Sports equipment monitoring management method and system based on Internet of Things

Abstract

The invention discloses a sports equipment monitoring management method and system based on the Internet of Things, and belongs to the technical field of equipment supervision. The method comprises the following steps: collecting an initial vector of sports equipment during lending and storing the initial vector to a block chain; during returning, performing special outline separation according to the type of the sports equipment, and outputting an outline set with a label; extracting a deformation parameter and a texture entropy from the labeled contour set, and outputting a first damage probability through a convolutional network; obtaining a sports equipment placement process video during returning, and outputting a second damage probability; and obtaining a comprehensive use probability according to the first damage probability and the second damage probability, generating a manual review sampling inspection sequence, and giving an alarm. According to the method, through rod body direction filtering and racket surface grid reconstruction of special contour separation of the net racket type equipment, high-precision segmentation of stacked equipment is achieved, and the problem that in the prior art, due to stacking and overlapping of latticed equipment such as badminton rackets, recognition errors are caused is solved.

Description

Technical Field

[0001] This invention relates to the field of equipment monitoring technology, and in particular to a method and system for monitoring and managing sports equipment based on the Internet of Things. Background Technology

[0002] In the management of sports equipment, mesh-type sports equipment, such as badminton rackets and tennis rackets, are prone to stacking and mixing during return due to their unique structural characteristics—including slender shafts and porous mesh surfaces. This leads to low accuracy in identifying such equipment. Misidentification can mask equipment damage, resulting in accidents where broken frames are not detected during stacking, causing racket heads to fly off during training.

[0003] Therefore, in the existing technology, when multiple pieces of equipment are stacked on storage boxes or shelves, their grid structure will form complex visual occlusions, causing traditional image recognition systems to misjudge multiple racket faces as a single continuous surface. Thus, there is a problem of recognition errors caused by the stacking and overlapping of grid-shaped equipment such as badminton rackets. Summary of the Invention

[0004] This application provides a method and system for monitoring and managing sports equipment based on the Internet of Things, which solves the problem of identification errors caused by stacking and overlapping of grid-like equipment such as badminton rackets in the prior art, and achieves high-precision segmentation of stacked equipment.

[0005] This application provides a method for monitoring and managing sports equipment based on the Internet of Things, including: S1, collecting RFID data and multi-angle images of sports equipment when it is borrowed, generating an initial vector and storing it in the blockchain;

[0006] S2: When returning the sports equipment, acquire the LiDAR point cloud data, perform specific contour separation according to the type of sports equipment, and output a set of tagged contours.

[0007] S3 extracts deformation parameters and texture entropy from the labeled contour set, and outputs the first damage probability through a convolutional network;

[0008] S4, obtain the video of the placement process of the sports equipment when it is returned, and output the second damage probability;

[0009] S5: Based on the first damage probability and the second damage probability, obtain the comprehensive usage probability, generate a manual review and sampling sequence, and issue a warning.

[0010] Furthermore, the specific contour separation includes:

[0011] After applying rod-direction filtering to net-play equipment and reconstructing the racket face mesh, a tagged outline set of net-play equipment is generated:

[0012] It also includes curvature smoothing segmentation of spherical and box-shaped equipment, calculating Gaussian curvature, extracting abrupt change points where the Gaussian curvature gradient value is greater than the gradient threshold to segment the contours, and generating a set of labeled contours for spherical and box-shaped equipment.

[0013] Furthermore, the rod direction filtering includes the following steps:

[0014] The normal vectors of each point in the LiDAR point cloud data of online auction equipment are obtained. Neighboring points are searched using the neighborhood radius, and the directional consistency index of each point is calculated. This directional consistency index is used to determine whether a point in the point cloud belongs to a straight line segment of a pole.

[0015]

[0016] Where I(p) is the directional consistency index, N is the total number of points in the neighborhood, i is the point number in the neighborhood, p is the number of points in the point cloud, and n p Let n be the normal vector of point p. i Let σ be the normal vector of the i-th point in the neighborhood of point p. d Here, e is the direction sensitivity coefficient, and e is the natural constant.

[0017] The point set with directional consistency index higher than the index threshold is retained, discrete noise points are removed, the linear characteristics of the pole are enhanced, and the pole directional filtering is completed.

[0018] Furthermore, after reconstructing the racket face mesh, generating a tagged outline set of net-playing equipment includes the following steps:

[0019] The racket face area of ​​net-playing equipment is identified based on multi-angle images. A point cloud normal vector field is constructed as an indicator vector field to define the reconstruction area. The continuous curved surface of the racket face mesh is reconstructed for regular holes.

[0020] The function for reconstructing the continuous surface of the beat face mesh using regular holes is:

[0021]

[0022] Where χ is the surface function to be reconstructed, Δ is the Laplace operator, V is the indicator vector field, and Ω is the reconstruction region;

[0023] Obtain the local curvature of the connection point between the shaft and the racket face. R is the radius of curvature, weighted by w = e -c Merge the contours to generate a set of labeled contours.

[0024] Furthermore, the deformation parameters are obtained as follows:

[0025] Extract the measured dimension vector of a specified part of the equipment from the tagged contour set, and obtain the standard dimension vector and strain tensor divergence.

[0026] The deformation parameters are obtained based on the measured size vector, the standard size vector, and the strain tensor divergence. The formula for obtaining the deformation parameters is as follows:

[0027]

[0028] Where, δ s For deformation parameters, This is the measured size vector. For standard size vectors, For strain tensor divergence, sgn() is the sign function, -1 for compression, 0 for no deformation, and 1 for tension.

[0029] Furthermore, the method for obtaining the first damage probability is as follows:

[0030] Collect historical equipment damage data to construct a training dataset, including the feature vector X = [δ s E t ,ρ m ,k s ] T And the damage grade label Y according to artificial standards, where δ s E is the deformation parameter. t ρ is the texture entropy value. m k is the density of the material. s The structural stiffness coefficient;

[0031] Construct a convolutional neural network model, which includes convolutional layers and fully connected layers, and use mean squared error as the loss function;

[0032] The model is trained using the Adam optimizer, and an early stopping mechanism is implemented.

[0033] The deformation parameters and texture entropy values ​​input in real time are standardized and preprocessed, and the predicted values ​​are output through the trained model.

[0034] The model's output predictions are converted into the first damage probability using the Sigmoid function.

[0035] Furthermore, the second damage probability is obtained as follows:

[0036] Obtain video of the placement process of sports equipment and extract the trajectory coordinate sequence of the equipment movement from it;

[0037] Discretize the trajectory coordinate sequence into N intervals, statistically analyze the distribution probability of coordinate values ​​in each interval, and calculate the trajectory fluctuation entropy.

[0038] When the trajectory fluctuation entropy is greater than the fluctuation entropy threshold, the second damage probability is calculated;

[0039] The trajectory coordinate sequence is as follows:

[0040]

[0041] in, Let t be the trajectory of the equipment, y be the time index, x(t), y(t), z(t) be the three-dimensional coordinates of the equipment center, t1 be the time when the equipment enters the warehouse, t2 be the time when the equipment completes its positioning, and [] be the time series vector.

[0042] The formula for calculating the trajectory fluctuation entropy is:

[0043]

[0044] Among them, H t Let p be the trajectory fluctuation entropy. i Let N be the probability distribution of the trajectory coordinates in interval i, where i is the discrete interval number and N is the number of discrete intervals.

[0045] The formula for calculating the second damage probability is:

[0046] in, Here, H represents the second damage probability, α is the damage sensitivity coefficient, and H... base This is the baseline entropy value.

[0047] Furthermore, the sampling sequence is obtained as follows:

[0048] The sampling priority is determined based on the overall probability of injury to sports equipment and the average monthly usage frequency.

[0049] Φ(pd ,k ,fu ,k )=pd ,k ·lg(1+fu ,k );

[0050] Wherein, Φ(pd) ,k ,fu ,k ) represents the sampling priority for sports equipment k, pd ,k For the comprehensive injury probability of sports equipment k, fu ,k The average monthly usage frequency of sports equipment k;

[0051] Based on the sampling priority of each piece of sports equipment, they are sorted from high to low to generate a sampling sequence;

[0052] The method for obtaining the comprehensive damage probability is as follows:

[0053]

[0054] Where, p d To determine the overall damage probability, ε1 is the weight of the first damage probability. ε1 represents the first damage probability, and ε2 represents the weight of the second damage probability. This represents the second probability of damage.

[0055] Furthermore, the issuance of the warning includes:

[0056] When the overall probability of damage exceeds the first overall threshold, a Level 1 warning is issued, notifying staff to conduct immediate testing.

[0057] When the overall damage probability is not greater than the first overall threshold but greater than the second overall threshold, a level two warning is issued, notifying staff to conduct testing within a specified time.

[0058] When the overall damage probability is not greater than the second overall threshold, it is marked as a safe state.

[0059] This application provides an Internet of Things-based sports equipment monitoring and management system, including: a loan data acquisition module, a return data acquisition module, a first injury probability assessment module, a second injury probability assessment module, and a sampling sequence generation module;

[0060] The data acquisition module for lending out sports equipment is used to collect RFID data and multi-angle images of the equipment when it is lent out, generate an initial vector, and store it in the blockchain.

[0061] The data acquisition module is used to acquire the lidar point cloud data of the sports equipment during the return process, perform specific contour separation according to the type of sports equipment, and output a set of tagged contours.

[0062] The first damage probability assessment module is used to extract deformation parameters and texture entropy from the labeled contour set, and output the first damage probability through a convolutional network;

[0063] The second damage probability assessment module is used to acquire video of the placement process of sports equipment upon return and output the second damage probability;

[0064] The sampling sequence generation module is used to obtain the comprehensive usage probability based on the first damage probability and the second damage probability, generate a manual review sampling sequence, and issue a warning.

[0065] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages:

[0066] 1. By using specialized contour separation of racket equipment, shaft direction filtering, and racket face grid reconstruction, the independent contours of complex structure equipment can be accurately identified. This enables high-precision segmentation of stacked equipment, reduces the error rate of stacked return recognition, and improves the separation speed of racket equipment. It effectively solves the problem of recognition errors caused by stacking and overlapping of grid-like equipment such as badminton rackets in existing technologies.

[0067] 2. By using a dual-channel damage probability fusion mechanism, including a physical damage channel and a behavioral damage channel, physical deformation and behavioral abnormalities are correlated, thereby enabling early warning of latent damage to equipment, reducing the false negative rate of latent damage to equipment, extending the service life of equipment, and solving the problem that it is difficult to detect internal structural damage caused by violent use by relying solely on manual visual inspection in existing technologies.

[0068] 3. By generating a sampling inspection sequence, damage risk and usage intensity are correlated, thereby enabling precise allocation of manual inspection resources, reducing the workload of manual sampling inspection, improving the timeliness of handling high-risk equipment, and effectively solving the problems of manpower waste and missed inspection of key equipment caused by blind full inspection in existing technologies. Attached Figure Description

[0069] Figure 1 A flowchart illustrating the IoT-based sports equipment monitoring and management method provided in this application embodiment;

[0070] Figure 2 This is a structural diagram of an IoT-based sports equipment monitoring and management system provided in an embodiment of this application. Detailed Implementation

[0071] This application provides a method and system for monitoring and managing sports equipment based on the Internet of Things, which solves the problem of identification errors caused by the stacking and overlapping of grid-like equipment such as badminton rackets in the prior art. By using special contour separation of racket-like equipment, shaft direction filtering and racket face grid reconstruction, high-precision segmentation of stacked equipment is achieved.

[0072] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.

[0073] like Figure 1 The diagram shown is a flowchart of a sports equipment monitoring and management method based on the Internet of Things (IoT) provided in this application embodiment. This method is applied in an IoT-based sports equipment monitoring and management system and includes the following steps: S1, collecting RFID data and multi-angle images of the sports equipment when it is borrowed, and generating an initial vector. And stored in the blockchain, the initial vector including the timestamp t o Physical property vector Image-aware hash value

[0074] S2: When returning the sports equipment, acquire the LiDAR point cloud data, perform specific contour separation according to the type of sports equipment, and output a set of tagged contours.

[0075] S3 extracts deformation parameters and texture entropy from the labeled contour set, and outputs the first damage probability through a convolutional network;

[0076] S4, obtain the video of the placement process of the sports equipment when it is returned, and output the second damage probability;

[0077] S5: Based on the first damage probability and the second damage probability, obtain the comprehensive usage probability, generate a manual review and sampling sequence, and issue a warning.

[0078] In this embodiment, S1 borrows and registers to generate an initial vector. Its physical property vector S2 serves as the basis for equipment classification; S2 performs type-adaptive contour separation on the LiDAR point cloud and outputs a labeled contour set to S3, which establishes a spatial reference for dimensional measurement; S3 extracts deformation parameters and texture entropy based on the contour set; S4 parses the warehouse camera video to obtain the placement trajectory; the placement trajectory parsed by S4 and the physical damage assessment of S3 form a dual-channel input S5; S5 integrates the first damage probability of S3 and the second damage probability of S4 to generate a comprehensive probability, and outputs the sampling sequence and warning instructions.

[0079] Furthermore, the specific contour separation includes:

[0080] After applying rod-direction filtering to net-play equipment and reconstructing the racket face mesh, a tagged outline set of net-play equipment is generated:

[0081] It also includes curvature smoothing segmentation of spherical and box-shaped instruments and calculation of Gaussian curvature κ. G Extract abrupt change points where the Gaussian curvature gradient value is greater than the gradient threshold to segment the contours and generate a set of labeled contours for spherical and box-shaped equipment.

[0082] In this embodiment, the parallel processing results of net-type and spherical / box-type equipment are synchronously integrated into a tagged contour set.

[0083] Furthermore, the rod direction filtering includes the following steps:

[0084] The normal vectors of each point in the LiDAR point cloud data of online auction equipment are obtained. Neighboring points are searched using the neighborhood radius, and the directional consistency index of each point is calculated. This directional consistency index is used to determine whether a point in the point cloud belongs to a straight line segment of a pole.

[0085]

[0086] Where I(p) is the directional consistency index, N is the total number of points in the neighborhood, i is the point number in the neighborhood, p is the number of points in the point cloud, and n p Let n be the normal vector of point p. i Let σ be the normal vector of the i-th point in the neighborhood of point p. d Here, e is the direction sensitivity coefficient, and e is the natural constant.

[0087] The point set with directional consistency index higher than the index threshold is retained, discrete noise points are removed, the linear characteristics of the pole are enhanced, and the pole directional filtering is completed.

[0088] In this embodiment, the filtering result includes the set of axial parameters of the rod, which serves as the positioning reference for surface reconstruction and eliminates connection errors caused by missing point clouds.

[0089] Furthermore, after reconstructing the racket face mesh, generating a tagged outline set of net-playing equipment includes the following steps:

[0090] The racket face area of ​​net-playing equipment is identified based on multi-angle images. A point cloud normal vector field is constructed as an indicator vector field to define the reconstruction area. The continuous curved surface of the racket face mesh is reconstructed for regular holes.

[0091] The function for reconstructing the continuous surface of the beat face mesh using regular holes is:

[0092]

[0093] Where χ is the surface function to be reconstructed, Δ is the Laplace operator, V is the indicator vector field, and Ω is the reconstruction region;

[0094] Obtain the local curvature of the connection point between the shaft and the racket face. R is the radius of curvature, weighted by w = e -c Merge the contours to generate a set of labeled contours.

[0095] In this embodiment, the topology-preserving reconstruction of the hole region solves the problem of measurement reference distortion caused by mesh fragmentation.

[0096] Furthermore, the deformation parameters are obtained as follows:

[0097] Extract the measured dimension vector of a specified part of the equipment from the tagged contour set, and obtain the standard dimension vector and strain tensor divergence.

[0098] The deformation parameters are obtained based on the measured size vector, the standard size vector, and the strain tensor divergence. The formula for obtaining the deformation parameters is as follows:

[0099]

[0100] Where, δ s For deformation parameters, This is the measured size vector. For standard size vectors, For strain tensor divergence, sgn() is the sign function, -1 for compression, 0 for no deformation, and 1 for tension.

[0101] Furthermore, the method for obtaining the first damage probability is as follows:

[0102] Collect historical equipment damage data to construct a training dataset, including the feature vector X = [δ s E t ,ρ m ,k s ] T And the damage grade label Y according to artificial standards, where δ s E is the deformation parameter. t ρ is the texture entropy value. m k is the density of the material. s The structural stiffness coefficient;

[0103] Construct a convolutional neural network model, which includes convolutional layers and fully connected layers, and use mean squared error as the loss function;

[0104] The model is trained using the Adam optimizer, and an early stopping mechanism is implemented.

[0105] The deformation parameters and texture entropy values ​​input in real time are standardized and preprocessed, and the predicted values ​​are output through the trained model.

[0106] The model's output predictions are converted into the first damage probability using the Sigmoid function.

[0107] In this embodiment, the feature encoding layer integrates size deviation, surface texture, and material properties; the convolutional kernel captures the spatial correlation patterns of local damage; and the fully connected layer maps global features to the probability space. This structure enables multi-scale assessment of physical damage and enhances the model's ability to identify complex damage patterns.

[0108] Furthermore, the second damage probability is obtained as follows:

[0109] Obtain video of the placement process of sports equipment and extract the trajectory coordinate sequence of the equipment movement from it;

[0110] Discretize the trajectory coordinate sequence into N intervals, statistically analyze the distribution probability of coordinate values ​​in each interval, and calculate the trajectory fluctuation entropy.

[0111] When the trajectory fluctuation entropy is greater than the fluctuation entropy threshold, the second damage probability is calculated;

[0112] The trajectory coordinate sequence is as follows:

[0113]

[0114] in, Let t be the trajectory of the equipment, y be the time index, x(t), y(t), z(t) be the three-dimensional coordinates of the equipment center, t1 be the time when the equipment enters the warehouse, t2 be the time when the equipment completes its positioning, and [] be the time series vector.

[0115] The formula for calculating the trajectory fluctuation entropy is:

[0116]

[0117] Among them, H t Let p be the trajectory fluctuation entropy. i Let N be the probability distribution of the trajectory coordinates in interval i, where i is the discrete interval number and N is the number of discrete intervals.

[0118] The formula for calculating the second damage probability is:

[0119] in, Here, H represents the second damage probability, α is the damage sensitivity coefficient, and H... base This is the baseline entropy value.

[0120] In this embodiment, use injury is indirectly assessed through behavioral pattern analysis, avoiding the problem of installing sensors on sports equipment that could affect athletes' use.

[0121] Furthermore, the sampling sequence is obtained as follows:

[0122] The sampling priority is determined based on the overall probability of injury to sports equipment and the average monthly usage frequency.

[0123] Φ(pd ,k ,fu ,k )=pd ,k ·lg(1+fu ,k );

[0124] Wherein, Φ(pd) ,k ,fu ,k ) represents the sampling priority for sports equipment k, pd ,k For the comprehensive injury probability of sports equipment k, fu ,k The average monthly usage frequency of sports equipment k;

[0125] Based on the sampling priority of each piece of sports equipment, they are sorted from high to low to generate a sampling sequence;

[0126] The method for obtaining the comprehensive damage probability is as follows:

[0127]

[0128] Where, p d To determine the overall damage probability, ε1 is the weight of the first damage probability. ε1 represents the first damage probability, and ε2 represents the weight of the second damage probability. This represents the second probability of damage.

[0129] In this embodiment, a nonlinear mapping between damage probability and usage intensity is established to optimize the allocation efficiency of manual inspection resources.

[0130] Furthermore, the issuance of the warning includes:

[0131] When the overall probability of damage exceeds the first overall threshold, a Level 1 warning is issued, notifying staff to conduct immediate testing.

[0132] When the overall damage probability is not greater than the first overall threshold but greater than the second overall threshold, a level two warning is issued, notifying staff to conduct testing within a specified time.

[0133] When the overall damage probability is not greater than the second overall threshold, it is marked as a safe state.

[0134] In this embodiment, the instruction distributor activates the hierarchical response protocol based on the probability threshold, and the parameter updater receives the calibration standard parameters from the manual detection results to realize the dynamic optimization of damage detection and continuously improve the accuracy and timeliness of damage warning.

[0135] like Figure 2 The diagram shown is a structural diagram of the sports equipment monitoring and management system based on the Internet of Things provided in this application embodiment. The sports equipment monitoring and management system based on the Internet of Things provided in this application embodiment includes: a loan data acquisition module, a return data acquisition module, a first injury probability assessment module, a second injury probability assessment module, and a sampling sequence generation module.

[0136] The data acquisition module for lending out sports equipment is used to collect RFID data and multi-angle images of the equipment when it is lent out, generate an initial vector, and store it in the blockchain.

[0137] The data acquisition module is used to acquire the lidar point cloud data of the sports equipment during the return process, perform specific contour separation according to the type of sports equipment, and output a set of tagged contours.

[0138] The first damage probability assessment module is used to extract deformation parameters and texture entropy from the labeled contour set, and output the first damage probability through a convolutional network;

[0139] The second damage probability assessment module is used to acquire video of the placement process of sports equipment upon return and output the second damage probability;

[0140] The sampling sequence generation module is used to obtain the comprehensive usage probability based on the first damage probability and the second damage probability, generate a manual review sampling sequence, and issue a warning.

[0141] In summary, the embodiments of this application accurately identify the independent contours of complex equipment by using rod direction filtering and racket face mesh reconstruction for specialized contour separation of net-playing equipment. This achieves high-precision segmentation of stacked equipment, reduces the error rate of stacked return identification, and improves the separation speed of net-playing equipment.

[0142] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0143] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0144] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0145] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0146] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the invention.

[0147] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for monitoring and managing sports equipment based on the Internet of Things, characterized in that, Includes the following steps: S1 collects RFID data and multi-angle images of sports equipment when it is borrowed, generates an initial vector and stores it in the blockchain; S2: When returning the sports equipment, acquire the LiDAR point cloud data, perform specific contour separation according to the type of sports equipment, and output a set of tagged contours. S3 extracts deformation parameters and texture entropy from the labeled contour set, and outputs the first damage probability through a convolutional network; S4, obtain the video of the placement process of the sports equipment when it is returned, and output the second damage probability; S5: Based on the first damage probability and the second damage probability, obtain the comprehensive usage probability, generate a manual review and sampling sequence, and issue a warning.

2. The IoT-based sports equipment monitoring and management method as described in claim 1, characterized in that, The specific contour separation includes: After applying rod-direction filtering to net-play equipment and reconstructing the racket face mesh, a tagged outline set of net-play equipment is generated: It also includes curvature smoothing segmentation of spherical and box-shaped instruments and calculation of Gaussian curvature κ. G Extract abrupt change points where the Gaussian curvature gradient value is greater than the gradient threshold to segment the contours and generate a set of labeled contours for spherical and box-shaped equipment.

3. The IoT-based sports equipment monitoring and management method as described in claim 2, characterized in that, The rod-body directional filtering includes the following steps: The normal vectors of each point in the LiDAR point cloud data of online auction equipment are obtained. Neighboring points are searched using the neighborhood radius, and the directional consistency index of each point is calculated. This directional consistency index is used to determine whether a point in the point cloud belongs to a straight line segment of a pole. Where I(p) is the directional consistency index, N is the total number of points in the neighborhood, i is the point number in the neighborhood, p is the number of points in the point cloud, and n p Let n be the normal vector of point p. i Let σ be the normal vector of the i-th point in the neighborhood of point p. d Here, e is the direction sensitivity coefficient, and e is the natural constant. The point set with directional consistency index higher than the index threshold is retained, discrete noise points are removed, the linear characteristics of the pole are enhanced, and the pole directional filtering is completed.

4. The IoT-based sports equipment monitoring and management method as described in claim 2, characterized in that, After reconstructing the racket face grid, generating a tagged outline set of net-playing equipment includes the following steps: The racket face area of ​​net-playing equipment is identified based on multi-angle images. A point cloud normal vector field is constructed as an indicator vector field to define the reconstruction area. The continuous curved surface of the racket face mesh is reconstructed for regular holes. The function for reconstructing the continuous surface of the beat face mesh using regular holes is: minutes χ ∫∫ Ω ‖Δχ-▽·V‖ 2 dΩ; Where χ is the surface function to be reconstructed, Δ is the Laplace operator, V is the indicator vector field, and Ω is the reconstruction region; Obtain the local curvature of the connection point between the shaft and the racket face. R is the radius of curvature, weighted by w = e -c Merge the contours to generate a set of labeled contours.

5. The IoT-based sports equipment monitoring and management method as described in claim 1, characterized in that, The deformation parameters are obtained as follows: Extract the measured dimension vector of a specified part of the equipment from the tagged contour set, and obtain the standard dimension vector and strain tensor divergence. The deformation parameters are obtained based on the measured size vector, the standard size vector, and the strain tensor divergence. The formula for obtaining the deformation parameters is as follows: Where, δ s For deformation parameters, This is the measured size vector. For standard size vectors, For strain tensor divergence, sgn() is the sign function, -1 for compression, 0 for no deformation, and 1 for tension.

6. The IoT-based sports equipment monitoring and management method as described in claim 1, characterized in that, The method for obtaining the first damage probability is as follows: Collect historical equipment damage data to construct a training dataset, including the feature vector X = [δ s E t ,ρ m ,k s ] T And the damage grade label Y according to artificial standards, where δ s E is the deformation parameter. t ρ is the texture entropy value. m k is the density of the material. s The structural stiffness coefficient; Construct a convolutional neural network model, which includes convolutional layers and fully connected layers, and use mean squared error as the loss function; The model is trained using the Adam optimizer, and an early stopping mechanism is implemented. The deformation parameters and texture entropy values ​​input in real time are standardized and preprocessed, and the predicted values ​​are output through the trained model. The model's output predictions are converted into the first damage probability using the Sigmoid function.

7. The IoT-based sports equipment monitoring and management method as described in claim 1, characterized in that, The second damage probability is obtained as follows: Obtain video of the placement process of sports equipment and extract the trajectory coordinate sequence of the equipment movement from it; Discretize the trajectory coordinate sequence into N intervals, statistically analyze the distribution probability of coordinate values ​​in each interval, and calculate the trajectory fluctuation entropy. When the trajectory fluctuation entropy is greater than the fluctuation entropy threshold, the second damage probability is calculated; The trajectory coordinate sequence is as follows: in, Let t be the trajectory of the equipment, y be the time index, x(t), y(t), z(t) be the three-dimensional coordinates of the equipment center, t1 be the time when the equipment enters the warehouse, t2 be the time when the equipment completes its positioning, and [] be the time series vector. The formula for calculating the trajectory fluctuation entropy is: Among them, H t Let p be the trajectory fluctuation entropy. i Let N be the probability distribution of the trajectory coordinates in interval i, where i is the discrete interval number and N is the number of discrete intervals. The formula for calculating the second damage probability is: in, Here, H represents the second damage probability, α is the damage sensitivity coefficient, and H... base This is the baseline entropy value.

8. The IoT-based sports equipment monitoring and management method as described in claim 1, characterized in that, The sampling sequence is obtained as follows: The sampling priority is determined based on the overall probability of injury to sports equipment and the average monthly usage frequency. Φ(p d,k ,f u,k )=p d,k ·lg(1+f u,k ); Wherein, Φ(p) d,k ,f u,k ) represents the sampling priority for sports equipment k, p d,k f represents the overall probability of injury to sports equipment k. u,k The average monthly usage frequency of sports equipment k; Based on the sampling priority of each piece of sports equipment, they are sorted from high to low to generate a sampling sequence; The method for obtaining the comprehensive damage probability is as follows: Where, p d To determine the overall damage probability, ε1 is the weight of the first damage probability. ε1 represents the first damage probability, and ε2 represents the weight of the second damage probability. This represents the second probability of damage.

9. The IoT-based sports equipment monitoring and management method as described in claim 1, characterized in that, The warning includes: When the overall probability of damage exceeds the first overall threshold, a Level 1 warning is issued, notifying staff to conduct immediate testing. When the overall damage probability is not greater than the first overall threshold but greater than the second overall threshold, a level two warning is issued, notifying staff to conduct testing within a specified time. When the overall damage probability is not greater than the second overall threshold, it is marked as a safe state.

10. A sports equipment monitoring and management system based on the Internet of Things, characterized in that, It includes a loan data acquisition module, a return data acquisition module, a first damage probability assessment module, a second damage probability assessment module, and a sampling sequence generation module; The data acquisition module for lending out sports equipment is used to collect RFID data and multi-angle images of the equipment when it is lent out, generate an initial vector, and store it in the blockchain. The data acquisition module is used to acquire the lidar point cloud data of the sports equipment during the return process, perform specific contour separation according to the type of sports equipment, and output a set of tagged contours. The first damage probability assessment module is used to extract deformation parameters and texture entropy from the labeled contour set, and output the first damage probability through a convolutional network; The second damage probability assessment module is used to acquire video of the placement process of sports equipment upon return and output the second damage probability; The sampling sequence generation module is used to obtain the comprehensive usage probability based on the first damage probability and the second damage probability, generate a manual review sampling sequence, and issue a warning.

Images