Intelligent badminton racket based on algorithm identification of racket action quality

Abstract

The present application relates to intelligent badminton racket based on algorithm identification swing action quality, specifically relates to the field of sports science, installs inertia measurement unit and pressure sensor to the inside center position of intelligent badminton racket handle and carries out calibration, improves the data acquisition accuracy of intelligent badminton racket, utilizes pressure sensor to collect the pressure distribution data of badminton racket handle when swinging, the comprehensive action data of badminton racket is collected through inertia measurement unit when swinging, the quality and the explainability of data are checked and handled to pressure distribution data, comprehensive action data, utilize wavelet reconstruction to remove the high-frequency noise and burst noise existing when swinging, extract the acceleration characteristics, direction characteristics and pressure characteristics when badminton racket swings, realize the multi-dimensional feature extraction of swing action, construct decision tree model output the identification prediction value of current node badminton racket swing, mapping numerical value range obtains swing action quality score, refines the identification result of swing action.

Description

Technical Field

[0001] This invention relates to the field of sports science, and more specifically, to an intelligent badminton racket based on algorithm-based recognition of swing quality. Background Technology

[0002] As people place increasing emphasis on healthy living, sports, as an important way to maintain health and relieve stress, are gradually gaining widespread recognition and popularity. Among many sports, badminton is particularly favored due to its ease of learning, high level of fun, and the dual requirements of physical fitness and technical skill. With the global spread of badminton, this sport is becoming an indispensable part of people's lives.

[0003] Over time, traditional badminton training methods have gradually revealed some shortcomings, forcing people to seek more technological and personalized training methods to improve training efficiency and competitive level. While traditional badminton training emphasizes honing fundamental skills and improving technique, it often lacks personalized guidance tailored to individual differences, failing to effectively address athletes' problems in technical details and tactical responses. Over-reliance on the experience and observation of human coaches introduces subjectivity and limitations, confining training to fixed patterns and content, lacking innovation and diversity, and failing to stimulate athletes' learning interest and potential. Summary of the Invention

[0004] This invention addresses the technical problems existing in the prior art by providing an intelligent badminton racket that identifies the quality of the swing action based on an algorithm, thereby solving the problems mentioned in the background art.

[0005] The technical solution of this invention to solve the above-mentioned technical problems is as follows: an intelligent badminton racket based on algorithm to identify the quality of the swing action, comprising the following steps:

[0006] S101. Install the inertial measurement unit and pressure sensor inside the center of the smart badminton racket handle and calibrate them. Use the pressure sensor to collect the pressure distribution data of the badminton racket handle when swinging the racket, and use the inertial measurement unit to collect the comprehensive motion data of the badminton racket when swinging the racket.

[0007] S102. Check and process the pressure distribution data and overall motion data, align the pressure distribution data, overall motion data and timestamp information, and use wavelet reconstruction to remove high-frequency noise and sudden noise during the swing motion;

[0008] S103. Extract the acceleration, direction and pressure features of the badminton racket swing, construct a decision tree model to output the recognition prediction value of the badminton racket swing of the current node, and obtain the swing action quality score by mapping the numerical range.

[0009] In a preferred embodiment, in step S101, the inertial measurement unit and the pressure sensor are installed at the center of the inside of the smart badminton racket handle. The inertial measurement unit integrates an accelerometer and a gyroscope. Zero-bias calibration is used to eliminate the zero-bias error of the inertial measurement unit and the pressure sensor in a motionless state. The zero-bias calibration uses the zero-bias value output by the inertial measurement unit and the pressure sensor in an environment without external force influence as a data acquisition reference. The scale factor calibration is used to adjust the amplitude of the output by the inertial measurement unit and the pressure sensor. The scale factor calibration uses a known reference standard to compare the known data output by the inertial measurement unit and the pressure sensor and perform calibration adjustment.

[0010] Furthermore, pressure sensors are used to collect pressure distribution data on the badminton racket handle during the swing, and an inertial measurement unit is used to collect comprehensive motion data of the badminton racket during the swing.

[0011] In a preferred embodiment, in step S102, missing values ​​in the pressure distribution data and overall motion data are detected and deleted, and the average values ​​of the pressure distribution data and overall motion data are used to fill the missing values. Outliers in the pressure distribution data and overall motion data are detected, duplicate records in the pressure distribution data and overall motion data are detected and selectively deleted, and the pressure distribution data, overall motion data and timestamp information are aligned using timestamp information as an index for the pressure distribution data and overall motion data.

[0012] Furthermore, wavelet reconstruction is performed on the pressure distribution data and overall motion data. The pressure distribution data and overall motion data are decomposed into components of different scales using wavelet transform and transformed into the wavelet coefficient domain. Thresholding is then applied to the wavelet coefficient domain. Finally, by substituting the inverse wavelet transform into the thresholded wavelet coefficient domain, the pressure distribution data and overall motion data are reconstructed back to the original domain, restoring their shape and features. The specific formula for wavelet reconstruction is as follows:

[0013]

[0014] Where O(n) represents the reconstructed pressure distribution data and overall action data, c j,k This represents the wavelet coefficients at the nth scale and mth timestamp in the wavelet coefficient domain after thresholding. Let represent the inverse wavelet transform function at the nth scale and the mth timestamp, where j represents the total scale and k represents the total timestamp.

[0015] In a preferred embodiment, in S103, the acceleration features of the badminton racket are extracted by using second-order difference to extract the difference between adjacent velocities of the swing action, the directional features of the swing action trajectory are extracted by the direction angle of the acceleration, the handle is divided into multiple dispersed region grids, and the pressure features are extracted by the average pressure value in each dispersed region grid.

[0016] Furthermore, a decision tree model is constructed, and pressure distribution data and full-scale motion data are input. Acceleration, direction, and pressure features are selected as the splitting features for the current decision tree node, and the splitting point is determined. Based on the splitting features and splitting points, the pressure distribution data and full-scale motion data are divided into multiple data subsets. The splitting steps are recursively repeated to build subtrees until the maximum depth of the decision tree is reached. The current node is then marked as a leaf node, and the predicted value of the leaf node is obtained. The predicted value is mapped to a numerical range of 0-100 using a linear mapping method. The specific formula is as follows:

[0017]

[0018] Where M represents the range of values ​​after mapping, X represents the predicted value, and V... min V represents the minimum predicted value. max Let A represent the maximum predicted value. Set the excellent prediction threshold as A, with a threshold range of 90≤M. Set the qualified prediction threshold as B, with a threshold range of 60≤M<90. Set the unqualified prediction threshold as C, with a threshold range of M<60. Divide the data into N equal-sized result subsets containing pressure distribution data and comprehensive action data. Use one result subset as the validation set in turn, and use the remaining N-1 result subsets as the training set. Repeat the division and step.

[0019] The beneficial effects of this invention are as follows: By eliminating errors through zero-bias calibration and scale factor calibration of the sensors, the accuracy and reliability of the collected data are ensured, improving the data acquisition precision of the intelligent badminton racket. The data collected by the pressure sensor and inertial measurement unit comprehensively reflects all aspects of the swing motion, making the analysis of the swing motion more comprehensive and in-depth. Outlier detection is used to eliminate anomalies in the pressure distribution data and overall motion data, making the analysis more accurate and reliable. Timestamp information is used as an index for the pressure distribution data and overall motion data, ensuring the consistency and correspondence of the pressure distribution data and overall motion data in time. Wavelet transform is used to decompose the pressure distribution data and overall motion data into components of different scales, and thresholding is performed. Value processing removes noise, preserves the main features of pressure distribution data and comprehensive action data while removing unnecessary interference, thereby improving data quality and interpretability. This enables multi-dimensional feature extraction of the swing action, providing a more comprehensive description of its characteristics. A decision tree model is used to classify and predict the extracted features, offering good interpretability and ease of understanding, aiding in comprehending the swing action discrimination process. A linear mapping method maps predicted values ​​to a range of 0-100, making the prediction results intuitively comparable. Graded evaluation of the results is achieved by setting excellent, qualified, and unqualified prediction thresholds, further refining the swing action recognition results. Cross-validation enhances the generalization ability and stability of the decision tree model. Attached Figure Description

[0020] Figure 1 This is a flowchart of the method of the present invention; Detailed Implementation

[0021] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0022] In the description of this application, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0023] In the description of this application, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this application is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed in this application.

[0024] This embodiment provides, for example Figure 1 The smart badminton racket shown here, which uses an algorithm to identify the quality of a swing motion, specifically includes the following steps:

[0025] S101. Install the inertial measurement unit and pressure sensor inside the center of the smart badminton racket handle and calibrate them. Use the pressure sensor to collect the pressure distribution data of the badminton racket handle when swinging the racket, and use the inertial measurement unit to collect the comprehensive motion data of the badminton racket when swinging the racket.

[0026] Furthermore, the inertial measurement unit (IMU) and pressure sensor are installed inside the center of the smart badminton racket handle. The IMU integrates an accelerometer and a gyroscope. Zero-bias calibration is used to eliminate the zero-bias error of the IMU and pressure sensor in a motionless state. This error is caused by the characteristics of the IMU and pressure sensor themselves and external environmental factors. The zero-bias calibration uses the zero-bias value output by the IMU and pressure sensor in an environment without external force influence as a data acquisition reference. The scale factor calibration is used to adjust the amplitude of the output of the IMU and pressure sensor. The scale factor calibration uses a known reference standard to compare the known data output by the IMU and pressure sensor and perform calibration adjustment.

[0027] Furthermore, pressure sensors are used to collect pressure distribution data of the badminton racket handle during the swing, which reflects the grip strength and handle changes during the swing. An inertial measurement unit is used to collect comprehensive motion data of the badminton racket during the swing, including the racket's acceleration, angular velocity, and direction information.

[0028] S102. Check and process the pressure distribution data and overall motion data, align the pressure distribution data, overall motion data and timestamp information, and use wavelet reconstruction to remove high-frequency noise and sudden noise during the swing motion;

[0029] Furthermore, missing values ​​in the pressure distribution data and overall motion data are detected and deleted. The data is then filled with the corresponding values ​​using the average values ​​of the pressure distribution data and overall motion data. Outliers in the pressure distribution data and overall motion data are detected, and these outliers are identified by using the standard deviation of the pressure distribution data and overall motion data to determine their deviation from their own mean. Data exceeding twice the standard deviation of the mean plus or minus two is considered outlier. Duplicate records in the pressure distribution data and overall motion data are detected and selectively deleted. Finally, timestamp information is used as an index to align the pressure distribution data, overall motion data, and timestamp information.

[0030] Furthermore, wavelet reconstruction is performed on the pressure distribution data and overall motion data. Wavelet transform is used to decompose the pressure distribution data and overall motion data into components of different scales and transform them into the wavelet coefficient domain, including approximation coefficients and detail coefficients. Thresholding is applied to the wavelet coefficient domain to remove high-frequency noise and sudden noise present during the swing motion. The thresholding process sets a threshold value; if the wavelet coefficient of a certain timestamp in the wavelet coefficient domain is greater than the threshold, the original wavelet coefficient is retained; otherwise, the wavelet coefficient is set to 0, indicating that it is ignored. The pressure distribution data and overall motion data are reconstructed to the original domain and their shape and features are restored by substituting the inverse wavelet transform into the thresholded wavelet coefficient domain. The specific formula for wavelet reconstruction is as follows:

[0031]

[0032] Where O(n) represents the reconstructed pressure distribution data and overall action data, c j,k This represents the wavelet coefficients at the nth scale and mth timestamp in the wavelet coefficient domain after thresholding. Let represent the inverse wavelet transform function at the nth scale and the mth timestamp, where j represents the total scale and k represents the total timestamp.

[0033] S103. Extract the acceleration, direction and pressure features of the badminton racket swing, construct a decision tree model to output the recognition prediction value of the badminton racket swing of the current node, and obtain the swing action quality score by mapping the numerical range.

[0034] Furthermore, the acceleration features of the badminton racket are extracted by using second-order difference to extract the difference between adjacent velocities of the swing motion, and the directional features of the swing motion trajectory are extracted by the direction angle of the acceleration. The handle is divided into multiple dispersed region grids, where each dispersed region grid represents a specific position, and the pressure features are extracted by the average pressure value within each dispersed region grid.

[0035] Furthermore, a decision tree model is constructed, and pressure distribution data and full-scale motion data are input. Acceleration features, direction features, and pressure features are selected as the splitting features of the current decision tree node, and the splitting point is determined. Based on the splitting features and splitting points, the pressure distribution data and full-scale motion data are divided into multiple data subsets, where each data subset corresponds to pressure distribution data and full-scale motion data at different timestamps. The splitting steps are recursively repeated to build subtrees until the maximum depth of the decision tree is reached. The current node is then marked as a leaf node, and the predicted value of the leaf node is obtained, i.e., the recognition prediction value of the badminton racket swing of the current node. The predicted value is mapped to a numerical range of 0-100 through a linear mapping method. The specific formula is as follows:

[0036]

[0037] Where M represents the range of values ​​after mapping, X represents the predicted value, and V... min V represents the minimum predicted value. max Let A represent the maximum predicted value. Set the excellent prediction threshold as A, with a threshold range of 90≤M. Set the qualified prediction threshold as B, with a threshold range of 60≤M<90. Set the unqualified prediction threshold as C, with a threshold range of M<60. Divide the data into N equal-sized result subsets containing pressure distribution data and comprehensive action data. Use one result subset as the validation set in turn, and use the remaining N-1 result subsets as the training set. Repeat the division and step.

[0038] It should be noted that the descriptions of each embodiment in the above embodiments have different focuses. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0039] 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.

[0040] 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 computer, 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. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0041] 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.

[0042] 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.

[0043] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make further changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention. Clearly, those skilled in the art can make various alterations and modifications to the invention without departing from its spirit and scope. Thus, if these modifications and modifications of the invention fall within the scope of the claims and their equivalents, the invention is also intended to include these modifications and modifications.

Claims

1. An intelligent badminton racket based on algorithm-based recognition of swing quality, characterized in that, Specifically, the following steps are included: S101. Install the inertial measurement unit and pressure sensor inside the center of the smart badminton racket handle and calibrate them. Use the pressure sensor to collect the pressure distribution data of the badminton racket handle when swinging the racket, and use the inertial measurement unit to collect the comprehensive motion data of the badminton racket when swinging the racket. S102. Check and process the pressure distribution data and overall motion data, align the pressure distribution data, overall motion data and timestamp information, and use wavelet reconstruction to remove high-frequency noise and sudden noise during the swing motion; The specific formula for wavelet reconstruction is as follows: ； in, This represents the reconstructed pressure distribution data and overall motion data. This represents the wavelet coefficients at the nth scale and mth timestamp in the wavelet coefficient domain after thresholding. This represents the inverse wavelet transform function at the nth scale and the mth timestamp. Indicates the total scale. Indicates the total number of timestamps; S103. Extract the acceleration, direction and pressure features of the badminton racket swing, construct a decision tree model to output the recognition prediction value of the badminton racket swing of the current node, and obtain the swing action quality score by mapping the numerical range. The acceleration features of a badminton racket are extracted by using second-order difference to extract the acceleration features of the racket swing trajectory by the difference between adjacent velocities of the swing motion. The handle is divided into multiple dispersed region grids, and the pressure features are extracted by the average pressure value within each dispersed region grid.

2. The intelligent badminton racket based on algorithm-based recognition of swing quality according to claim 1, characterized in that: In step S101, calibration is divided into zero bias calibration and scale factor calibration. The zero bias calibration uses the zero bias value output by the inertial measurement unit and the pressure sensor in an environment without external force influence as a data acquisition reference. The scale factor calibration uses a known reference standard to compare the known data output by the inertial measurement unit and the pressure sensor and perform calibration adjustment.

3. The intelligent badminton racket based on algorithm-based recognition of swing quality according to claim 1, characterized in that: In step S102, the specific steps for checking and processing the pressure distribution data and the overall motion data are as follows: detecting and deleting missing values ​​in the pressure distribution data and the overall motion data; filling in the missing values ​​using the average values ​​of the pressure distribution data and the overall motion data; detecting outliers in the pressure distribution data and the overall motion data; and detecting and selectively deleting duplicate records in the pressure distribution data and the overall motion data.

4. The intelligent badminton racket based on algorithm-based recognition of swing quality according to claim 1, characterized in that: In step S102, the timestamp information is used as an index for the pressure distribution data and the overall action data to align the pressure distribution data, the overall action data, and the timestamp information.

5. The intelligent badminton racket based on algorithm-based recognition of swing quality according to claim 1, characterized in that: In step S102, wavelet reconstruction is performed on the pressure distribution data and the overall action data. The pressure distribution data and the overall action data are decomposed into components of different scales through wavelet transform and transformed into the wavelet coefficient domain. Threshold processing is performed on the wavelet coefficient domain. The pressure distribution data and the overall action data are reconstructed to the original domain and their shape and features are restored by substituting the inverse wavelet transform into the thresholded wavelet coefficient domain.

6. The intelligent badminton racket based on algorithm-based recognition of swing quality according to claim 1, characterized in that: In step S103, the specific steps for constructing the decision tree model are as follows: input pressure distribution data and full action data, select acceleration features, direction features and pressure features as the splitting features of the current decision tree node and determine the splitting point, divide the pressure distribution data and full action data into multiple data subsets according to the splitting features and splitting points, recursively repeat the splitting steps to build subtrees until the maximum depth of the decision tree is reached, and mark the current node as a leaf node and obtain the predicted value of the leaf node.

7. The intelligent badminton racket based on algorithm-based recognition of swing quality according to claim 1, characterized in that: In step S103, the specific formula for obtaining the swing quality score by mapping the numerical range is as follows: ； in Indicates the range of values ​​after mapping. Indicates the predicted value. This represents the minimum predicted value. This represents the maximum predicted value.

Images