A method and system for analyzing sleep pillow pressure distribution data

Abstract

The present application relates to the field of sleep health monitoring and textile ergonomics evaluation, in particular to a sleep pillow pressure distribution data analysis method and system, comprising: collecting multi-point pressure distribution data on a sleep pillow, and preprocessing and fusing the multi-point pressure distribution data to generate a two-dimensional pressure distribution matrix; adopting an adaptive clustering method based on a contour coefficient to determine an optimal clustering number K, constructing an improved Euclidean distance as a similarity measure by taking pressure values as weights, executing a pressure weighted K-Means clustering algorithm, and automatically segmenting the two-dimensional pressure distribution matrix into K pressure-bearing areas with physiological significance; adopting a fuzzy C-means clustering algorithm to perform secondary processing on the segmentation result, and outputting a membership probability distribution of each pressure measuring point belonging to each pressure-bearing area. The present application solves the problem of presetting the number of areas in the existing sleep pillow pressure distribution data analysis method.

Description

Technical Field

[0001] This invention relates to the fields of sleep health monitoring and textile ergonomics evaluation, and in particular to a method and system for analyzing pillow pressure distribution data. Background Technology

[0002] As a crucial bedding item for maintaining the normal physiological curvature of the head and neck during sleep, the support performance of a pillow directly affects sleep quality and cervical spine health. By collecting and analyzing pressure distribution data on the pillow surface, ergonomic indicators such as peak pressure, average pressure, and pressure gradient at the contact interface between the human body and the pillow can be objectively assessed. This provides data support for pillow structure optimization, personalized selection, and pressure sore risk warning. Pillow pressure distribution data analysis methods have become a research hotspot in the fields of sleep health monitoring and textile ergonomic evaluation.

[0003] Common methods for analyzing pillow pressure distribution data primarily rely on clustering and partitioning techniques. This involves automatically dividing the entire pressure distribution map into several physically meaningful pressure-bearing regions based on the spatial proximity and pressure value similarity of pressure measurement points, and then extracting feature parameters from each region to evaluate its support performance. Existing technologies generally employ traditional K-Means clustering algorithms or hierarchical clustering methods based on Euclidean distance for partitioning. These methods typically require users to pre-define the number of cluster regions and use standard Euclidean distance as a similarity measure between measurement points. Furthermore, after region partitioning, existing methods often employ a hard boundary partitioning strategy, meaning each pressure measurement point is uniquely assigned to a specific region, with no overlap or ambiguity between regions.

[0004] Pre-setting the number of clustering regions relies heavily on human experience. The optimal number of pressure-bearing regions varies depending on the pillow structure and individual body characteristics; a fixed number cannot be adaptively adjusted, easily leading to over-segmentation or under-segmentation. Standard Euclidean distance only considers the spatial coordinate difference or pressure value difference between measuring points, failing to incorporate the pressure value itself as a weight in the distance metric. This distorts the similarity representation between high-pressure and low-pressure areas, affecting the physiological rationality of the zoning results. Hard boundary division ignores the fuzzy characteristics of regional transition zones in the pressure distribution map. Measuring points at the boundaries are often simultaneously influenced by adjacent regions; forcibly assigning them to a particular region results in the loss of important boundary information, leading to deviations in subsequently extracted peak pressure, pressure gradient, and other feature parameters, ultimately affecting the accuracy and robustness of the pillow's support performance evaluation. Summary of the Invention

[0005] To address the problems in existing methods for analyzing pillow pressure distribution data, such as the need to pre-determine the number of regions for clustering and partitioning, the lack of consideration for pressure value weights in distance metrics, and the inaccurate segmentation of pressure-bearing areas due to rigid partition boundaries, this invention provides a method and system for analyzing pillow pressure distribution data.

[0006] The technical solution adopted by this invention to solve its technical problem is: a method for analyzing pillow pressure distribution data. Preferably, S1: collect multi-point pressure distribution data on the pillow, and preprocess and fuse the multi-point pressure distribution data to generate a two-dimensional pressure distribution matrix; S2: use an adaptive clustering method based on silhouette coefficients to determine the optimal number of clusters K, construct an improved Euclidean distance as a similarity measure using pressure values ​​as weights, and execute a pressure-weighted K-Means clustering algorithm to automatically divide the two-dimensional pressure distribution matrix into K physiologically significant pressure-bearing areas; S3: use fuzzy C... The mean clustering algorithm performs secondary processing on the automatic segmentation results of the two-dimensional pressure distribution matrix, outputting the probability distribution of the membership degree of each pressure measurement point to each pressure-bearing area, so as to retain the fuzzy information of the area boundary; S4: For each pressure-bearing area, five feature parameters are extracted respectively: peak pressure, average pressure, pressure gradient, pressure concentration and comfort contact ratio, to construct a five-dimensional pressure morphology feature vector; S5: The five-dimensional pressure morphology feature vector is input into the pre-constructed fuzzy comprehensive evaluation model to calculate the comprehensive support performance score, and outputs the support performance level of the pillow according to the preset level threshold.

[0007] Preferably, in step S2, an improved Euclidean distance is constructed using pressure values ​​as weights, and the distance between any two pressure measurement points i and j is calculated according to the following formula: ; in, and These represent the measuring points. and measuring points Pressure value, , and , Representing the measuring points respectively and measuring points Spatial coordinates on the plane of the pillow.

[0008] Preferably, in step S2, an adaptive method based on the silhouette coefficient is used to determine the optimal number of clusters. Specifically, this includes the number of candidate clusters. The profile coefficient s(i) for each pressure measuring point i is calculated as follows: ; in, For measuring points The average distance to all other measurement points within its cluster. For measuring points The average distance to all measuring points within its nearest neighbor cluster; the overall profile coefficient is the arithmetic mean of the profile coefficients of all measuring points, calculated using the following formula: ; in The total number of pressure measurement points is within the preset range of candidate cluster numbers, K= Within, select the one that maximizes the overall profile coefficient. The value is used as the optimal cluster number K.

[0009] Preferably, the formula for calculating the pressure concentration is: ; in This represents the total number of pressure measuring points within the current pressure-bearing area. For the first in the region Pressure values ​​at each measuring point Peak pressure within the region; preset comfort threshold. The formula for calculating the comfort contact ratio is: ; in, If the pressure value in the current pressure zone is lower than the preset comfort threshold The number of measuring points, The total number of measuring points in the area; the preset comfort threshold The value range is 30-40 mmHg.

[0010] Preferably, the fuzzy comprehensive evaluation model in S5 is constructed by including: determining the set of evaluation factors as the five-dimensional pressure morphology feature vector. Determine the set of comments , The number of performance levels is determined; a trapezoidal membership function is used to determine the membership degree of each evaluation factor to each rating level, forming a membership matrix. ,in Indicates the first The evaluation factor belongs to the first... Membership degree of each rating level; weight vector of each evaluation factor determined based on analytic hierarchy process (AHP). ,satisfy ,in For the first The weight of each evaluation factor; The comprehensive evaluation results are calculated using fuzzy synthesis operators: ; in For fuzzy synthesis operators, This is the comprehensive score vector. The evaluation result belongs to the first A comprehensive score for each rating level; based on the principle of maximum membership or the weighted average method. Defuzzify and output the final support performance level; synthesize the weight vector and membership matrix to obtain the comprehensive support performance score.

[0011] On the other hand, a pillow pressure distribution data analysis system includes: a data acquisition and fusion module for acquiring multi-point pressure distribution data on the pillow, preprocessing and fusing it, and outputting a two-dimensional pressure distribution matrix; an adaptive weighted clustering and partitioning module for adaptively determining the optimal number of clusters K based on the silhouette coefficient, and performing pressure-weighted K-Means clustering using an improved Euclidean distance with pressure values ​​as the weight, dividing the pressure distribution map into K pressure-bearing regions; a fuzzy boundary partitioning module for processing the boundaries of the pressure-bearing regions using a fuzzy C-means clustering algorithm, and outputting the probability distribution of the membership degree of each pressure measurement point to each pressure-bearing region; a feature extraction module for extracting the peak pressure, average pressure, pressure gradient, pressure concentration, and comfort contact ratio of each pressure-bearing region, and constructing a five-dimensional pressure morphology feature vector; and a comprehensive evaluation module, which embeds a fuzzy comprehensive evaluation model for calculating and outputting the pillow's support performance level based on the input five-dimensional pressure morphology feature vector.

[0012] The beneficial effects of this invention are as follows: It automatically determines the optimal number of clusters using an adaptive method based on contour coefficients, eliminating the need for manual pre-setting and avoiding over- or under-segmentation. By introducing an improved Euclidean distance weighted by pressure values ​​as a similarity metric, the clustering sensitivity of high-pressure areas is enhanced, and the partitioning results better conform to the physiological pressure distribution pattern. Furthermore, fuzzy C-means clustering is used to output the membership probability of measurement points to each region, preserving fuzzy information in boundary transition zones and overcoming feature extraction bias caused by rigid partitioning. Based on this, peak pressure, average pressure, pressure gradient, pressure concentration, and comfortable contact ratio are extracted to construct a five-dimensional pressure morphology feature vector, which is then combined with a fuzzy comprehensive evaluation model to output the pillow's support performance level, significantly improving the comprehensiveness and accuracy of the evaluation. This invention provides reliable technical support for pillow comfort assessment, pressure ulcer risk warning, and personalized bedding design. Attached Figure Description

[0013] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary, and those skilled in the art can derive other embodiments based on the provided drawings without creative effort.

[0014] Figure 1 This is a schematic diagram of the optimal embodiment of the pillow pressure distribution data analysis method and system of the present invention. Detailed Implementation

[0015] 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.

[0016] like Figure 1 As shown, this invention provides a method and system for analyzing pillow pressure distribution data. This method and system address the problems in existing technologies where clustering partitioning requires a preset number of regions, distance metrics do not consider pressure value weights, and rigid partition boundaries lead to inaccurate segmentation of pressure-bearing areas and an inability to retain ambiguous boundary information. The method described in this invention will be explained in detail below.

[0017] Example 1 A method for analyzing pillow pressure distribution data includes: S1: Collect multi-point pressure distribution data on the pillow, and preprocess and fuse the multi-point pressure distribution data to generate a two-dimensional pressure distribution matrix; First, multi-point pressure distribution data is collected on the pillow. In practice, sensors are attached to the head, neck, and shoulder areas of the body, and the body is placed on the pillow. Pressure distribution data at the body-pillow interface is collected using a pressure sensor array, such as a dot matrix composed of multiple capacitive or resistive pressure sensors. The spatial spacing between sensors is set according to the measurement accuracy requirements, preferably 10 to 20 millimeters between the centers of adjacent sensors. The sensor array collects the pressure values ​​at the contact interface between the body's head and neck and the pillow in real time. The raw data for each pressure measurement point includes the spatial coordinates of that point and the pressure value measured at that point. The collected raw data contains noise or outliers, so preprocessing is performed. Preprocessing operations include, but are not limited to, median filtering to remove salt-and-pepper noise and using linear interpolation or spline interpolation to fill in missing data caused by sensor failure. After single-point preprocessing, the data from all measurement points are fused according to their spatial coordinates to generate a two-dimensional pressure distribution matrix. The row and column indices of this matrix correspond to the spatial locations of the pressure measurement points, and the element values ​​in the matrix are the pressure values ​​at the corresponding locations. This two-dimensional pressure distribution matrix fully and accurately reflects the overall pressure distribution at the interface between the pillow and the human body.

[0018] S2: The optimal number of clusters K is determined by an adaptive clustering method based on the silhouette coefficient. An improved Euclidean distance is constructed with pressure value as the weight as a similarity measure. The pressure-weighted K-Means clustering algorithm is executed to automatically divide the two-dimensional pressure distribution matrix into K pressure-bearing areas with physiological significance. This step first uses an adaptive method based on the silhouette coefficient to determine the optimal number of clusters K. A reasonable range for the number of candidate clusters is set, for example, from a minimum of 2 to a maximum of 10. This range is adaptively adjusted according to the total number of pressure measuring points, and generally the maximum value does not exceed the square root of the total number of measuring points. For each integer K within the candidate range, standard K-Means clustering is performed once to obtain preliminary clustering results. Then, the silhouette coefficient of each pressure measuring point is calculated. For any pressure measuring point i, its silhouette coefficient si is calculated according to the following formula: ; in, For measuring points The average distance between the two points and all other measuring points within the same cluster. The average distance is calculated using Euclidean distance, which is the straight-line distance between the two measuring points in spatial coordinates. For measuring points The average distance to all measurement points within its nearest neighbor cluster. The nearest neighbor cluster refers to the cluster excluding the measurement point. The cluster with the smallest average distance to the measurement point, outside of its own cluster. Profile coefficient. The value is between negative one and positive one. When When it is close to positive one, it indicates that the measuring point Well-clustered, far from neighboring clusters; when When it approaches zero, it indicates that the measuring point Located on the boundary between two clusters; when When it is negative, it indicates the measuring point. It may have been incorrectly assigned to the current cluster.

[0019] After calculating the profile coefficients of all pressure measurement points, the overall profile coefficient for the current candidate cluster number K is calculated. The overall profile coefficient is the arithmetic mean of the profile coefficients of all N pressure measurement points, and the calculation formula is as follows: ; Where N is the total number of pressure measuring points, and the summation symbol indicates that for each measuring point from the first to the Nth measuring point... Accumulate the results. Within the preset range of candidate cluster numbers K, belong to the set... Calculate each one separately. The overall profile coefficient corresponding to the value is selected based on the value that maximizes the overall profile coefficient. The value is used as the optimal number of clusters K. This achieves adaptive determination of the number of clusters without manual intervention.

[0020] After determining the optimal number of clusters K, this invention executes the pressure-weighted K-Means clustering algorithm. Traditional K-Means clustering uses standard Euclidean distance as a similarity measure, calculating only the square root of the sum of the squares of the spatial coordinate differences between two measurement points, ignoring the influence of the pressure value itself on the partitioning. Therefore, this invention proposes an improved Euclidean distance that uses pressure value as the weight. For any two pressure measurement points... and The weighted distance between them is calculated using the following formula: ; in, and Representing the measuring points respectively and measuring points The pressure value, whose physical dimension is the pressure unit, is millimeters of mercury or pascals in practical applications. , and , Representing the measuring points respectively and measuring points The spatial coordinates on the pillow plane are expressed in units of length, such as millimeters or centimeters. The physical meaning of this formula is that the weight of the spatial distance between two measuring points is modulated by the arithmetic mean of the pressure values ​​at the two points. When both measuring points are in high-pressure areas, their average pressure is large, and the weighted distance is amplified. This means that measuring points in high-pressure areas have a higher requirement for spatial proximity and are more likely to cluster together. Conversely, measuring points in low-pressure areas have a smaller weighted distance, and the influence of spatial dispersion is relatively reduced.

[0021] As an equivalent alternative, the following formula can also be used: ; The equivalent formula uses the geometric mean of pressure values ​​as weights. Both methods can achieve the weighting effect of pressure values ​​on distance metrics. Using the improved Euclidean distance as the similarity metric, the standard K-Means clustering algorithm is executed: first, K cluster centers are randomly initialized; then, measurement point allocation and cluster center updates are iteratively performed until the cluster centers no longer change significantly or the preset iteration limit is reached. After clustering, the original two-dimensional pressure distribution matrix is ​​automatically divided into K pressure-bearing regions with clear physiological significance, such as the head-pillow region and neck-pillow region.

[0022] S3: The fuzzy C-means clustering algorithm is used to perform secondary processing on the automatic segmentation results of the two-dimensional pressure distribution matrix, and outputs the membership probability distribution of each pressure measuring point to each pressure-bearing area, so as to preserve the fuzzy information of the area boundary; The pressure-weighted K-Means clustering in the second step described above is a hard clustering algorithm. Its output is that each pressure point is uniquely assigned to a pressure-bearing region, and the boundaries between regions are distinct. However, in actual pillow pressure distribution, there is usually a transition zone between two pressure-bearing regions. Pressure points within this transition zone are simultaneously influenced by two or more adjacent regions, making their physiological significance ambiguous. To overcome this problem, this invention uses a fuzzy C-means clustering algorithm to further process the segmentation results of the second step. Based on the clustering results of the second step, each pressure-bearing region is considered as an initial fuzzy cluster, and the association of each pressure point with that region is transformed from an absolute true / false value to a probability value. Specifically, the fuzzy C-means clustering algorithm achieves this by optimizing the objective function: minimizing the sum of squared weighted distances from all pressure points to all cluster centers, where the weight is the m-th power of the membership degree of the pressure point to that cluster center, and m is the fuzzy weighting exponent, typically taken as 2. The algorithm iteratively calculates each cluster center and the membership degree of each pressure point to each cluster center. Finally, for each pressure measurement point, the algorithm outputs a probability distribution vector of membership degrees among K pressure-bearing regions. Each component of this vector represents the degree to which the measurement point belongs to the corresponding pressure-bearing region, and the sum of all components is 1. For example, a measurement point located on the boundary between the neck-occipital region and the head-occipital region might have a membership vector with a probability of 0.6 belonging to the neck-occipital region, a probability of 0.4 belonging to the head-occipital region, and a probability of 0 belonging to other regions. In this way, the originally rigid region boundaries are transformed into a smooth, fuzzy transition zone, fully preserving the fuzzy information of the boundary regions.

[0023] S4: For each pressure zone, extract five feature parameters: peak pressure, average pressure, pressure gradient, pressure concentration, and comfort contact ratio, and construct a five-dimensional pressure morphology feature vector. After completing the fuzzy boundary partitioning, five key feature parameters are extracted for each identified pressure-bearing area.

[0024] The first characteristic parameter is peak pressure. Peak pressure is defined as the maximum pressure value among all pressure measuring points within the pressure-bearing area. It reflects the maximum pressure exerted on the area and is a core indicator for assessing the risk of localized pressure overload and pressure sores. Peak pressure equals the pressure values ​​at all measuring points within the area. The maximum value, expressed by the symbol express.

[0025] The second characteristic parameter is the average pressure. Average pressure is defined as the arithmetic mean of the pressure values ​​at all pressure measuring points within the pressure-bearing area. It reflects the overall pressure level experienced by the area.

[0026] The third characteristic parameter is the pressure gradient. The pressure gradient describes the degree of pressure variation with spatial location within a pressure-bearing region. For each measuring point within the region, the pressure difference between it and its adjacent measuring points in both the horizontal and vertical directions is calculated. Then, the rate of change in these directions is obtained. Finally, the root mean square of the rates of change for all measuring points, or the maximum value, is taken as the pressure gradient index for the entire region. A larger pressure gradient indicates more drastic pressure changes and a higher risk of shear stress on human tissue.

[0027] The fourth characteristic parameter is pressure concentration. The formula for calculating pressure concentration provided in this invention is as follows: ; in This represents the total number of pressure measuring points within the current pressure-bearing area. For the first in the region Pressure values ​​at each measuring point The peak pressure within the region is the denominator of the formula. The numerator is the root mean square (RMS) of the pressure values ​​within the region, and the denominator is the peak pressure. The RMS value is more sensitive to larger pressure values. When the pressure distribution within the region is highly uneven, i.e., a few measuring points have extremely high pressure while most measuring points have low pressure, the RMS will be significantly smaller than the peak pressure, resulting in low pressure concentration. Conversely, when the pressure distribution within the region is relatively uniform, the RMS is close to the peak pressure, and the pressure concentration is close to the value of 1. Therefore, pressure concentration measures the degree of pressure accumulation within a region; the smaller the value, the more concentrated the pressure and the higher the risk of local overload.

[0028] The fifth characteristic parameter is the comfort contact ratio. This invention presets a comfort pressure threshold. This threshold has a clear physiological basis and is preferably the capillary closure threshold pressure in the human body. Physiological studies have shown that when external pressure consistently exceeds the capillary closure pressure, it hinders tissue microcirculation, leading to ischemia and hypoxia, and consequently causing pressure ulcers. The capillary closure pressure in the human body is typically between 30 mmHg and 40 mmHg. Therefore, a preset comfort threshold is necessary. The value ranges from 30 mmHg to 40 mmHg, with a typical preferred value of 32 mmHg. The formula for calculating the comfort contact ratio is as follows: ; in, If the pressure value in the current pressure zone is lower than the preset comfort threshold The number of measuring points, This represents the total number of measuring points within the area. This ratio indicates the percentage of the area within the pressure zone where the pressure is within a safe range. The closer the value is to 1, the better the contact comfort and the lower the risk of pressure sores.

[0029] The five characteristic parameters mentioned above—peak pressure, average pressure, pressure gradient, pressure concentration, and comfort contact ratio—are combined in sequence to form a five-dimensional pressure morphology feature vector. ,in Represents peak pressure. Represents average pressure. Represents the pressure gradient. Represents the degree of pressure concentration. This represents the comfort contact ratio. This vector comprehensively quantifies the morphological and mechanical properties of a pressure zone from multiple dimensions.

[0030] S5: Input the five-dimensional pressure morphology feature vector into the pre-constructed fuzzy comprehensive evaluation model, calculate the comprehensive score of support performance, and output the support performance level of the pillow according to the preset level threshold. The five-dimensional pressure morphology feature vector U constructed in the fourth step is input into a pre-constructed fuzzy comprehensive evaluation model. The model automatically calculates the comprehensive score of the pillow's support performance and outputs the final support performance level according to the preset level threshold, such as setting four levels: excellent, good, medium and poor.

[0031] The construction process of this fuzzy comprehensive evaluation model includes the following sub-steps: Determine the evaluation factor set. The evaluation factor set is the set of all indicators involved in the evaluation; in this invention, the evaluation factor set is the five-dimensional pressure morphology feature vector. ,Include to There are five evaluation factors in total.

[0032] Determine the set of comments. The set of comments is the set of possible values ​​for the evaluation results, denoted as . , .in, This represents the total number of performance levels. When equal to 4, set Excellent Good, For the middle, It is poor.

[0033] Establish the membership function and calculate the membership matrix. For each evaluation factor, such as factor μi, define its membership function relative to each level in the comment set. Define the membership function. For the... The evaluation factor for the first Membership of each rating level To form a membership matrix ; Determine the weight vector. Different evaluation factors have varying degrees of influence on the final support performance evaluation result. This invention uses the analytic hierarchy process (AHP) or entropy weight method to determine the weight of each evaluation factor. Taking the AHP as an example, firstly, a 5x5 pairwise comparison judgment matrix is ​​constructed, and the relative importance between each factor is assessed through expert experience. For example, peak pressure and comfort contact ratio are the most important for assessing pressure ulcer risk and are given higher weights; pressure gradient is second; mean pressure and pressure concentration are third. Then, the eigenvector of this judgment matrix is ​​calculated, and a consistency test is performed. After passing the test, the normalized eigenvector obtained is the weight vector. Weight vector ,in For the first The weights of each evaluation factor are such that the sum of all five weights equals 1, which satisfies the condition... ; Fuzzy synthesis and defuzzification. The above weight vector... With membership matrix Perform a synthesis operation to obtain a comprehensive evaluation result. The mathematical expression for composition is: ; in For the fuzzy synthesis operator, this invention preferably employs a weighted average synthesis operator, i.e., matrix multiplication. Specifically, the calculation is as follows: [The text abruptly ends here, so the translation stops as well.] The first in element equal to from Equal to 1 to 5 Multiply The sum is calculated using the following formula: ; This operator comprehensively considers the influence of all factors, minimizing information loss. The result of the synthesis operation is a 1-row, L-column composite score vector. ,Right now ,in The evaluation result belongs to the first The overall score for each rating level is a value between 0 and 1.

[0034] Finally, the final support performance level is output according to the defuzzification principle. Defuzzification is the process of transforming a fuzzy evaluation vector into a clear evaluation result. This invention employs one of two methods: the first is the maximum membership principle, which involves finding the comprehensive score vector. The component with the largest median value corresponds to the final support performance rating. For example, if... Given the value [0.1, 0.7, 0.15, 0.05], the maximum value is 0.7, corresponding to the second level. If the rating is "Good", then output "Good". The second method is a weighted average method, which quantifies each rating level into a numerical value, for example, Excellent is 100 points, Good is 75 points, Average is 50 points, and Poor is 25 points. Then, the overall score vector is calculated. A weighted average is calculated to obtain a final score, which is then mapped to a preset level threshold range to output the level. This completes the automatic, quantitative, and comprehensive evaluation of the pillow's support performance level.

[0035] Example 2 A pillow pressure distribution data analysis system, the system being equipped with a pillow pressure distribution data analysis method, comprising: Data acquisition and fusion module; this module is used to perform the first step of the above method. Its function is to collect pressure distribution data at multiple points on the pillow, and perform preprocessing operations such as median filtering and interpolation on the collected raw data, and finally fuse all the measurement point data and output a complete two-dimensional pressure distribution matrix.

[0036] The adaptive weighted clustering partitioning module is used to execute the second step of the above method. This module further includes an adaptive clustering number unit. The specific operation of this adaptive clustering number unit is as follows: iterates through a preset candidate range of clustering numbers, for example, from 2 to 10. For each candidate clustering number, it calls the pressure-weighted K-Means clustering algorithm for preliminary clustering; then it calculates the profile coefficients of all pressure measurement points under that clustering number and calculates the arithmetic mean of these profile coefficients; finally, it selects the clustering number with the largest average as the optimal clustering number K. After determining the value of K, the module uses the improved Euclidean distance, weighted by pressure values, as the metric to perform pressure-weighted K-Means clustering, thereby automatically dividing the input two-dimensional pressure distribution map into K physiologically significant pressure-bearing regions.

[0037] The fuzzy boundary partitioning module performs the third step of the method described above. This module receives the hard clustering segmentation results from the adaptive weighted clustering partitioning module and then performs secondary processing on these results using the fuzzy C-means clustering algorithm. During this processing, the region boundaries are no longer abrupt hard boundaries but are transformed into a fuzzy transition zone. Ultimately, this module outputs a membership probability distribution for each pressure measurement point across all K pressure-bearing regions, fully preserving the fuzzy information of the boundary regions.

[0038] The feature extraction module performs the fourth step of the above method. Its function is to extract five feature parameters from each pressure-bearing region output by the fuzzy boundary partitioning module: peak pressure, average pressure, pressure gradient, pressure concentration, and comfort contact ratio. It is important to note that the pressure concentration in this module is explicitly defined as the ratio of the average of the squares of the pressure values ​​at each measuring point within the region to the peak pressure, i.e., the root mean square pressure divided by the peak pressure. The comfort contact ratio in this module is explicitly defined as the ratio of the number of measuring points within the region whose pressure values ​​are less than or equal to the human capillary closure threshold pressure to the total number of measuring points in the region. The human capillary closure threshold pressure is a physiological constant, typically taken as 32 mmHg. The feature extraction module combines the five extracted parameters into a five-dimensional pressure morphology feature vector and outputs it.

[0039] The comprehensive evaluation module embeds a fuzzy comprehensive evaluation model. This module is used to execute the fifth step of the above method. It receives a five-dimensional pressure morphology feature vector from the feature extraction module as input, calls the internally constructed evaluation factor set, comment set, trapezoidal membership function, weight vector, and fuzzy synthesis operator to automatically calculate the comprehensive score of support performance, and outputs a clear and accurate pillow support performance level (e.g., excellent, good, average, poor) based on a preset level mapping relationship, as the final result of the entire system output.

[0040] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. The above descriptions are only preferred embodiments of this application. It should be noted that due to the limitations of written expression, while there are objectively infinite specific structures, those skilled in the art can make several improvements, modifications, or changes without departing from the principles of this application, and can also combine the above technical features in an appropriate manner. These improvements, modifications, changes, or combinations, or the direct application of the inventive concept and technical solution to other situations without modification, should all be considered within the scope of protection of this application.

Claims

1. A method for analyzing the pressure distribution data of a sleeping pillow, characterized in that, include: S1: Collect multi-point pressure distribution data on the pillow, and preprocess and fuse the multi-point pressure distribution data to generate a two-dimensional pressure distribution matrix; S2: The optimal number of clusters K is determined by an adaptive clustering method based on the silhouette coefficient. An improved Euclidean distance is constructed with pressure value as the weight as a similarity measure. The pressure-weighted K-Means clustering algorithm is executed to automatically divide the two-dimensional pressure distribution matrix into K pressure-bearing areas with physiological significance. S3: The fuzzy C-means clustering algorithm is used to perform secondary processing on the automatic segmentation results of the two-dimensional pressure distribution matrix, and outputs the membership probability distribution of each pressure measuring point to each pressure-bearing area, so as to preserve the fuzzy information of the area boundary; S4: For each pressure zone, extract five feature parameters: peak pressure, average pressure, pressure gradient, pressure concentration, and comfort contact ratio, and construct a five-dimensional pressure morphology feature vector. S5: Input the five-dimensional pressure morphology feature vector into the pre-constructed fuzzy comprehensive evaluation model, calculate the comprehensive score of support performance, and output the support performance level of the pillow according to the preset level threshold.

2. The method for analyzing pillow pressure distribution data according to claim 1, characterized in that, In S2, an improved Euclidean distance is constructed using pressure values ​​as weights. The distance between any two pressure measurement points i and j is calculated using the following formula: ； in, and These represent the measuring points. and measuring points Pressure value, , and , Representing the measuring points respectively and measuring points Spatial coordinates on the plane of the pillow.

3. The method for analyzing pillow pressure distribution data according to claim 1, characterized in that, In step S2, an adaptive method based on the silhouette coefficient is used to determine the optimal number of clusters. ; Specifically, this includes the number of candidate clusters. The profile coefficient s(i) for each pressure measuring point i is calculated as follows: ； in, For measuring points The average distance to all other measurement points within its cluster. For measuring points The average distance between it and all measurement points in its nearest neighbor cluster; The overall profile coefficient is the arithmetic mean of the profile coefficients of all measuring points. The calculation formula is: ； in The total number of pressure measurement points is within the preset range of candidate cluster numbers, K= Within, select the one that maximizes the overall profile coefficient. The value is used as the optimal cluster number K.

4. The method for analyzing pillow pressure distribution data according to claim 1, characterized in that, The S4 further includes: The formula for calculating the pressure concentration is: ； in This represents the total number of pressure measuring points within the current pressure-bearing area. For the first in the region Pressure values ​​at each measuring point This represents the peak pressure within the region. Preset comfort threshold The formula for calculating the comfort contact ratio is: ； in, If the pressure value in the current pressure zone is lower than the preset comfort threshold The number of measuring points, The total number of measuring points in the area; the preset comfort threshold The value range is 30-40 mmHg.

5. The method for analyzing pillow pressure distribution data according to claim 1, characterized in that, The fuzzy comprehensive evaluation model in S5 is constructed as follows: The evaluation factor set is determined as the five-dimensional pressure morphology feature vector. ; Determine the comment set , Support performance rating; The trapezoidal membership function is used to determine the membership degree of each evaluation factor to each rating level, forming a membership matrix. ,in Indicates the first The evaluation factor belongs to the first... Membership degree of each rating level; The weight vector of each evaluation factor is determined based on the analytic hierarchy process. ,satisfy ,in For the first The weight of each evaluation factor; The comprehensive evaluation results are calculated using fuzzy synthesis operators: ； in For fuzzy synthesis operators, This is the comprehensive score vector. The evaluation result belongs to the first A comprehensive score based on each rating level; Based on the principle of maximum membership or the weighted average method, Defuzzify and output the final support performance level; The weight vector and membership matrix are combined to obtain the overall support performance score.

6. A pillow pressure distribution data analysis system, equipped with the pillow pressure distribution data analysis method according to any one of claims 1-5, characterized in that, include: The data acquisition and fusion module is used to collect multi-point pressure distribution data on the pillow, perform preprocessing and fusion, and output a two-dimensional pressure distribution matrix. The adaptive weighted clustering partitioning module is used to adaptively determine the optimal number of clusters K based on the silhouette coefficient, and perform pressure-weighted K-Means clustering using the improved Euclidean distance with pressure value as the metric to divide the pressure distribution map into K pressure-bearing regions. The fuzzy boundary partitioning module is used to process the boundary of the pressure-bearing area using the fuzzy C-means clustering algorithm and output the probability distribution of the membership degree of each pressure measuring point to each pressure-bearing area. The feature extraction module is used to extract the peak pressure, average pressure, pressure gradient, pressure concentration and comfort contact ratio of each pressure zone, and construct a five-dimensional pressure morphology feature vector. The comprehensive evaluation module, which incorporates a fuzzy comprehensive evaluation model, is used to calculate and output the support performance level of the pillow based on the input five-dimensional pressure morphology feature vector.

7. A pillow pressure distribution data analysis system according to claim 6, characterized in that, Also includes: The adaptive weighted clustering partitioning module further includes a cluster number adaptive unit, which is used to traverse a preset cluster number candidate range, calculate the profile coefficient of all measurement points under each cluster number, and select the cluster number with the largest average profile coefficient as the optimal cluster number K.

8. A pillow pressure distribution data analysis system according to claim 6, characterized in that, Also includes: The pressure concentration in the feature extraction module is defined as the ratio of the average of the sum of squares of the pressure values ​​at each measuring point within the region to the peak pressure. The comfort contact ratio is defined as the ratio of the number of measuring points in the area whose pressure values ​​are less than or equal to the human capillary closure threshold pressure to the total number of measuring points in the area.

Images