic reference surface fitting operation;

[0067] Figure 3 Flowchart for constructing a meshed substrate surface model;

[0068] Figure 4 A correlation analysis diagram showing the relationship between the multispectral reflectance and the estimated porosity of sanitary napkin substrate;

[0069] Figure 5 This is a nonlinear correlation analysis diagram between the average reflectance of the substrate and the estimated porosity. Detailed Implementation

[0070] 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.

[0071] Please see Figure 1 This invention provides an online high-speed sanitary napkin spraying decoration method. The method includes: arranging an optical sensor array upstream of a conveyor line to capture real-time surface image streams of a sanitary napkin substrate flowing below it. These image streams simultaneously contain multispectral reflectance data and surface microstructure data of the substrate surface. Subsequently, a dynamic reference plane fitting operation is performed on the acquired real-time surface image stream. This operation generates a virtual reference plane adapted to the currently moving substrate based on the statistical median of the surface microstructure data. This plane provides a reference for subsequent precise spatial positioning. Next, multiple spraying control grids are systematically divided on the virtual reference plane, and the multispectral reflectance data is mapped to each grid point according to its spatial coordinates, thereby constructing a gridded substrate surface model with reflectance characteristics. Then, based on the reflectance characteristics of each grid point in the model, the theoretical saturation point of pigment adhesion in that area is deduced, and the initial spraying dosage required for each grid point is calculated based on this saturation point. Finally, this series of initial spray dosages and substrate delivery speeds are input into the spray timing planner, which coordinates the working phases and dosages of multiple spray nozzles to generate an executable spray instruction table.

[0072] Example 1: See Figure 2From a real-time surface image stream, surface elevation datasets of sanitary napkin substrates at multiple consecutive sampling times are extracted. Outlier filtering and smoothing are performed on the surface elevation dataset to obtain a clean elevation dataset, where outlier filtering is based on the statistical outlier degree of elevation values ​​within their neighborhood. The least squares plane fitting rule is used to calculate the fitted plane equation describing the overall tilt and bending trend of the substrate. The theoretical height values ​​calculated by the fitted plane equation at each point are subtracted from the elevation values ​​of each point in the clean elevation dataset to obtain a set of relative height residuals. Using the median absolute value of the relative height residual set as the baseline offset, the fitted plane equation is translated and corrected to generate the final virtual reference plane.

[0073] The process of calculating the fitted plane equation using the least squares plane fitting rule based on the cleanroom elevation dataset involves: reading the three-dimensional spatial coordinates of all sampling points from the cleanroom elevation dataset; pre-setting the fitted plane equation as a standard plane equation containing three undetermined plane coefficients; constructing an objective function with the sum of squares of the differences between the actual height values ​​of all sampling points and the theoretical height values ​​calculated from the fitted plane equation; obtaining a system of linear equations with respect to the three undetermined plane coefficients by taking the partial derivatives of the objective function with respect to each of the three undetermined plane coefficients and setting each partial derivative to zero; and solving this system of linear equations to obtain the specific values ​​of the three undetermined plane coefficients, thereby determining the fitted plane equation.

[0074] In practice, surface elevation datasets of sanitary napkin substrates at multiple consecutive sampling times are extracted from real-time surface image streams acquired by an optical sensor array located upstream of the conveyor line. The surface elevation dataset undergoes outlier filtering and smoothing based on the statistical outlier degree of elevation values ​​within their neighborhoods, resulting in a clean elevation dataset. Outlier filtering identifies and removes data points in the surface elevation dataset that significantly deviate from the statistical characteristics of local neighborhood elevation values. Smoothing further reduces random fluctuation noise in the data. The clean elevation dataset provides a stable data foundation for subsequent plane fitting. Based on the clean elevation dataset, the least squares plane fitting rule is used to calculate the fitting plane equation describing the overall tilt and bending trend of the substrate. The theoretical height value calculated by the fitting plane equation at each point is subtracted from the elevation value of each point in the clean elevation dataset to obtain a set of relative height residuals. The median absolute value of the relative height residual set is used as the baseline offset to translate and correct the fitting plane equation, ultimately generating a virtual reference plane for subsequent steps.

[0075] In some embodiments, the process of calculating the fitted plane equation using the least squares plane fitting rule based on a cleanroom elevation dataset is specifically unfolded as follows: The three-dimensional spatial coordinates of all sampling points are read from the cleanroom elevation dataset, and the fitted plane equation is preset to a standard form containing three undetermined plane coefficients. An objective function is constructed using the sum of squares of the differences between the actual height values ​​of all sampling points and the theoretical height values ​​calculated from the fitted plane equation. By taking the partial derivatives of the objective function with respect to the three undetermined plane coefficients and setting each partial derivative to zero, a system of linear equations with respect to the three undetermined plane coefficients is obtained. Solving the system of linear equations yields the specific values ​​of the three undetermined plane coefficients, thereby determining the fitted plane equation. The objective function used in the least squares fitting process can be expressed as:

[0076]

[0077] Where: symbol Represents the objective function value, symbol Represents the total number of sampling points in the clean elevation dataset, symbol Indicates the first The actual measured height value of each sampling point, symbol and They represent the first The x and y coordinates of each sampling point in a Cartesian coordinate system, with symbols... , and These are the coefficients of the fitted plane equation to be determined.

[0078] Optionally, outlier filtering is based on the statistical outlier degree of elevation values ​​within their neighborhood. This process is accomplished by calculating the elevation value distribution characteristics of each data point and its spatial neighborhood, such as calculating the local elevation mean and standard deviation of each point. Points deviating from the mean by more than a certain multiple of the standard deviation are identified as outliers and removed. It is understood that smoothing can be performed using moving averages or Gaussian filtering on the data sequence before cleaning to suppress high-frequency noise without significantly altering the macroscopic morphology of the substrate surface, thereby ensuring the accuracy of subsequent fitting. In some embodiments, the translation correction operation for generating the virtual reference plane involves calculating the absolute values ​​of all residual values ​​in the relative height residual set, finding the median of these absolute values ​​as the reference offset, and then translating the original fitted plane equation along the normal direction by the distance of the reference offset.

[0079] Example 2: See Figure 3A Cartesian coordinate system is established with the geometric center of the virtual reference plane as the origin. Within this system, a uniform grid array is created along the length and width of the substrate according to predetermined physical dimensions. Each cell of this grid array represents a spray control grid. Multispectral reflectance data acquired at each sampling moment is mapped to one or more grid points within the spray control grid based on their corresponding spatial coordinates through coordinate transformation. For each spray control grid, all associated multispectral reflectance data within its lifetime are aggregated, and the average reflectance and reflectance variance of each grid point across multiple spectral channels are calculated. The average reflectance and reflectance variance are then used as the reflectance characteristics of the grid points to assign to the meshed substrate surface model, completing the model's attribute construction.

[0080] In practice, the process of generating a meshed substrate surface model with reflective features begins with an established virtual reference plane. A Cartesian coordinate system is established with the geometric center of the virtual reference plane as the origin. Within this Cartesian coordinate system, a uniform mesh array is divided along the length and width of the sanitary napkin substrate according to predetermined physical dimensions. Each cell of the mesh array is defined as a spraying control mesh. The predetermined physical dimensions are determined based on the spray coverage accuracy of the spray nozzle and the resolution requirements of the decorative pattern. The partitioning operation discretizes the continuous substrate surface into a series of regularly arranged geometric units, providing a spatial framework for subsequent local attribute analysis and dosage control. The multispectral reflectance data acquired by the optical sensor array at each sampling moment is associated with one or more grid points of the spraying control mesh based on the actual spatial coordinates corresponding to the multispectral reflectance data through coordinate transformation. The coordinate transformation involves rotation and translation calculations from the sensor coordinate system to the virtual reference plane coordinate system, ensuring that each reflectance data point can be accurately assigned to its own or nearest neighbor spraying control mesh.

[0081] In some embodiments, for each spray control grid, it is necessary to aggregate all multispectral reflectance data associated with the spray control grid within its lifetime. The lifetime refers to the time window from when the region corresponding to the spray control grid on the substrate enters the sensing region to when it leaves the sensing region. After aggregation, the average reflectance and reflectance variance of the grid points of the spray control grid in multiple spectral channels are calculated. The calculated average reflectance and reflectance variance are then used as the reflectance characteristics of the grid points of the spray control grid and assigned to the gridded substrate surface model, thereby completing the attribute construction of the gridded substrate surface model. The calculation of the average reflectance can be expressed as:

[0082]

[0083] Where: symbol This indicates the spray control grid in the spectral channel. Average reflectance, symbol This represents the total number of multispectral reflectance data points associated with the spray control grid during its lifecycle, denoted by [symbol]. Indicates the first Data points in the spectral channel The reflectance value obtained from the measurement.

[0084] Optionally, when associating multispectral reflectance data with grid points of the spray control grid, nearest neighbor matching or bilinear interpolation algorithms can be used. Nearest neighbor matching directly assigns data points to the spray control grid to which the center of the nearest grid point belongs, while bilinear interpolation assigns the reflectance value of a single data point to the corresponding spray control grid according to its distance weights from its four surrounding grid points. In some embodiments, the construction of the gridded substrate surface model is dynamic and continuous. As the substrate moves, new sensor data is continuously collected and mapped to the spray control grid at the front of the model. The spray control grid that has completed attribute calculations then enters a ready state to await spray planning.

[0085] Example 3: An empirical mapping database of reflection characteristics and surface porosity is established. The estimated porosity of the grid region is obtained from the database based on the average reflectance of the grid points. The microscopic volume that pigment can fill per unit area is calculated based on the estimated porosity and pigment characteristic parameters. This microscopic volume is defined as the theoretical pigment capacity of the grid points. Reflectance variance is introduced as a surface uniformity influencing factor to correct the theoretical pigment capacity, resulting in the corrected theoretical pigment capacity. The visual saturation level of the target decorative layer is set and converted into the required pigment coverage thickness. Combining the corrected theoretical pigment capacity and the required pigment coverage thickness, the maximum pigment dosage that does not exceed the capacity while meeting the coverage thickness requirement is calculated through iterative approximation. This maximum pigment dosage is the initial spraying dosage.

[0086] In practical implementation, the theoretical saturation point of pigment adhesion and the initial spraying dosage are calculated by reverse-engineering the reflection characteristics of each grid point in the gridded substrate surface model. An empirical mapping relationship library between reflection characteristics and surface porosity is established. The estimated porosity of the grid point region is obtained from the empirical mapping relationship library based on the average reflectance of the grid points. The microscopic volume that pigment can fill per unit area is calculated based on the estimated porosity and pigment characteristic parameters. This microscopic volume is defined as the theoretical pigment capacity of the grid point. Pigment characteristic parameters include, but are not limited to, pigment density, solid content, and its filling factor within the substrate pores. The reflectance variance of the grid points is introduced as a surface uniformity influencing factor to correct the theoretical pigment capacity for uniformity. A large reflectance variance indicates non-uniform surface optical properties, corresponding to non-uniform microstructure or material distribution. The theoretical pigment capacity needs to be lowered to avoid the risk of overfilling in local areas, thus obtaining the corrected theoretical pigment capacity.

[0087] In some embodiments, a visual saturation level of the target decorative layer is set. This visual saturation level is a predefined quantitative indicator corresponding to the product's decorative requirements. The visual saturation level is then converted into the required pigment coverage thickness. This conversion is performed based on the optical properties of the pigment and a calibration curve of coating thickness versus visual saturation. Combining the corrected theoretical pigment capacity with the required pigment coverage thickness, an iterative approximation calculation is used to determine the maximum pigment dosage that does not exceed the capacity while still meeting the coverage thickness requirement. The calculated maximum pigment dosage is then determined as the initial spraying dosage corresponding to that grid point. The calculation for uniformity correction of the theoretical pigment capacity can be expressed as follows:

[0088]

[0089] Where: symbol Indicates the revised theoretical capacity of the pigment, symbol Indicates the uncorrected theoretical pigment capacity, symbol This represents a correction factor related to the substrate material and pigment properties, with the symbol... This represents the variance of the reflectance of this grid point in the gridded substrate surface model.

[0090] Optionally, when querying the estimated porosity from the empirical mapping relation library, the input is a reflection feature vector composed of the average reflectance of grid points under multiple spectral channels. The query operation can be based on forward calculation using a trained nonlinear regression model, or it can be based on multidimensional interpolation using a pre-stored reflectance-porosity lookup table. In establishing the empirical mapping relation library between reflectance features and surface porosity, the specific implementation of the nonlinear regression model involves using pre-collected standard substrate samples with various known porosities. Under standard illumination conditions, multispectral reflectance data for each sample is acquired, and the average reflectance under each spectral channel is calculated. After associating the known porosity with the average reflectance data to form a mapping relation pair, the mapping relation pair is trained using a nonlinear regression analysis algorithm (such as multinomial regression, support vector regression, or neural network) to learn the complex nonlinear function mapping relationship between average reflectance and porosity. After training, the obtained function mapping relationship is stored as the empirical mapping relation library. In the query operation, the average reflectance of grid points in the gridded substrate surface model is input into this trained function mapping relationship for forward calculation, and the corresponding estimated porosity value is directly output. In some embodiments, the process of calculating the initial spraying dose by iterative approximation involves setting the starting point of the iteration to a base dose value estimated based on the target coverage thickness, calculating the distribution state of the pigment on the substrate surface under the current assumed dose in each iteration, and determining whether the coverage thickness requirement and capacity constraint condition are met simultaneously.

[0091] See Figure 4This is a correlation analysis chart showing the relationship between the multispectral reflectance and estimated porosity of a sanitary napkin substrate. Reflectance and porosity are positively correlated; higher reflectance corresponds to a higher estimated porosity. This is because higher substrate porosity indicates a more porous surface microstructure, resulting in stronger light reflection. The high degree of fit between the measured data points and the fitted curve demonstrates that the polynomial regression model effectively characterizes the nonlinear relationship between the two, serving as an empirical mapping basis for "reflectance characteristics → porosity." This type of chart is a core parameter correlation tool for the sanitary napkin spray coating decoration process. It is used to construct an empirical mapping library of "reflectance-porosity," quickly estimating substrate porosity using real-time collected reflectance data. Porosity is a key fundamental parameter for calculating pigment capacity and spray dosage; this correlation model directly affects the accuracy of the spray dosage, thereby ensuring the visual saturation and uniformity of the decorative layer.

[0092] Example 4: Obtaining a constant conveyor speed and a fixed spatial distribution of multiple spray nozzles along the conveyor direction. The zero point is defined as the projection position of the front end of the gridded substrate surface model reaching the first spray nozzle along the conveyor direction. Based on the conveyor speed, the precise time points at which the center point of each spray control grid passes under each spray nozzle are calculated. According to the initial spray dosage sequence, each spray control grid is assigned a dosage ratio to be sprayed when passing different nozzles. The dosage ratio is determined based on the atomization characteristics and overlap area of ​​each nozzle. The precise time points of each grid are combined with the corresponding dosage ratios to generate a time-ordered instruction list for each spray nozzle. The instruction list contains instructions to apply a specific dosage at a specific time. All nozzle instruction lists constitute a spray phase and dosage instruction table.

[0093] The method for determining the zero point of time, specifically the projection position of the front end of the meshed substrate surface model reaching the first spray nozzle in the conveying direction, involves acquiring the geometric shape data of the meshed substrate surface model to determine its front boundary coordinates in the conveying direction, and obtaining the precise spatial position coordinates of the first spray nozzle in the conveying line coordinate system. The front boundary coordinates are then projected along a plane perpendicular to the conveying line onto the vertical projection line where the first spray nozzle is located. The straight-line distance required to move from the front boundary coordinates to coincide with the vertical projection line of the first spray nozzle in the conveying direction is calculated. The time required to move this straight-line distance is calculated based on the conveying speed, and the moment after the meshed substrate surface model begins to move and this calculated time has elapsed is set as the zero point of time.

[0094] In practice, the initial spray dosage sequence is input into the spraying timing planner and combined with the substrate conveying speed to generate a spraying phase and dosage instruction table for multi-nozzle collaborative operation. This obtains the constant conveying speed of the production line and the fixed spatial distribution of multiple spraying nozzles in the conveying direction. Taking the projection position of the first spraying nozzle at the front end of the gridded substrate surface model in the conveying direction as the zero point of time, the precise time points at which the center point of each spraying control grid passes under each spraying nozzle are calculated based on the conveying speed. According to the initial spray dosage sequence, each spraying control grid is assigned the dosage ratio to be sprayed when passing different nozzles. The dosage ratio is determined based on the atomization characteristics and overlap area of ​​each nozzle. Atomization characteristics include the atomization cone angle and particle size distribution, and the overlap area is defined by the intersection of the spray patterns of adjacent nozzles on the substrate surface. The precise time points of each spraying control grid are combined with the corresponding dosage ratio to generate a time-ordered instruction list for each spraying nozzle. The instruction list contains instructions to apply a specific dosage at a specific time. The instruction lists of all nozzles constitute a complete spraying phase and dosage instruction table.

[0095] In some embodiments, the method for determining the zero point of time as the projection position of the front end of the meshed substrate surface model reaching the first spray nozzle in the conveying direction is as follows: Obtain the geometric shape data of the meshed substrate surface model and determine its front end boundary coordinates in the conveying direction; obtain the precise spatial position coordinates of the first spray nozzle in the conveying line coordinate system. Project the front end boundary coordinates along a plane perpendicular to the conveying line onto the vertical projection line where the first spray nozzle is located; calculate the straight-line distance required to move from the front end boundary coordinates to coincide with the vertical projection line of the first spray nozzle in the conveying direction. Calculate the time required to move this straight-line distance based on the conveying speed; set the moment after the meshed substrate surface model starts moving and the calculated time has elapsed as the zero point of time. Calculating the precise time point at which the center point of each spraying control grid passes below the nozzle depends on the conveying speed and the relative position relationship; the calculation relationship can be expressed as:

[0096]

[0097] Where: symbol Indicates the first The center point of the spray control grid passes through the first... The precise time point when directly below the spray nozzle, symbol Indicates a predetermined zero point in time, symbol This represents the position defined from the zero point of time to the [missing information]. The center point of the grid reaches the first The straight-line distance that needs to be moved directly below each nozzle, denoted by [symbol]. This indicates the constant conveying speed of the production line.

[0098] Optionally, when assigning different dosage ratios to each spray control grid through different nozzles, the allocation strategy must consider the uniformity of the spray pattern. For grids located in areas where multiple nozzles overlap, the total dosage is shared by the multiple nozzles according to a preset ratio. It can be understood that the preset dosage ratio is based on calibration data of the spatial distribution of the nozzle atomization field, ensuring that the pigment deposition obtained across the entire substrate surface is spatially continuous and uniform. Referring to Table 1, a simplified example is shown, which includes the dosage ratio allocation of three spray nozzles to five consecutively arranged spray control grids.

[0099] Table 1: Dosage Ratio Allocation Table for Spraying Control Grid

[0100] Spraying control grid number Dosage ratio at nozzle N1 Dosage ratio when N2 passes through nozzle Dosage ratio when passing through nozzle N3 total G1 1.0 0.0 0.0 1.0 G2 0.7 0.3 0.0 1.0 G3 0.0 1.0 0.0 1.0 G4 0.0 0.3 0.7 1.0 G5 0.0 0.0 1.0 1.0

[0101] In some embodiments, when generating a time-ordered instruction list, each instruction in the list contains at least two fields: a timestamp and a dose value. The timestamp is calculated using a formula, and the dose value is derived by multiplying the initial spray dose of the grid by the dose ratio when passing through the corresponding nozzle. It can be understood that the final output of the spraying phase and dose instruction table from the spraying timing planner is essentially a precise sequence of time and action. This sequence is synchronously sent to the independent controllers of each nozzle, driving multiple nozzles to release a predetermined dose of pigment at strictly agreed-upon times, thereby achieving the spraying of decorative patterns that precisely correspond to the surface model of the gridded substrate on a high-speed moving substrate.

[0102] Example 5: The process of establishing an empirical mapping relationship library between reflection characteristics and surface porosity involves pre-collecting various standard substrate samples with known porosity. Under standard illumination conditions, multispectral reflectance data for each standard substrate sample is acquired, and its average reflectance in each spectral channel is calculated. For each standard substrate sample, its known porosity and the calculated average reflectance are correlated to form a mapping relationship pair. Using all the mapping relationship pairs of the standard substrate samples, a functional mapping relationship between average reflectance and porosity is trained through nonlinear regression analysis, and this functional mapping relationship is stored as an empirical mapping relationship library. During querying, the average reflectance of any grid point in the gridded substrate surface model is input into the functional mapping relationship for calculation, and the estimated porosity of the grid point region is output.

[0103] The process of calculating the initial coating dose through iterative approximation by combining the corrected theoretical pigment capacity and the required pigment coverage thickness involves setting an initial iterative dose value. Based on this initial iterative dose value, the predicted coverage thickness formed after pigment deposition at the current dose is calculated according to the physical model of pigment coverage on the surface. The predicted coverage thickness is compared with the required pigment coverage thickness, and it is determined whether the total volume of pigment at the predicted coverage thickness exceeds the corrected theoretical pigment capacity. If the predicted coverage thickness is less than the required pigment coverage thickness and the total pigment volume does not exceed the corrected theoretical pigment capacity, the iterative dose value is increased by a preset step size. If the predicted coverage thickness is greater than or equal to the required pigment coverage thickness, or the total pigment volume has reached the corrected theoretical pigment capacity, the iteration terminates. The iterative dose value obtained in the last iteration that simultaneously satisfies the thickness and capacity constraints is set as the maximum pigment dose, i.e., the initial coating dose.

[0104] In practical implementation, establishing an empirical mapping database between reflectance characteristics and surface porosity requires the prior acquisition of various standard substrate samples with known porosity, obtained through laboratory physical measurement methods. Under standard illumination conditions, multispectral reflectance data for each standard substrate sample is acquired, and the average reflectance of each sample in each spectral channel is calculated. Standard illumination conditions refer to a measurement environment where the light source spectrum, illumination angle, and sensor receiving angle are calibrated and fixed. For each standard substrate sample, its known porosity and calculated average reflectance are correlated to form a mapping pair. Using these mapping pairs for all standard substrate samples, a functional mapping relationship between average reflectance and porosity is trained through nonlinear regression analysis, and this functional mapping relationship is stored in an empirical mapping database. During querying, the average reflectance of any grid point in the gridded substrate surface model is input into the functional mapping relationship for calculation, and the estimated porosity of the grid area is output. Nonlinear regression analysis aims to find a mathematical model that best fits the complex relationship between reflectance and porosity data; its functional relationship can be generally expressed as:

[0105]

[0106] Where: symbol This represents the estimated porosity calculated using a function mapping relationship, with the symbol... This represents the mapping function obtained through nonlinear regression training, with the symbol... This indicates that the grid points in the meshed substrate surface model are... The average reflectance value calculated under different spectral channels.

[0107] In some embodiments, the initial spraying dose is calculated iteratively by combining the corrected theoretical pigment capacity with the required pigment coverage thickness. An initial iterative dose value is set, which can be initially estimated based on the required pigment coverage thickness and an empirical unit consumption. Based on the initial iterative dose value, the predicted coverage thickness formed after pigment deposition at the current dose is calculated according to a physical model of pigment coverage on the surface. The physical model considers the pigment deposition efficiency, leveling characteristics, and filling behavior in the substrate pores. The predicted coverage thickness is compared with the required pigment coverage thickness, and it is determined whether the total volume of pigment at the predicted coverage thickness exceeds the corrected theoretical pigment capacity. If the predicted coverage thickness is less than the required pigment coverage thickness and the total pigment volume does not exceed the corrected theoretical pigment capacity, the iterative dose value is increased by a preset step size. If the predicted coverage thickness is greater than or equal to the required pigment coverage thickness, or the total pigment volume has reached the corrected theoretical pigment capacity, the iteration is terminated, and the iterative dose value obtained in the last iteration that simultaneously satisfies the thickness and capacity constraints is set as the maximum pigment dose, i.e., the initial spraying dose.

[0108] Optionally, nonlinear regression analysis can be performed using multinomial regression, support vector regression, or neural network algorithms. In some embodiments, the preset step size in the iterative approximation calculation can be set according to the computational accuracy requirements and computational efficiency. The smaller the step size, the higher the accuracy of the final dose value, but the more iterations are required. It is understood that the termination condition of the iteration is a dual judgment, which must simultaneously consider the requirements of visual coverage thickness and the limitations of the physical capacity of the substrate, so as to ensure that the calculated initial spraying dose meets the visual requirements of the decorative appearance, and does not cause pigment accumulation or waste due to exceeding the pore saturation point.

[0109] See Figure 5 This is a nonlinear correlation analysis chart showing the relationship between the average reflectance of the substrate and the estimated porosity. The average reflectance and estimated porosity are negatively correlated; the higher the reflectance, the lower the estimated porosity. This is because the lower the porosity of the substrate (the denser the structure), the stronger its ability to reflect light. Conversely, a loose structure with high porosity increases light scattering and absorption, reducing reflectance. The high degree of fit between the experimental data points and the fitted curve indicates that this nonlinear regression model can accurately characterize the relationship between the two, serving as a quantitative prediction tool for "reflectance → porosity". This type of chart is a core tool for analyzing material surface properties, used to quickly infer substrate porosity from reflectance data, providing a basis for subsequent calculations of process parameters.

[0110] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0111] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An in-line high speed sanitary napkin spray decoration process characterized by, Includes the following steps: An optical sensor array positioned upstream of the conveyor line is used to acquire a real-time surface image stream of the sanitary napkin substrate, which includes multispectral reflectance data and surface micromorphology data. A dynamic reference plane fitting operation is performed on the real-time surface image stream to generate a virtual reference plane of the sanitary napkin substrate in the current motion state. The virtual reference plane is generated based on the statistical median of the surface micromorphology data. Multiple spray control grids are defined on the virtual reference plane, and the multispectral reflectance data is mapped to each grid point of the spray control grid to generate a gridded substrate surface model with reflectance features. Based on the reflection characteristics of each grid point in the gridded substrate surface model, the theoretical saturation point of pigment adhesion is solved in reverse, and the initial spraying dose corresponding to each grid point is calculated based on the theoretical saturation point. The sequence of the initial spraying dosage is input into the spraying timing planner, and combined with the substrate conveying speed, a spraying phase and dosage instruction table for multi-nozzle collaborative operation is generated. The step of performing a dynamic reference plane fitting operation on the real-time surface image stream to generate a virtual reference plane for the sanitary napkin substrate in the current motion state includes: From the real-time surface image stream, extract the surface elevation dataset of the sanitary napkin substrate at multiple consecutive sampling times; The surface elevation dataset is subjected to outlier filtering and smoothing to obtain a clean elevation dataset. The outlier filtering is based on the statistical outlier degree of the elevation value in the neighborhood. Based on the clean elevation dataset, the least squares plane fitting rule is used to calculate the fitting plane equation describing the overall tilting and bending trend of the substrate. Subtract the theoretical height values ​​calculated by the fitted plane equation at each point from the elevation values ​​of each point in the clean elevation dataset to obtain the set of relative height residuals; Using the median absolute value of the relative height residual set as the reference offset, the fitted plane equation is translated and corrected to generate the final virtual reference plane.

2. The online high-speed sanitary napkin spraying decoration method according to claim 1, characterized in that, The process of defining multiple spray control grids on the virtual reference plane and mapping the multispectral reflectance data to each grid point of the spray control grids to generate a gridded substrate surface model with reflectance features includes: A Cartesian coordinate system is established with the geometric center of the virtual reference plane as the origin; Within the Cartesian coordinate system, a uniform grid array is divided along the length and width of the substrate according to predetermined physical dimensions. Each cell of the grid array is a spray control grid. The multispectral reflectance data acquired at each sampling time is associated with one or more grid points in the spraying control grid through coordinate transformation based on its corresponding actual spatial coordinates. For each of the spray control grids, all multispectral reflectance data associated with its lifetime are aggregated, and the average reflectance and reflectance variance of the grid points in multiple spectral channels are calculated. The average reflectivity and reflectivity variance are used as the reflection characteristics of the grid points and assigned to the gridded substrate surface model to complete the model's attribute construction.

3. The in-line high speed sanitary napkin spray decoration process according to claim 2, characterized in that, The process involves calculating the theoretical saturation point of pigment adhesion based on the reflection characteristics of each grid point in the gridded substrate surface model, and then calculating the initial spraying dosage corresponding to each grid point based on the theoretical saturation point, including: An empirical mapping relationship library between reflection characteristics and surface porosity is established. Based on the average reflectance of the grid points, the estimated porosity of the grid point region is obtained by querying the empirical mapping relationship library. Based on the estimated porosity and pigment characteristic parameters, the micro-volume that pigment can fill per unit area is calculated, and the micro-volume is defined as the theoretical pigment capacity of the grid point. The reflectance variance is introduced as a surface uniformity influencing factor to correct the theoretical pigment capacity for uniformity, thus obtaining the corrected theoretical pigment capacity. Set the visual saturation level of the target decorative layer, and convert the visual saturation level into the desired pigment coverage thickness; By combining the corrected theoretical pigment capacity with the required pigment coverage thickness, the maximum pigment dosage that does not exceed the capacity under the premise of satisfying the coverage thickness is calculated through iterative approximation calculation. The maximum pigment dosage is the initial spraying dosage.

4. The in-line high speed sanitary napkin spray decoration process according to claim 3, characterized in that, The step of inputting the sequence of the initial spray dosage into the spray timing planner, and generating a spray phase and dosage instruction table for multi-nozzle collaborative operation in combination with the substrate delivery speed, includes: Obtain the constant conveying speed of the production line, as well as the fixed spatial distribution of multiple spray nozzles in the conveying direction; The zero point of time is defined as the projection position of the front end of the gridded substrate surface model in the conveying direction reaching the first spray nozzle; Based on the conveying speed, calculate the precise time points at which the center point of each of the spraying control grids passes under each of the spraying nozzles in sequence; Based on the sequence of initial spraying doses, each spraying control grid is assigned a dose ratio that should be sprayed when passing through different nozzles, the dose ratio being determined based on the atomization characteristics and overlapping areas of each nozzle; By combining the precise time point of each grid with the corresponding dose ratio, a time-sorted instruction list is generated for each spray nozzle. The instruction list contains instructions to apply a specific dose at a specific time. The instruction lists of all nozzles constitute the spray phase and dose instruction table.

5. The in-line high speed sanitary napkin spray decoration process according to claim 4, characterized in that, Based on the clean elevation dataset, the least squares plane fitting method is used to calculate the fitting plane equation describing the overall tilt and bending trend of the substrate, including: Read the three-dimensional spatial coordinates of all sampling points from the clean elevation dataset; The fitted plane equation is preset to a standard plane equation form containing three undetermined plane coefficients; Construct an objective function that is the sum of squares of the differences between the actual height values ​​of all sampling points and the theoretical height values ​​calculated from the fitted plane equation; By taking the partial derivatives of the objective function with respect to the three undetermined plane coefficients, and setting each partial derivative to zero, a system of linear equations with respect to the three undetermined plane coefficients is obtained. Solving the system of linear equations yields the specific values ​​of the three undetermined plane coefficients, thereby determining the equation of the fitted plane.

6. The in-line high speed sanitary napkin spray decoration process according to claim 5, characterized in that, The step of establishing an empirical mapping database between reflection characteristics and surface porosity, and querying the database to obtain the estimated porosity of the grid region based on the average reflectance of the grid points, includes: Pre-collect standard substrate samples with various known porosities; Under standard illumination conditions, multispectral reflectance data of each standard substrate sample were acquired, and its average reflectance in each spectral channel was calculated. For each of the aforementioned standard substrate samples, its known porosity is correlated with the calculated average reflectance to form a mapping relationship pair; Using the mapping pairs of all standard substrate samples, a functional mapping relationship between average reflectance and porosity is obtained through nonlinear regression analysis, and the functional mapping relationship is stored as the empirical mapping relationship library; During the query, the average reflectance of any grid point in the gridded substrate surface model is input into the function mapping relationship for calculation, and the estimated porosity of the grid point region is output.

7. The in-line high speed sanitary napkin spray decoration process according to claim 6, characterized in that, The modified theoretical pigment capacity and the required pigment coverage thickness are combined, and through iterative approximation calculations, the maximum pigment dosage that does not exceed the capacity while satisfying the coverage thickness is determined. This maximum pigment dosage is the initial spraying dosage, including: Set an initial iterative dose value; Based on the initial iterative dose value, and according to the physical model of pigment coverage on the surface, the predicted coverage thickness formed after pigment deposition at the current dose is calculated. The predicted coverage thickness is compared with the required pigment coverage thickness, and it is determined whether the total volume of pigment under the predicted coverage thickness exceeds the corrected theoretical pigment capacity. If the predicted coverage thickness is less than the required pigment coverage thickness, and the total volume of the pigment does not exceed the corrected theoretical pigment capacity, then the iterative dose value is increased by a preset step size. If the predicted coverage thickness is greater than or equal to the required pigment coverage thickness, or if the total volume of the pigment has reached the corrected theoretical pigment capacity, then the iteration terminates. The iterative dose value obtained in the last iteration that simultaneously satisfies the thickness and capacity constraints is set as the maximum pigment dose, i.e., the initial spraying dose.

8. The in-line high speed sanitary napkin spray decoration process according to claim 7, characterized in that, The step of taking the projection position of the front end of the gridded substrate surface model in the conveying direction reaching the first spray nozzle as the zero point of time includes: Obtain the geometric shape data of the surface model of the gridded substrate and determine its front boundary coordinates in the conveying direction; Obtain the precise spatial position coordinates of the first spray nozzle in the conveyor line coordinate system; Project the coordinates of the front-end boundary along a plane perpendicular to the conveyor line onto the vertical projection line where the first spray nozzle is located; Calculate the straight-line distance required to move in the conveying direction from the front boundary coordinates to coincide with the vertical projection line of the first spray nozzle; Calculate the time required to travel the straight-line distance based on the conveying speed; The moment after the meshed substrate surface model begins to move and the time required to move the calculated straight-line distance is set as the zero point of time.

9. An online high-speed sanitary napkin spraying decoration system, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, The processor, when executing the computer program, implements the steps of the online high-speed sanitary napkin spray decoration method according to any one of claims 1 to 8.

Images