Method for measuring the spread of a disposable hygiene article
By constructing a local diffusion tensor and a Riemannian metric space, the problem of accurately quantifying the anisotropy of liquid diffusion in the core material of hygiene products in traditional methods is solved, enabling accurate quantification and evaluation of liquid diffusion and improving the accuracy and reliability of the evaluation.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- BEIJING BEISHUTE MATERNITY & CHILD ARTICLES CO LTD
- Filing Date
- 2026-01-15
- Publication Date
- 2026-04-17
AI Technical Summary
Traditional methods cannot accurately quantify the anisotropic properties of liquid diffusion in the core material of hygiene products, leading to discrepancies between the evaluation results and actual user experience. Furthermore, existing technologies lack mathematical characterization of diffusion direction, resulting in distorted reconstruction results.
By acquiring time-series signals from multiple sensors at multiple time points, the liquid arrival time difference and spatial distance between paired sensors are calculated, a local diffusion tensor is constructed, the anisotropic effective distance is calculated using the Riemann metric space, a continuous spatiotemporal saturation map is reconstructed using an interpolation algorithm, and a quality assessment is performed by combining a dynamic time warping algorithm.
This method enables accurate quantification of liquid diffusion in the core material of hygiene products, solving the map distortion problem caused by neglecting material anisotropy in traditional methods, and improving the accuracy and reliability of the assessment.
Smart Images

Figure CN121540595B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of data processing technology, and more particularly to a method for measuring the spread of disposable hygiene products. Background Technology
[0002] In the field of hygiene products, the diffusion performance of liquids within the core material is a key indicator for evaluating product comfort and effectiveness. Traditional evaluation methods primarily rely on linear measurements in a single direction or macroscopic observations of overall immersion time. These methods fail to capture the multidimensional and non-uniform diffusion behavior of liquids in complex porous media. Due to the anisotropic characteristics of fiber orientation, density distribution, and pore structure within the core material, liquids exhibit significantly different permeation rates in different directions. Traditional single-point or unidirectional measurements cannot fully reflect this spatial heterogeneity, leading to discrepancies between evaluation results and actual user experience.
[0003] Existing technologies attempt to improve measurement accuracy by increasing sensor density, but significant shortcomings remain at the data analysis level. Most solutions process or simply average the signals from each sensor independently, ignoring the dynamic correlation information of liquid propagation between points. This approach cannot construct accurate local diffusion models, failing to quantify the degree of anisotropy and struggling to identify preferred paths and hindrance regions during diffusion. More importantly, the lack of a mathematical representation of diffusion directionality forces subsequent spatial interpolation reconstruction to rely on simplified isotropic assumptions, leading to distorted reconstruction results.
[0004] In spatial reconstruction, traditional methods generally employ interpolation algorithms based on Euclidean distance, which has inherent limitations in practical applications. Because the influence of material anisotropy on the liquid propagation path is not considered, Euclidean distance cannot reflect the actual propagation cost of the liquid in different directions, resulting in a reconstructed diffusion map that obscures the true diffusion front and directional preferences. This systematic error masks the true diffusion characteristics of the material, making performance evaluation and quality control lack reliable basis. Therefore, a spatial reconstruction method that can integrate physical diffusion characteristics is urgently needed. Summary of the Invention
[0005] To address the technical problem that traditional testing methods cannot accurately quantify the anisotropic properties of liquid diffusion in the core materials of hygiene products, this invention provides solutions in the following aspects.
[0006] Methods for measuring the spread of disposable hygiene products include:
[0007] Acquire the timing signals of multiple discrete coordinates at multiple time points inside the core of the sanitary product under test;
[0008] Based on the timing signal, the time difference of liquid arrival between any two sensors is calculated, and combined with the spatial distance between the two sensors, a pair of sensor activation speeds characterizing the local diffusion rate is constructed.
[0009] For each of the aforementioned sensors, the distribution of the activation velocities of the paired sensors in its neighborhood is summarized, and a local diffusion tensor is constructed through statistical analysis to characterize the local diffusion anisotropy.
[0010] Based on the local diffusion tensor field composed of all the sensors, the anisotropic effective distance between the interpolation point and each of the sensors is calculated.
[0011] Based on the value of the time series signal and the effective distance of the anisotropy, an interpolation algorithm is used to reconstruct a continuous spatiotemporal saturation map.
[0012] Quality assessment was conducted based on a continuous spatiotemporal saturation map, and the diffusion measurement results of disposable hygiene products were obtained.
[0013] Preferably, the method for constructing the local diffusion tensor includes: representing the activation velocities of multiple pairs of sensors in the neighborhood as velocity vectors; calculating the covariance matrix of the velocity vectors; performing eigenvalue decomposition on the covariance matrix, taking the direction of the obtained eigenvector as the principal direction of local diffusion, and taking the eigenvalue as the diffusion intensity in that direction.
[0014] Preferably, the calculation of the anisotropic effective distance includes: treating the local diffusion tensor field as a Riemann metric space; solving the geodesic problem from the sensor to the interpolation point in the metric space; and defining the geodesic length as the anisotropic effective distance.
[0015] Preferably, the geodesic problem is obtained by solving the Eikonal equation.
[0016] Preferably, the method further includes quality assessment based on the continuous spatiotemporal saturation map, specifically including: extracting diffusion features from the map, the diffusion features including the shape of the diffusion front, the anisotropy ratio, and the curve of saturation changing over time; matching the diffusion features with a pre-stored standard sample feature map; and outputting the quality assessment result based on the matching degree.
[0017] Preferably, the matching process uses a dynamic time warping algorithm to calculate the similarity between the curve and the standard curve.
[0018] Preferably, the quality assessment includes: calculating a global diffusion anisotropy index based on the local diffusion tensors of all sensors; comparing the index with a preset threshold; and determining the product as unqualified when the index exceeds the threshold range.
[0019] Secondly, this application provides a system for measuring the spread of disposable hygiene products, employing the following technical solution:
[0020] A system for measuring the diffusion of disposable hygiene products includes a processor and a memory, wherein the memory stores computer program instructions that, when executed by the processor, implement the method for measuring the diffusion of disposable hygiene products as described above.
[0021] The present invention has the following effects:
[0022] 1. By converting discrete sensor signals into tensor fields that characterize the physical properties of materials, and redefining spatial distances for interpolation, the fundamental problem of severe distortion of diffusion maps caused by neglecting material anisotropy in traditional methods is solved.
[0023] 2. By utilizing the relative temporal relationships in sensor networks, traditional absolute time monitoring is transformed into the quantification of local propagation dynamics, providing a unique and reliable data source for the subsequent construction of a local diffusion tensor that can reveal the dominant direction and intensity of diffusion.
[0024] 3. By treating the tensor field as a metric space and calculating the geodesic distance, the interpolation weights are automatically increased in directions where diffusion is easy and automatically decreased in directions where diffusion is difficult, thus naturally presenting diffusion anisotropy in the final map. Attached Figure Description
[0025] Figure 1 This is a flowchart of steps S1-S6 in the method for measuring the spread of disposable hygiene products according to an embodiment of the present invention. Detailed Implementation
[0026] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.
[0027] Reference Figure 1 The method for measuring the diffusion of disposable hygiene products includes steps S1-S6, as detailed below:
[0028] S1: Obtain the timing signals of multiple discrete coordinates at multiple time points inside the core of the sanitary product under test.
[0029] In measuring the diffusion of the core of hygiene products, traditional methods often rely on static detection at a single location or a limited number of points, which cannot fully capture the complex process of dynamic diffusion of liquid inside the core, resulting in significant randomness and inaccuracy in the measurement results.
[0030] Specifically, core materials typically have anisotropic structures, and the liquid diffusion rate varies significantly in different directions. If only a few detection points are used, it will be difficult to accurately reflect the overall diffusion behavior, especially missing key information about local fast or slow diffusion areas.
[0031] Furthermore, the diffusion process changes rapidly over time, and the lack of high temporal resolution monitoring methods makes it impossible to accurately record the movement trajectory of the liquid front, thus affecting the reliability of subsequent anisotropy analysis. Therefore, there is an urgent need for a method that can simultaneously acquire high temporal resolution signals from multiple locations within the core to lay a data foundation for constructing an accurate diffusion model.
[0032] An array-style sensing system is deployed to simultaneously acquire time-series signals from multiple discrete locations within the core, overcoming the limitations of traditional methods. Multiple sensors are embedded in a grid pattern within the core or at contact surfaces. Each sensor can independently detect changes in physical or chemical signals, such as resistance, capacitance, or optical properties, upon the arrival of the liquid. i Each sensor is denoted as i=1,2,..., N , N The total number of sensors is and the sampling interval is . Δt =0.2 seconds.
[0033] By analyzing the time-series signals of different sensors under continuous conditions, the dynamic propagation path of liquid in space can be reconstructed. This can be used to solve the problem of the one-sidedness of single-point detection and to capture the details of diffusion anisotropy through high-density point deployment.
[0034] In practical implementation, sensors typically employ miniature electrodes or fiber optic sensors, arranged in a matrix. The number of rows and columns can be adjusted according to the core size. For example, for a standard hygiene product, it can be set to... M Line × K An array of columns, where M and K All values are integers, and the empirical value range is... M =12, K =10.
[0035] Each sensor is connected to the control host via a data acquisition card to record signal values in real time, with the sampling frequency set to [value to be specified]. f experience value f =5Hz to ensure capture of rapid diffusion events. At the start of the measurement, a fixed volume of test liquid is applied to the core, simultaneously triggering the data acquisition system to record the signal timing data of each sensor from the initial state to the arrival of the liquid, denoted as . ,in, i Number the sensor. tFor a given point in time, this sensor data is typically output in the form of voltage or current and stored digitally via an analog-to-digital converter.
[0036] To ensure the quality of timing signals, a data preprocessing step is performed after acquisition to eliminate the influence of noise. Specifically, a filtering algorithm is applied to the raw signal of each sensor, for example, using a cutoff frequency set to... A low-pass filter is used to smooth the data. The empirical value is 2Hz, which is used to denoise the acquired time-series data and convert the sensor measurement signal values into corresponding physical quantities through a calibration procedure. The physical quantities can be saturation or humidity, for example.
[0037] In addition, data from all sensors is synchronized via timestamps to ensure time alignment of signals from different locations, thus guaranteeing the accuracy of the entire measurement process.
[0038] S2: Based on the time-series signal, calculate the time difference of liquid arrival between any two sensors, and combine it with the spatial distance between the two sensors to construct the paired sensor activation rate characterizing the local diffusion rate.
[0039] The time-series signal acquired in step S1 is isolated point information. To reveal the propagation of the liquid in the sensor network, it is necessary to analyze the correlation between points. Therefore, the pairwise sensor activation velocities between any two sensors are calculated, transforming the discrete absolute time signal into a vector relationship characterizing the local propagation rate.
[0040] In traditional evaluation of the diffusion performance of liquids in hygiene products, the diffusion rate is usually estimated by a single direction or overall averaging method. This method cannot effectively capture the anisotropic diffusion characteristics caused by fiber orientation or structural inhomogeneity within the material.
[0041] When a liquid propagates within a core, the diffusion rate in different directions may vary significantly. Simple averaging calculations can mask these critical local dynamic characteristics, leading to misjudgments of the product's actual performance.
[0042] Furthermore, the local interaction information between points is weakened during the overall averaging process, making it difficult to accurately identify bottleneck areas or fast channels in the diffusion path, thus affecting the comprehensive evaluation of the product's water absorption and liquid retention performance.
[0043] To address the aforementioned issues, this study reveals the intrinsic diffusion characteristics of materials by quantifying the liquid propagation relationship between any two sensors. Liquid diffusion is viewed as a networked process, where each sensor pair provides propagation information within a local region. By analyzing these point-to-point dynamic relationships, a diffusion velocity field is constructed. This approach effectively avoids information loss associated with overall averaging methods, preserves the direction dependence during diffusion, and improves the accuracy of subsequent anisotropy analysis.
[0044] In the specific implementation process, the liquid arrival time of each sensor is determined, for the first... i Timing signals of each sensor After smoothing, the precise arrival time of the liquid is determined using a thresholding method. .
[0045] Specifically, when the signal strength exceeds the baseline noise standard deviation k When the time exceeds the minimum stability threshold and the duration exceeds the minimum stability threshold, the time is doubled. This means that it is considered a valid activation, where, k This is an empirical coefficient, usually taken as... k =2, These are empirical values, usually... =0.5 seconds. When a valid activation is determined, the time corresponding to the valid activation is obtained as the liquid arrival time for each sensor.
[0046] After obtaining the activation time of each sensor, calculate the time difference between any two sensors. For sensor pairs The formula for calculating the time difference is: ,in and These represent the liquid arrival times of the two sensors, respectively.
[0047] Due to shared N One sensor, so it can generate The time difference data points constitute a complete time difference matrix. This matrix captures the temporal relationship of liquid propagation between different locations, reflecting the spatial dynamic characteristics of the diffusion process.
[0048] Simultaneously, the spatial distance between each sensor pair is calculated. Given the sensor coordinates, the actual physical distance between sensor pairs is calculated using the Euclidean distance formula. Since the core is approximately planar after unfolding, the row and column numbers of different sensors in the array can be used as the coordinate values of each sensor.
[0049] Based on time difference and spatial distance data, and combined with the well-known physical velocity formula, the paired sensor activation velocity characterizing the local diffusion rate can be constructed. ,in This represents the average propagation speed from the first activated sensor to the last activated sensor.
[0050] when When the time sampling interval is less than the time sampling interval, it is considered to be activated simultaneously. Record as the maximum value . use The propagation rate of the liquid between different sensor pairs is described, which provides crucial vector information for the subsequent construction of the diffusion tensor.
[0051] S3: For each sensor, summarize the distribution of paired sensor activation velocities in its neighborhood, and construct a local diffusion tensor to characterize local diffusion anisotropy through statistical analysis.
[0052] In liquid diffusion analysis, the activation rate of a single pair of sensors can only reflect the linear propagation characteristics between two points and cannot fully characterize the overall diffusion pattern within a local area.
[0053] Especially in anisotropic materials, the permeability of liquids varies significantly in different directions. Relying solely on isolated rate values makes it difficult to quantify this direction-dependent diffusion behavior. Traditional methods often ignore the spatial distribution characteristics of velocity vectors, resulting in an inability to accurately capture the diffusion heterogeneity caused by the internal structure of the material, thus affecting the precise evaluation of core performance.
[0054] The paired sensor activation velocities obtained in step S2 are discrete vectors with different orientations. In order to form a unified characterization of anisotropy at each sensor location, statistical analysis of these vectors is required. The local diffusion tensor can simultaneously describe the magnitude and orientation preference of the diffusion rate.
[0055] Furthermore, to describe the diffusion anisotropy characteristics within the neighborhood of each sensor, a local diffusion tensor is constructed. Velocity vector information from multiple directions around a single sensor is integrated and statistically analyzed to form a second-order tensor. This tensor simultaneously reflects the magnitude and directionality of the diffusion rate, transforming discrete vector observations into a continuous tensor field, thereby characterizing the diffusion anisotropy of the material.
[0056] In the specific implementation process, a neighborhood range is defined for each sensor. Neighborhood radius R This is a key parameter, and its empirical value is usually set to 3 times the sensor spacing to ensure that enough velocity vector samples are included while preserving local characteristics.
[0057] Furthermore, all data related to the first... i One sensor Distance less R Sensor composition neighborhood set .
[0058] Determining the neighborhood range Then, extract the activation velocity vectors of all relevant pairs of sensors within the region. For each sensor pair in the neighborhood, where at least one point is located within the neighborhood, its velocity vector can be represented as a quantity with magnitude and direction.
[0059] The direction of the vector is determined by the spatial direction from the first activated sensor to the second activated sensor, and its magnitude is the previously calculated pairwise sensor activation velocity. V Furthermore, the velocity vector can be used to construct discrete samples of the local velocity vector field.
[0060] After obtaining the number i One sensor After obtaining the set of local velocity vectors, the local diffusion tensor is constructed by calculating the covariance matrix of this set. The covariance matrix can effectively characterize the dispersion and directional trend of a set of vectors distributed around its mean. The direction of its eigenvectors represents the dominant direction of local diffusion, and the magnitude of the eigenvalues represents the diffusion intensity in that direction.
[0061] ;
[0062] in, n This represents the number of effective velocity vectors within the neighborhood.
[0063] Indicates the first k A velocity vector.
[0064] m It is the mean of all effective velocity vectors in the neighborhood.
[0065] The covariance matrix describes the distribution characteristics of the velocity vector in a local region. The eigenvectors of the covariance matrix represent the dominant direction of local diffusion, and the eigenvalues represent the diffusion intensity, thus quantifying anisotropy.
[0066] For the constructed diffusion tensor T By performing eigenvalue decomposition, we can obtain ,in, Q It is an orthogonal matrix composed of eigenvectors. L It is a diagonal matrix whose eigenvalues are the diagonal elements.
[0067] The eigenvector corresponding to the largest eigenvalue indicates the main local diffusion direction, while the ratio between eigenvalues quantifies the degree of diffusion anisotropy.
[0068] Using the diffusion tensor T The decomposition results of the eigenvalues yield the dominant direction of liquid diffusion and the intensity of anisotropy.
[0069] S4: Based on the local diffusion tensor field composed of all sensors, calculate the anisotropic effective distance between the interpolation point and each sensor.
[0070] In liquid diffusion analysis, traditional distance metrics such as Euclidean distance cannot reflect the influence of material anisotropy on the liquid propagation path. When the core material has a significant directional structure, the diffusion resistance of the liquid varies significantly in different directions, and uniform isotropic distance calculations will distort the true propagation path length, leading to incorrect representation of spatial relationships in subsequent interpolation processes.
[0071] To address the aforementioned issues, this paper treats the local diffusion tensor as a Riemannian metric tensor and redefines the shortest path between two points in curved space, thereby incorporating the anisotropic properties of the material into distance calculations. This results in a relatively shorter path length in directions with lower diffusion resistance and a relatively longer path length in directions with higher resistance, more accurately reflecting the propagation cost of liquids in real materials and providing a correct spatial relationship basis for subsequent anisotropic interpolation.
[0072] In practical implementation, due to diffusion anisotropy, the cost of liquid propagation varies in different directions. Riemannian metric spaces can accurately model this non-uniformity, transforming the local diffusion tensor field on a discrete sensor into a continuous Riemannian metric field. By using radial basis function interpolation, for any spatial location... x Its metric tensor The diffusion tensor of surrounding sensors The weighted calculation yields:
[0073] ;
[0074] in, Radial basis functions, using Gaussian functions Scale parameters s The empirical value is 1.5 times the average spacing between sensors.
[0075] N The number of neighboring sensors involved in the interpolation.
[0076] Then, using the metric tensor Discrete tensor observations can be transformed into continuous tensor fields.
[0077] Based on the established Riemannian metric field, the points to be interpolated x With sensors Anisotropic effective distance between It is obtained by solving the geodesic equation.
[0078] In Riemannian geometry, the length of a geodesic is obtained by path integral. The integral calculation is the minimum cost path length after considering local anisotropy, which reflects the ease or difficulty of liquid propagation along the path. The process of obtaining geodesics using path integral is a well-known technique and will not be elaborated further.
[0079] In practical numerical calculations, the fast travel method is used to solve the Eikonal equations to efficiently calculate the effective anisotropic distance.
[0080] Specifically, solving the equation ,in Indicating in measurement G The vector norm under the boundary conditions is =0. This partial differential equation describes the equation from the sensor The propagation process of a departing wavefront in an anisotropic medium, and its numerical solution. u(x) This is the desired anisotropic effective distance. The time-space discretization step size is calculated. h It is usually set to 1 / 5 of the sensor spacing, but can be adjusted by the implementer according to the specific implementation scenario.
[0081] S5: Based on the value of the time series signal and the effective distance of anisotropy, an interpolation algorithm is used to reconstruct a continuous spatiotemporal saturation map.
[0082] After obtaining the time-series signals and anisotropic effective distances from discrete sensors, accurate reconstruction of the continuous diffusion state of the entire core region is crucial. Traditional spatial interpolation methods often assume that the medium is isotropic, which severely distorts the actual diffusion morphology of liquids in anisotropic materials.
[0083] Because such methods fail to account for the directional differences in diffusion resistance, the maps generated will blur the true diffusion front and fail to accurately show the preferred direction of liquid flow and the blocked areas, leading to misjudgments of product performance.
[0084] To address the aforementioned issues, an interpolation algorithm that integrates anisotropic effective distances is used to incorporate the directional diffusion characteristics of materials into the spatial reconstruction process.
[0085] Using the previously calculated anisotropic effective distance as the weight basis for interpolation, which is shorter in directions where diffusion is easy and longer in directions where diffusion is difficult, the interpolation results naturally conform to the actual diffusion path of the material. This ensures that the reconstructed continuous map not only depends on geometric proximity but also reflects physical diffusion connectivity, thus conforming to the dynamic distribution of the liquid in the core.
[0086] In the specific implementation process, the first step is to determine the continuous spatial grid to be reconstructed. For a two-dimensional core, the region is discretized into... M × N Regular grid dot matrix, grid spacing Δx and Δy It should be set according to the accuracy requirements of the application, and is usually 1 / 3 of the original sensor spacing.
[0087] Each grid point This refers to a point to be interpolated, where the anisotropic effective distance between it and all sensors is calculated. Based on the anisotropic effective distance, the saturation estimate of each grid point is calculated using the inverse distance weighted algorithm.
[0088] For grid point g, the interpolation formula is: ;
[0089] Among them, weight .
[0090] For sensors The timing signal value.
[0091] p The weighting index is based on empirical values. p =1.5.
[0092] Using interpolation formulas The closer the sensor is to the grid point, the greater its influence. The effective distance of anisotropy already implies the influence of diffusion directionality, thus making the interpolation results more accurate.
[0093] In the time dimension, the above spatial interpolation process is performed independently for each time frame to construct a complete spatiotemporal saturation map. To ensure temporal smoothness, temporal filtering is applied between adjacent time frames, using a first-order IIR filter for data filtering. The empirical value for the smoothing factor in the first-order IIR filter is 0.7.
[0094] Finally, a continuous spatiotemporal saturation map is generated, corresponding to spatial coordinates ( x , y ) and time t The dynamic evolution of liquid diffusion was obtained.
[0095] S6: Quality assessment is performed based on a continuous spatiotemporal saturation map to obtain the diffusion measurement results of disposable hygiene products.
[0096] In the quality assessment of disposable hygiene products, it is difficult to quantify the anisotropic diffusion behavior, resulting in inaccurate and inconsistent assessment results. In order to accurately extract the dynamic characteristics of diffusion and reduce the probability that defective products may enter the market or qualified products may be misjudged, continuous spatiotemporal saturation maps are used for quality assessment.
[0097] From continuous spatiotemporal saturation map Extract diffusion features from the data. Define the shape feature vector of the diffusion front. ,in, Indicates the eccentricity of the leading edge, used to describe the degree of elongation of the leading edge shape. The frontal area reflects the diffusion range. The dominant direction angle indicates the direction of diffusion dominance.
[0098] The diffusion anisotropy of the core material originates from the inhomogeneity of internal pores and fiber orientation; the eigenvalues of the local diffusion tensor characterize the differences in diffusion intensity in different directions. Therefore, the anisotropy ratio is obtained by calculating the global average of the eigenvalues of the local diffusion tensors of all sensors. ,in For the number of sensors, and They are respectively The maximum and minimum eigenvalues of the sensor tensor, where the maximum eigenvalue corresponds to the intensity in the dominant diffusion direction and the minimum eigenvalue corresponds to the intensity in the perpendicular direction, are used as a ratio to eliminate the influence of absolute magnitude and highlight relative differences, thus revealing the anisotropy ratio. It can reflect the intensity difference in the direction of diffusion.
[0099] Liquid diffusion is a dynamic process that requires time series analysis to capture the overall absorption behavior. Based on the principle of integral averaging, this curve maps the spatially distributed saturation to a time function, avoiding the random errors of single-point measurements. This allows for the generation of a saturation-time curve through a spatial integral map. ,in, This represents the total area of the core. This is a continuous spatiotemporal saturation map, representing the location... and time The liquid saturation (between 0 and 1), so The curve is a function of time; its shape reflects changes in diffusion motion. A rapid rise in the curve indicates efficient absorption, while a decline indicates saturation. Used to characterize the overall liquid absorption process.
[0100] In actual diffusion processes, both anisotropy ratio and shape differences affect the uniformity of liquid distribution. Combining these two factors allows for the quantification of the deviation of the test sample from the ideal state, utilizing the anisotropy ratio... and shape difference Theoretical correction factor for calculating saturation curve:
[0101] Define diffusion coordination coefficient ,in, The anisotropy ratio of the test sample is calculated from sensor data; The optimal anisotropy ratio is taken as an empirical value of 2.2, which can be determined based on the average value of the best-performing sample in historical data, representing the ideal equilibrium state. As a tolerance scale, the sensitivity to anisotropic deviation can be controlled by taking an empirical value of 0.8 from historical data; the larger the value, the higher the tolerance. To determine the shape difference, the test data and reference shape features are calculated using Euclidean distance. Differences; To determine the maximum permissible shape variation, an empirical value of 1.5 can be used, based on the maximum shape deviation of qualified products in historical data; furthermore... Values between 0 and 1 indicate better compatibility; a larger value signifies better anisotropy. The exponential part emphasizes that anisotropy is close to optimal, while the linear part emphasizes shape similarity. Multiplying the two ensures that only when both anisotropy and shape are close to ideal is the compatibility achieved. Only then can it have a high value.
[0102] The measured saturation curve may be affected by random noise or local anomalies, requiring smoothing and standardization using a compatibility coefficient. By combining the test curve with the reference curve and correcting the error of the measured saturation curve based on the consistency coefficient, a standardized saturation index is obtained:
[0103]
[0104] in, This is the original saturation curve of the test sample.
[0105] The saturation curve of the reference sample is shown and is from the standard database.
[0106] This is the time scaling factor. , This represents the characteristic time to reach 50% saturation. This factor normalizes the time axis and eliminates the influence of overall differences in diffusion rates.
[0107] Furthermore, in In the curve, when High-frequency bias test curve, when At low levels, it tends to gravitate towards the reference curve because Capable of controlling correction intensity, high This indicates that the test curve is reliable and retains more original information; low This indicates a large deviation, requiring a reference curve. Ensure time alignment to avoid phase deviation.
[0108] Dynamic Time Warping (DTW) can be used to compare the similarity of time series, but traditional DTW often ignores the trend of change, making it difficult to capture the similarity of diffusion rates. Therefore, a standardized saturation curve is used. Calculate the cost of improved dynamic time warping:
[0109] ;
[0110] in, and This indicates the standardized test curve and the reference curve at time points. and The value of .
[0111] The gradient curve, calculated through numerical differentiation (such as central difference), represents the instantaneous diffusion rate.
[0112] The gradient weight coefficient can be taken as an empirical value of 0.4. The balance value and gradient contribution are determined by optimization based on historical data.
[0113] For the optional regular path, the minimum cumulative cost can be solved by dynamic programming.
[0114] The smaller the value, the higher the similarity. During the calculation process, the gradient term ensures that not only the values match, but the trends of change are also consistent. This is because the value difference captures the overall absorption, while the gradient difference captures the diffusion dynamics. For example, even if the values are similar, a large gradient difference may indicate that the diffusion is unstable.
[0115] However, during quality assessment, the small deviations between different values may lead to false positives and false negatives. Therefore, exponential and hyperbolic functions are used to enhance the sensitivity to deviations and construct a final quality index based on the integrity of the diffusion process.
[0116] ;
[0117] in, This is the coordination coefficient.
[0118] Indicates the use of e An exponential function with base 0 converts similarity cost into similarity score, with a range of (0,1]. The smaller the cost, the higher the score.
[0119] This indicates the relative deviation in the anisotropy ratio, ensuring... near .
[0120] This indicates that shape differences are handled based on the hyperbolic tangent function. This is the critical value, representing the critical shape difference. An empirical value of 1.2 can be taken. near At that time, the value dropped rapidly.
[0121] and then, The value range is [0,1], and the higher the value, the better the quality.
[0122] Furthermore, it can be based on The value is used to determine the final quality, and a QR threshold is set. If the value is displayed correctly, it indicates that the disposable hygiene product is a qualified product; otherwise, it is a substandard product. An empirical value of 0.8 is recommended.
[0123] It should be noted that those skilled in the art can make various modifications and improvements without departing from the inventive concept, and these all fall within the scope of protection of this invention. Therefore, the scope of protection of this patent should be determined by the appended claims.
Claims
1. A method for measuring the spreadability of a disposable hygiene article, characterized in that, include: Acquire the timing signals of multiple discrete coordinates at multiple time points inside the core of the sanitary product under test; Based on the timing signal, the time difference of liquid arrival between any two sensors is calculated, and combined with the spatial distance between the two sensors, a pair of sensor activation speeds characterizing the local diffusion rate is constructed. For each of the aforementioned sensors, the distribution of activation velocities of the paired sensors within its neighborhood is summarized, and a local diffusion tensor is constructed through statistical analysis to characterize the local diffusion anisotropy. The method for constructing the local diffusion tensor includes: The activation velocities of multiple pairs of sensors within the neighborhood are represented as velocity vectors; Calculate the covariance matrix of the velocity vector; The covariance matrix is subjected to eigenvalue decomposition, and the direction of the obtained eigenvector is taken as the main direction of local diffusion, and the eigenvalue is taken as the diffusion intensity in that direction. Based on the local diffusion tensor field composed of all the aforementioned sensors, the anisotropic effective distance between the interpolation point and each of the aforementioned sensors is calculated. The calculation of the anisotropic effective distance includes: The local diffusion tensor field is considered as a Riemannian metric space; In the metric space, solve the geodesic problem from the sensor to the interpolation point; The geodesic length is defined as the effective anisotropic distance; Based on the value of the time series signal and the effective distance of the anisotropy, an interpolation algorithm is used to reconstruct a continuous spatiotemporal saturation map. Quality assessment was conducted based on a continuous spatiotemporal saturation map, and the diffusion measurement results of disposable hygiene products were obtained.
2. The method of measuring the dispersibility of a disposable hygienic article according to claim 1, characterized by, The geodesic problem is obtained by solving the Eikonal equation.
3. The method of measuring the dispersibility of a disposable hygienic article according to claim 1, characterized by, The method further includes quality assessment based on the continuous spatiotemporal saturation map, specifically including: Diffusion features are extracted from the map, including the shape of the diffusion front, the anisotropy ratio, and the curve of saturation changing over time. The diffusion characteristics are matched with pre-stored standard sample feature maps; Output quality assessment results based on the matching degree.
4. The method of measuring the dispersibility of a disposable hygienic article according to claim 3, characterized in that, The matching process uses a dynamic time warping algorithm to calculate the similarity between the curve and the standard curve.
5. The method of measuring the dispersibility of a disposable hygienic article according to claim 1, characterized by, The quality assessment includes: calculating a global diffusion anisotropy index based on the local diffusion tensor of all sensors; comparing the index with a preset threshold; and determining the product as unqualified when the index exceeds the threshold range.
6. A system for measuring the spread of a disposable hygiene product, characterized in that include: A processor and a memory, the memory storing computer program instructions that, when executed by the processor, implement the method for measuring the spread of disposable hygiene products according to any one of claims 1-5.
Citation Information
Patent Citations
Test method for absorption rate, diffusion length, leakage amount and reverse osmosis amount of absorbent sanitary products
CN109142182A
Mathematical model modeling method based on data analysis operation processing
CN120893334A