Stretching fabric strain sensor layout design-oriented system

Abstract

The invention discloses a stretching fabric strain sensor layout design-oriented system, which comprises a fabric sensor layout algorithm, which comprises the following steps of: processing a human body motion data set, and calculating a stretch rate of an edge to represent skin deformation in motion; according to the skin deformation data and four feature edge selection methods including a stretching extreme value method, a variance method, an auto-covariance and correlation coefficient method and a composite standard method, feature edges capable of representing skin deformation caused by movement are screened, and a stretching hot area is formed; through a K-Means clustering algorithm, classifying the stretched hot areas according to an expected number of sensors, and generating a virtual sensor in each cluster; and defining the coverage rate as a main objective function and the redundancy as an auxiliary objective function, optimizing the sensor layout to maximize the coverage rate of the sensor to the stretched hot area, and finally outputting a fabric sensor layout scheme. According to the method, the layout precision is remarkably improved, the layout design cost is greatly reduced, and the design iteration efficiency is greatly improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of fabric sensors, and particularly relates to a system for layout design of stretch fabric strain sensor. BACKGROUND

[0002] Human motion capture technology can monitor and record human motion data in real time, and has been widely used in many fields. Among them, human motion capture based on fabric strain sensors has become a very promising research direction. Fabric sensors convert the stretching or compression deformation of the skin during movement into resistance changes through the piezoresistive properties of their conductive fiber structure. These resistance changes can be measured and analyzed to achieve the capture and monitoring of human motion. Fabric sensors have the advantages of comfort, flexibility and non-invasiveness in motion monitoring, making the motion capture process more natural and unconstrained, and suitable for continuous monitoring of body movement. Fabric strain sensors have important application value in joint angle prediction, posture estimation and action classification, and have great potential in rehabilitation medicine, sports and human-computer interaction.

[0003] Human motion is driven by muscle contraction to produce the skeletal system, resulting in stretching or compression of the surrounding skin tissue. When the fabric stretch sensor is in contact with the skin, it will change similarly due to the deformation of the skin, thereby achieving motion monitoring. In the layout design of the sensor, the skin deformation is often used to approximate the deformation of tight clothing. Due to the complex characteristics of the skin, including high elasticity, pre-stress and anisotropy, the deformation mode becomes complex and diverse, and the skin deformation model of people with different body characteristics is significantly different. This variability requires targeted sensor layout for different populations, application scenarios, etc. to avoid problems such as sensors failing to produce effective responses and sensor output exceeding the response range, in order to achieve accurate motion capture. Therefore, the layout position of the fabric stretch sensor is crucial to the results of human motion monitoring.

[0004] There are currently various layout methods for fabric sensors. Empirical layout methods mainly rely on researchers' past practices and personal intuition to determine sensor positions and quantities for specific scenarios and simple movements without complex computing devices, but lack systematicity and scientificity, making it difficult to cope with different populations and complex movements. Anatomical layout methods use human anatomical structures and movement physiology knowledge to place sensors at key muscle groups and related parts of skeletal joints, but this method requires a high level of professional knowledge. Layout methods based on optical motion capture systems can accurately measure human movement skin deformation to plan sensor layout, providing accurate data to optimize layout and detail capture, but are costly, complex to operate, require a high environment and depend on professionals, limiting their widespread application. The difficulties in these layout methods greatly limit the effectiveness of fabric sensor technology in movement monitoring, which in turn hinders its wider application and development. SUMMARY

[0005] In view of the deficiencies in the prior art, the purpose of the present application is to provide a system for layout design of stretch fabric strain sensors based on three-dimensional human point cloud data analysis, which realizes intelligent layout of sensor positions through calculation of skin deformation hot area, feature edge screening and K-Means clustering optimization. The system uses a modular architecture, integrating a parameterized customization interface, an automated computing engine and a visualization output module, providing an efficient and accurate sensor layout design tool for non-professionals. In order to achieve the above-mentioned purposes and other advantages according to the present application, a system for layout design of stretch fabric strain sensors is provided, comprising:

[0006] A fabric sensor layout algorithm generates a sensor layout scheme through data-driven and multiple parameters, specifically including the following steps:

[0007] Processing human movement data sets, calculating the stretch rate of the edge to represent skin deformation in motion;

[0008] Through skin deformation data, and through four feature edge selection methods of stretch extremum method, variance method, autocovariance and correlation coefficient method, and composite standard method, the feature edges that can represent the skin deformation caused by motion are screened to form a stretch hot area;

[0009] Through the K-Means clustering algorithm, the stretch hot area is classified by the expected number of sensors, and virtual sensors are generated in each cluster;

[0010] Defining coverage as the main objective function and redundancy as the auxiliary objective function, the sensor layout is optimized to maximize the coverage of the stretch hot area by the sensors, and finally the fabric sensor layout scheme is output.

[0011] Preferably, the processing human motion dataset specific steps are:

[0012] Creating an adjacency matrix of the adjacency information of the connection points and the point coordinate information;

[0013] Using the adjacency matrix to construct a point-edge matrix recording the coordinates and serial number information of the two end points of an edge;

[0014] Constructing a point-coordinate matrix corresponding to each action and part;

[0015] Using the point-edge matrix and the point-coordinate matrix to identify the change of the end points of each edge, calculate the edge length, and construct an edge-length matrix corresponding to each motion;

[0016] Using the edge-length matrix to calculate the stretch rate of each edge in each frame relative to the first frame, and constructing an edge-stretch rate matrix, which is used to calculate the skin deformation caused by the motion.

[0017] Preferably, the method of screening characteristic edges that can represent the skin deformation caused by the motion includes:

[0018] Stretch extreme value method: when the maximum stretch of an edge during the motion exceeds a self-defined threshold, the edge is included in the characteristic set;

[0019] Difference method: calculate the variance of the stretch rate of each edge during the entire duration of the action. The larger the variance of the stretch rate of an edge, the more significant its stretch change, and then the edge is selected as a characteristic edge;

[0020] Autocovariance and correlation coefficient method: calculate the autocovariance and correlation coefficient of each edge in different action sequences to evaluate its similarity in different actions. The edges with large autocovariance and small correlation coefficient can better distinguish the motion, and then the edges are selected as characteristic edges;

[0021] Composite standard: rank the edges according to the combination of stretch rate, variance, autocovariance and correlation coefficient to establish a comprehensive ranking index.

[0022] Preferably, the classification of the stretch hot zone by the K-Means clustering algorithm with an expected number of sensors and the generation of virtual sensors in each cluster specifically includes the following steps:

[0023] Divide the selected characteristic edges into clusters equal to the number of sensors by K-Means clustering;

[0024] Take the center of the characteristic edges as the center of the sensors, and generate virtual sensors according to the self-defined sensor length;

[0025] The direction of each sensor is determined by target optimization calculation, and finally the sensor layout scheme is generated.

[0026] Compared with the prior art, the present application has the beneficial effects that:

[0027] 1. Precise determination of sensor layout

[0028] The present application realizes scientific and precise positioning of sensor location through four-dimensional point cloud data analysis and multiple feature extraction algorithms. Based on high-precision three-dimensional human body point cloud data and time-series motion information, the system can accurately calculate the deformation pattern of the skin during the motion process. Specifically, the system calculates a point cloud matrix of 6890x3x200 based on the 40,000+ mesh points and 129 dynamic sequences of the D-Faust dataset, accurately capturing the skin deformation pattern. Four feature selection methods, including the stretching extreme value method, variance method, autocovariance and correlation coefficient method, and composite standard method, can adaptively match different joint types and motion patterns. For example, the stretching extreme value method can be used to improve the accuracy of single-degree-of-freedom joints, while the variance method can effectively suppress skin wrinkle interference for multi-degree-of-freedom joints. At the same time, the system supports parameterized customization based on gender, BMI, body part and motion pattern, ensuring that each layout scheme can accurately adapt to the target population.

[0029] 2. Reduce technical complexity and cost

[0030] The present application simplifies the workflow of traditional sensor layout design, transforming manual methods that rely on expensive equipment and complex operations into an automated solution based on a software platform. Through pre-computed and stored human body deformation models, the system does not need to conduct on-site testing and equipment calibration, eliminating the dependence on optical motion capture equipment. The digital design process supports batch processing and template reuse, with very low marginal cost. The entire layout design process can be completed with standard computing equipment, significantly reducing hardware investment. At the same time, the automation characteristics of the system reduce labor costs and time investment, making it affordable for small institutions and resource-limited scenarios to have professional-level sensor layout design services.

[0031] 3. Reduce dependence on professional knowledge

[0032] The present application transforms complex professional knowledge into standardized algorithms and intuitive operation interfaces, significantly reducing the use threshold. The system, through modular design, encapsulates professional knowledge in algorithms from multiple fields such as human anatomy, sensor technology and data analysis, allowing users to complete layout design without related professional background. The visual operation interface and parameterized design process enable non-professionals to quickly get started and obtain high-quality design results. The standardized operation process ensures the consistency and reliability of the design results, unaffected by the personal experience and knowledge level of the operator. This design concept greatly promotes the popularization and application of fabric sensor technology.

[0033] 4. Improve design efficiency

[0034] The present application greatly improves the efficiency of sensor layout design through algorithm automation and process optimization. The system realizes full-process automation from parameter setting to layout scheme generation, and users only need to complete a small amount of parameter selection to obtain a complete design scheme. Real-time 3D visualization preview function supports rapid iteration and optimization, and users can immediately view the effect of parameter adjustment and make fine adjustments. Batch design function enables the system to handle multiple design requirements simultaneously, suitable for large-scale application scenarios. Compared with traditional manual design methods, the present application shortens the design cycle from several weeks to several hours, with an efficiency improvement of two orders of magnitude. This high efficiency not only saves a lot of time and labor costs, but also makes rapid prototyping and agile design methods possible.

[0035] 5. High system stability and strong scalability

[0036] The present application adopts modular architecture design, and each functional module is relatively independent with standardized interfaces, ensuring the stability and maintainability of the system. The data processing module supports input data in multiple formats, with good compatibility and scalability. The layout algorithm uses a verified mathematical model, and the calculation result is stable and reliable with strong reproducibility. The system reserves interfaces for algorithm upgrade and function expansion, supporting integration and upgrade of future technologies. The standardized output format facilitates integration with other tools and platforms, supporting digital management throughout the process. The modular design concept and open system architecture ensure that the present application has long-term technical vitality and broad development prospects. BRIEF DESCRIPTION OF DRAWINGS

[0037] Figure 1 Figure 1 is a software tool core interface diagram of a system for stretch fabric strain sensor layout design according to the present application;

[0038] Figure 2 Figure 2 is a layout optimization algorithm flowchart of a system for stretch fabric strain sensor layout design according to the present application;

[0039] Figure 3 Figure 3 is a feature edge graph selected by the autocovariance and correlation coefficient method of a system for stretch fabric strain sensor layout design according to the present application;

[0040] Figure 4 Figure 4 is a sensor layout method flowchart based on K-Means clustering optimization of a system for stretch fabric strain sensor layout design according to the present application; (a) clustering feature edges into six groups (b) generating virtual sensors in each cluster (c) determining the layout scheme;

[0041] Figure 5 Figure 5 is a coverage and redundancy calculation schematic diagram of a sensor of a system for stretch fabric strain sensor layout design according to the present application;

[0042] Figure 6 UI interface diagram for selecting gender, body type, placement site, and motion parameters of the system for stretch fabric strain sensor layout design according to the present application;

[0043] Figure 7 UI interface diagram of the system for stretch fabric strain sensor layout design according to the present application, (a) stretch hot zone display interface screened by four characteristic edge selection methods (b) layout scheme display and adjustment interface. DETAILED DESCRIPTION

[0044] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0045] Reference Figure 1 A system for stretch fabric strain sensor layout design, comprising:

[0046] A fabric sensor layout algorithm generates a sensor layout scheme through data driving and multiple parameters, specifically comprising the following steps:

[0047] Processing human motion data sets, calculating the stretch rate of edges to represent skin deformation in motion; using a digital human model to calculate skin deformation. The open source data set D-Faust contains two genders, five body types and 129 dynamic sequences. More than 40,000 grid points on the human body surface are captured, each point corresponding to a specific position on the human body surface. The accuracy of this data set in reproducing human motion is close to the optical capture system, with an error of 93% of the data points within 1mm, which ensures one-to-one correspondence between the data set and the actual human body surface points, ensuring the reliability of the calculation results. The data set is reshaped into a format of 6890x3x200, i.e. containing three-dimensional coordinate data of 6890 vertices on the human body surface in 200 frames of time. The distance between two points on the human body surface is defined as an edge. With human motion, the coordinates of the points change, resulting in changes in the length and direction of the edges in different human data sets. To capture this time change, the original data is processed as follows:

[0048] An adjacency matrix is created to connect the adjacency information of the points and the coordinates of the points.

[0049] A point-edge matrix is constructed using the adjacency matrix to record the coordinates and serial number information of the two end vertices of an edge.

[0050] Construct the point-coordinate matrix corresponding to each action and part.

[0051] Identify the change of each edge endpoint using the point-edge matrix and the point-coordinate matrix, calculate the edge length, and construct the edge-length matrix corresponding to each motion.

[0052] Calculate the stretch ratio of each edge in each frame relative to the first frame, which is set as the reference frame, using the edge-length matrix, and construct the edge-stretch ratio matrix. This matrix is used to calculate the skin deformation of the human body surface caused by motion.

[0053] By skin deformation data, and by four feature edge selection methods: stretch extreme method, variance method, autocovariance and correlation coefficient method, and composite standard method, filter the feature edges that can represent the skin deformation caused by motion to form the stretch hot zone; The four feature edge selection methods filter the feature edges that can represent the skin deformation caused by motion, which specifically includes:

[0054] Stretch extreme method: when the maximum stretch of an edge during the motion process exceeds a self-defined threshold, the edge is included in the feature set.

[0055] Variance method: calculate the variance of the stretch ratio of each edge throughout the duration of the action. The larger the variance of the stretch ratio of an edge, the more significant its stretch change, and then the edge is selected as a feature edge.

[0056] Autocovariance and correlation coefficient method: calculate the autocovariance and correlation coefficient of each edge in different action sequences to evaluate its similarity in different actions. The edges with large autocovariance and small correlation coefficient can better distinguish the motion, and such edges are selected as feature edges.

[0057] Composite standard: rank the edges according to the combination of stretch ratio, variance, autocovariance and correlation coefficient to establish a comprehensive ranking index:

[0058]

[0059] where λ represents the stretch ratio of the edge, s 2 represents the variance of the edge, and α, β and γ are the weights assigned to the stretch extreme method, variance method, autocovariance and correlation coefficient method respectively. f(n) is the ranking sum based on these weighted standards.

[0060] Using these four feature selection methods, a set of feature edges can be determined, which form the stretch hot zone of the human body surface and represent the skin deformation region caused by motion.

[0061] The K-Means clustering algorithm is used to classify the stretch hot zones with the expected number of sensors and generate virtual sensors in each cluster. The selected feature edges are divided into clusters equal to the number of sensors by K-Means clustering. The center of the feature edge is taken as the center of the sensor, and the virtual sensors are generated according to the custom sensor length. The direction of each sensor is determined by target optimization calculation, and the final sensor layout scheme is generated. The whole process is as shown in Figure 4 .

[0062] To determine the sensor layout scheme corresponding to the stretch hot zone, two indicators are defined: coverage rate as the main objective function, which measures the capture effect of the sensor on the stretch hot zone; redundancy as an auxiliary evaluation function, which evaluates the degree of overlap of the sensor action range. The following assumptions are made for the placement of the sensor on the skin:

[0063] The sensor is placed close to the skin surface, and the thickness can be ignored.

[0064] The longitudinal deformation of the sensor is much larger than the transverse deformation, so the width w and the transverse stretch ratio λ β of the sensor are considered as constant values.

[0065] The effective working range of the sensor is determined by the product of the length l and the width w of the sensor, and is proportionally adjusted according to the longitudinal stretch ratio λ α and the transverse stretch ratio λ β .

[0066] The coverage range of each sensor is defined by the center coordinates (C x , C y , C z ) and the coordinates of the four corners (x1, y1, z1) to (x4, y4, z4). If a point in the human point cloud falls within the coverage range of the sensor, it is considered to be covered by the sensor. At the same time, the redundancy rate is calculated by determining whether a point is captured by multiple sensors. The schematic diagram of the equivalent model is shown in Figure 5 . The calculation methods of coverage rate and redundancy are as follows:

[0067]

[0068] where m is the number of points in the stretch hot zone covered by each generated sensor, n is the number of sensors, and N is the total number of coordinate points in all stretch hot zones. s min and s max are the minimum and maximum longitudinal stretch ratios, respectively. In the redundancy calculation, C j represents the number of sensors covering a particular stretch hot zone. When a point is covered by multiple sensors, the value of the function I(C j >1) is 1, otherwise it is 0.

[0069] The coverage is defined as the main objective function, and the redundancy is defined as the auxiliary objective function. By optimizing the sensor layout to maximize the coverage of the sensor on the tensile hot zone, the final output of the fabric sensor layout scheme is obtained.

[0070] Further, based on the aforementioned fabric sensor layout algorithm, a digital sensor layout design system is developed on the MATLAB platform (MathWorks, R2021b). The system is used to design and optimize the layout of fabric strain sensors in human motion capture. Although MATLAB is used as the implementation platform, the algorithm logic is universal and can be ported to other development platforms such as Python and C++. The system includes:

[0071] A data initialization module is used to load and parse human motion data. The system uses the open-source D-Faust human motion dataset as the basis data. This dataset is obtained through high-precision optical scanning and contains two genders (male and female), five body types (based on BMI classification: lean, standard, overweight, obese, and severely obese), and 129 dynamic sequences. Each sequence records the deformation information of the human body surface during a specific motion process. The system designs a flexible data interface that supports users to import custom datasets. Users only need to prepare data files according to the SMPL model-based data format (including point cloud format, marker point information, time sequence label, etc.), and the system can automatically complete data loading and preprocessing. This design enables the system to adapt to different research scenarios and special application requirements.

[0072] A parameter setting module is used to support users to adjust multi-dimensional parameters through a visual interface. The system provides a graphical user interface (GUI) and uses a tabbed layout to classify and organize complex parameter settings. Through the visual interactive interface, users can select gender, body type, body part, and monitored motion type by clicking the interface. The system provides three gender options (female, male, and no limit), five different BMI value ranges, and users can select from pre-set body parts or use coordinates to define the region to select the sensor placement part, and can select a single action or action combination to meet specific needs. These selections are then used to process the data loaded in the initialization stage, as shown in FIG. 4. Figure 6

[0073] ​Further, the system's visual interface design fully considers user-friendliness and functional integrity, providing gender selection functionality (male / female / neutral design option without limitation), body type customization based on BMI value (covering 5 standard intervals), flexible selection of body parts (both preset parts and coordinate customization are supported), and comprehensive coverage of exercise types (single action or combined action options). Users can adjust key parameters of the sensor according to actual needs, including number, length, and stretch ratio, and all setting changes can be previewed in real time in the 3D model to ensure that users can intuitively evaluate and optimize the design scheme. This integrated design concept not only improves operational efficiency but also significantly reduces the threshold of professional knowledge, enabling non-professionals to quickly master and use the system for high-quality sensor layout design.

[0074] The layout generation module is used to automatically calculate the optimal position of the sensor according to an algorithm; the deformation calculation engine: the system performs real-time calculation of the skin deformation pattern based on preprocessed point cloud data. This engine uses a hierarchical calculation architecture, first establishes the topological structure of the human body surface, and then realizes real-time updating of the distance between any two points on the human body surface through dynamic maintenance of the edge-edge length matrix. The system takes the initial frame of the motion sequence as the reference benchmark and calculates the deformation rate of all edges in each subsequent frame, thereby quantifying the degree of skin stretching and compression. Through an efficient mapping algorithm, the system remaps the two-dimensional stretch rate data onto the three-dimensional human body surface model to generate an intuitive deformation heat map.

[0075] Feature extraction algorithm implementation: the system uses a parallel computing architecture to simultaneously execute four feature extraction algorithms, and users can select a single method or combine the results of multiple methods by setting weight parameters according to actual needs.

[0076] The stretch extremum algorithm sets a threshold parameter based on the minimum stretching range of the sensor, systematically scans all edges, marks the edges with a stretch rate exceeding the threshold, generates a binary feature map, and automatically selects the top 5% edges with the largest stretch amplitude as candidates.

[0077] The variance analysis algorithm calculates the stretch variance of each edge in the complete motion sequence, uses an adaptive threshold mechanism to identify the edges with the largest stretch fluctuation, and usually selects the top 5% edges in terms of variance.

[0078] The correlation analysis algorithm constructs a correlation matrix between different motion patterns, calculates the autocovariance and cross-correlation coefficient of each edge, sorts based on a composite index, and selects the characteristic edges that best distinguish different motions.

[0079] The composite evaluation algorithm integrates the calculation results of the first three methods, allows users to adjust the weight parameters of each component, and finally outputs the edge set with the highest overall score.

[0080] Layout generation process: Based on the output of the four feature edge selection methods, the system calculates and visualizes the skin deformation area caused by motion, and generates the corresponding layout scheme after the user determines the sensor parameters. The system uses an optimized K-Means clustering algorithm to process the feature edge data. First, randomly select k edge center points as the initial clustering centers, then calculate the distance from all feature edges to each clustering center, and assign each edge to the nearest center. After clustering is completed, the system traverses all edges in each cluster based on the geometric definition of coverage rate and calculates the coverage rate index. Through iterative optimization, the edge position that can achieve the maximum coverage rate is found. When the user selects one of the four feature methods, the system will display the layout scheme generated by the corresponding method, and provide detailed quantitative analysis results, such as Figure 7 as shown.

[0081] Adjustment optimization module, which is used to allow users to adjust parameters in real time and recalculate the layout scheme; the system presents the layout scheme generated based on default parameters in three-dimensional form, and users can make real-time adjustments in multiple ways. The interface provides intuitive slider controls to allow users to adjust the number and standard length parameters of the sensors, and modifies the sensor stretch ratio settings through a dedicated input window. Whenever the user adjusts the parameters, the system will immediately trigger a recalculation process to generate an updated sensor layout scheme and display the adjustment effect in real time on the current interface. This instant feedback mechanism enables users to quickly iterate and optimize the design scheme, and intuitively assess the impact of different parameter combinations on the layout effect.

[0082] Output module, which is used to finally generate visual layout diagrams and detailed quantitative evaluation reports. The system generates rich visual output content to provide users with comprehensive design information. In terms of three-dimensional layout diagrams, the system accurately displays the specific position of each sensor on the human body, supports users to view detailed information including sensor direction, angle, and other key parameters through rotation and zoom operations. The deformation heat map function uses color coding technology to intuitively display the spatial distribution of skin deformation during motion, with different colors representing different stretching intensities. Users can clearly observe the key deformation areas. The coverage analysis chart provides a quantitative relationship curve between sensor coverage rate and sensor number, helping users find the best balance point between sensor cost and monitoring accuracy. All visual results support high-resolution export, which is convenient for subsequent report making and technical exchange.

[0083] The number of devices and processing stages described herein are intended to simplify the description of the application and are not intended to limit the application to the embodiments described. Although embodiments of the application have been disclosed in connection with the specified materials and procedures, it is to be understood that the application is not limited to the disclosed materials and procedures, but rather, the intended application is to suggest that the application is broadly applicable to the pertinent art. Various modifications and alterations of this application will become apparent to those skilled in the art from the disclosure herein, and it is intended that the application shall be limited only by the scope of the claims appended hereto and equivalents thereof.

Claims

1. A system for designing the layout of strain sensors for stretched fabrics, characterized in that: include: The fabric sensor layout algorithm generates a sensor layout solution through data-driven and multiple parameters, specifically including the following steps: Processing human motion datasets, calculating edge stretching rates to represent skin deformation during motion; Using skin deformation data and four feature edge selection methods, namely the stretching extreme value method, variance method, autocovariance and correlation coefficient method, and composite standard method, we screen feature edges that can represent skin deformation caused by movement to form stretching hot zones. The stretching hotspots are classified with the desired number of sensors through the K-Means clustering algorithm, and virtual sensors are generated in each cluster; Coverage is defined as the main objective function and redundancy as the auxiliary objective function. The sensor layout is optimized to maximize the coverage of the stretching hot zone, and the fabric sensor layout plan is finally output.

2. The system for designing a layout of strain sensors for stretched fabrics according to claim 1, wherein: The specific steps of processing the human motion data set are: Create an adjacency matrix connecting the adjacency information of the points and the point coordinate information; Use the adjacency matrix to construct a point-edge matrix that records the coordinates and sequence information of the vertices at both ends of an edge; Construct the point-coordinate matrix corresponding to each action and part; Use the point-edge matrix and point-coordinate matrix to identify the changes in the endpoints of each edge, calculate the edge length, and construct the edge-edge length matrix corresponding to each movement; The edge-edge length matrix is ​​used to calculate the stretching rate of each edge in each frame relative to the first frame, and an edge-stretching rate matrix is ​​constructed. This matrix is ​​used to calculate the skin deformation of the human body caused by movement.

3. The system for designing a layout of strain sensors for stretched fabrics according to claim 1, wherein: The method for screening feature edges that can represent skin deformation caused by motion includes: Stretching extreme value method: When the maximum stretching of an edge during movement exceeds a custom threshold, the edge is included in the feature set; Difference method: Calculate the variance of the stretch rate of each edge during the entire duration of the action. A large variance of the stretch rate of an edge indicates a significant change in stretch rate, and the edge is selected as a feature edge. Autocovariance and correlation coefficient method: Calculate the autocovariance and correlation coefficient of each edge in different action sequences to assess its similarity in different actions. Edges with large autocovariance and small correlation coefficient can better distinguish movements and are selected as feature edges; Composite criteria: Rank edges based on a combination of stretch ratio, variance, autocovariance, and correlation coefficient to create a comprehensive ranking index.

4. The system for designing a layout of strain sensors for stretched fabrics according to claim 1, wherein: The method of classifying the stretching hotspots with the desired number of sensors by using the K-Means clustering algorithm and generating virtual sensors in each cluster specifically includes the following steps: K-Means clustering is used to divide the selected feature edges into clusters equal to the number of sensors; The center of the feature edge is used as the center of the sensor, and a virtual sensor is generated according to the customized sensor length; The orientation of each sensor is determined through target optimization calculation, and finally a sensor layout solution is generated.

5. The system for designing a layout of strain sensors for stretched fabrics according to claim 1, wherein: Also included is a data initialization module, which is used to load and parse human motion data; Parameter setting module, which is used for visual interface to support users to adjust multi-dimensional parameters; A layout generation module, configured to automatically calculate an optimal position of a sensor according to an algorithm; An adjustment and optimization module, which is used to allow the user to adjust parameters in real time and recalculate the layout plan; The output module is used to finally generate a visual layout diagram and a detailed quantitative evaluation report.

Images