Swing analysis device and swing analysis method

Abstract

The swing is classified using time-series data of the torque exerted during the swing motion. [Solution] A swing analysis device 1 for multiple players, comprising: an acquisition unit 14a that acquires time-series data of torque exerted during the swing of a hitting tool for multiple players; an expansion unit 14b that expands the time-series data of torque into modes; a determination unit 14c that determines the adopted mode from the results obtained by the expansion unit 14b; and a stratification unit 14d that stratifies the swing using the determined adopted mode and the results obtained by the expansion unit 14b.

Description

[Technical Field]

[0001] The present invention relates to a swing analysis device and a swing analysis method. [Background technology]

[0002] Non-patent document 1 below describes a method for measuring the swing of a golf club and calculating time-series data of the torque exerted during the swing using an inverse dynamics model. [Prior art documents] [Patent Documents]

[0003] [Non-Patent Document 1] Hiroyuki Okazaki, et al., "Development of a Swing Estimation Method Using Vibration Analysis Results of Torque During Swing and a Mechanical Model," Proceedings of the Dynamics and Design Conference 2020, The Japan Society of Mechanical Engineers, September 2020, No. 20-11. [Disclosure of the Invention] [Problems that the invention aims to solve]

[0004] Time-series data of torque generated during a swing can be considered information corresponding to each player's swing, so it would be extremely useful if we could classify swings (players) from such time-series data.

[0005] This invention was devised in view of the above circumstances, and its main objective is to provide a swing analysis device that can classify swings using time-series data of torque exerted during the swing motion of a striking tool. [Means for solving the problem]

[0006] The present invention relates to a swing analysis device for classifying the swings of multiple players, comprising: an acquisition unit that acquires time-series data of torque exerted during the swing of a hitting tool for the multiple players; an expansion unit that modally expands the time-series data of torque; a determination unit that determines an adopted mode from the results obtained by the expansion unit; and a stratification unit that stratifies the swings of the multiple players using the determined adopted mode and the results obtained by the expansion unit. [Effects of the Invention]

[0007] By adopting the above configuration, the swing analysis device of the present invention can classify swings using time-series data of torque exerted during the swing motion. [Brief explanation of the drawing]

[0008] [Figure 1] This is an overall configuration diagram of the swing analysis system of this embodiment. [Figure 2] This flowchart shows the procedure for the swing analysis method of this embodiment. [Figure 3] This is a perspective on golf clubs as striking tools. [Figure 4] This is a block diagram showing an example of a swing analysis device. [Figure 5] This is a schematic diagram illustrating a 3D rigid body 2-link model. [Figure 6] (a), (b), and (c) are graphs showing the relationship between the x, y, and z axis components of torque around the center of gravity of the arm and time, respectively. [Figure 7] (a), (b), and (c) are graphs showing the relationship between the x, y, and z-axis components of torque around the club's center of gravity and time, respectively. [Figure 8] (a), (b), and (c) are graphs showing the contribution rates of the 1st to 10th modes when the torque time series data in Figures 6(a), (b), and (c) are decomposed by singular value decomposition, respectively. [Figure 9](a), (b), and (c) are graphs showing the contribution rates of the first to tenth modes when the torque time-series data of FIGS. 7(a), (b), and (c) are subjected to singular value decomposition, respectively. [Figure 10] (a), (b), and (c) are graphs with the golfer on the horizontal axis and the magnitudes of the right singular vectors of the x-axis, y-axis, and z-axis of the torque data around the arm center of gravity on the vertical axis. [Figure 11] As an example of the cluster analysis result, it is a dendrogram created using the right singular vectors of the torque data around the arm center of gravity. [Figure 12] (a), (b), and (c) are graphs with the magnitudes of the x-axis, y-axis, and z-axis of the torque data around the arm center of gravity on the vertical axis and time on the horizontal axis, and different line types are used for each grouped group. [Figure 13] It is a further enlarged view of the dashed-two-dot line regions in FIGS. 12(a), (b), and (c). [Figure 14] (a) is the trajectory of the swing motion on the Y-Z plane of the absolute coordinate system, and (b) is the trajectory of the swing motion on the X-Z plane of the absolute coordinate system. [Figure 15] It is a graph showing the average value of the swing plane angles of each grouped group. [[ID=^{19}]]

Mode for Carrying Out the Invention

[0009] Hereinafter, an embodiment of the present invention will be described based on the drawings. The drawings include exaggerated expressions and expressions different from the actual structural dimensional ratios in order to assist in understanding the present invention. Also, when there are multiple embodiments, the same or common elements are denoted by the same reference numerals throughout the specification, and duplicate explanations are omitted. Furthermore, the specific configurations shown in the embodiments and the drawings are for understanding the content of the present invention, and the present invention is not limited to the specific configurations shown.

[0010] [Swing Analysis System] Figure 1 shows an overall configuration diagram of the swing analysis system 100, including the swing analysis device 1 of this embodiment. The swing analysis system 100 is composed of, for example, the swing analysis device 1 and the measurement device 2. The swing analysis device 1 is a device for classifying multiple swings using various hitting tools 3 into several groups. Classifying swings can also be said to be classifying golfers. Therefore, in this specification, the classification of swings may be read as the classification of golfers 5.

[0011] In the swing analysis system 100 of this embodiment, a golf club 30 is used as the striking tool 3. Therefore, the swing analysis system 100 of this embodiment is suitably implemented for analyzing the swing of a golfer 5 as a player using a golf club 30. Please note that the following description is based on this embodiment. However, in addition to a golf club, various rackets, bats, etc., may also be used as the striking tool 3.

[0012] [Swing Analysis Process] Figure 2 is a flowchart showing the procedure for swing analysis (swing classification) using the swing analysis system 100 of this embodiment, and its outline is as follows. (1) Swing measurement (Step S1): Here, the swings of multiple players are measured using measuring device 2. (2) Calculation of time-series data of torque (Step S2): Here, time-series data of the torque exerted during each player's swing is calculated based on their swing measurement data. (3) Acquisition of time-series torque data and mode expansion (step S3): Here, the swing analysis device 1 acquires time-series data of the above torque for each player and expands it into multiple modes. (4) Determination of the adoption mode (steps S4, S4a, S4b): Here, the swing analysis device 1 determines, for each player, an adopted mode that includes one or more modes from among several modes that best represent the characteristics of the torque time series data. (5) Swing stratification (Step S5): Here, the swing analysis device 1 stratifies each player's swing based on the adopted mode and the results of the mode progression.

[0013] As described above, the swing analysis device 1 of the present invention can extract commonalities from time-series data of torque exerted during a swing for multiple players and stratify their swings. This is useful for various applications, such as fitting a shaft 31 or golf club 30 suitable for the swing, improving swing form, and developing golf equipment. The swing analysis device 1 of this embodiment, the swing analysis system 100 including it, and the swing analysis method will be described in detail below in the order of each of the above processes (1) to (5).

[0014] [Swing Measurement (Step S1)] As shown in Figure 1, the swing of the golf club 30 is measured, for example, by measuring the actual swing of the golf club 30 by golfer 5.

[0015] Golfer 5 is intended for a variety of individuals whose swing types need to be categorized. Therefore, there are no particular restrictions on the Golfer 5 individuals to be measured. In this embodiment, 55 right-handed average golfers (average score 70-120) were selected as multiple Golfer 5 individuals, and each performed one swing.

[0016] Figure 3 shows a typical golf club 30. As shown in Figure 3, the golf club 30 includes a shaft 31, a club head 32 fixed to one end (sometimes called the "tip") of the shaft 31, and a grip 33 fixed to the other end of the shaft 31. Although Figure 3 shows a wood-type golf club 30, the golf club 30 is not limited to a wood type and various commercially available clubs can be used. Furthermore, the golf club 30 may be one that each player normally uses.

[0017] Swing measurement can employ various methods and apparatus known prior to the filing of this application, as long as they can measure the swing, which is the time-series behavior of the golf club 30. For example, swing measurement may employ any of the measurement methods described in the applicant's Japanese Patent Publication No. 6851038. One such measurement method uses an inertial sensor unit (not shown) attached to the shaft 31 or grip 33 of the golf club 30. The inertial sensor unit can, for example, acquire time-series data (measurement data) relating to acceleration in three axes, angular velocity, and geomagnetic field in a predefined local coordinate system (described later) at the grip end 33a during the swing.

[0018] Another example of swing measurement is a method of capturing the swing of golfer 5 using a distance image sensor (not shown). The distance image sensor can acquire, for example, two systems of time-series video data (measurement data) including an IR image and a depth image, both of which include golfer 5 and the golf club 30 during the swing. To improve measurement accuracy, the distance image sensor may be used in combination with the inertial sensor unit described above.

[0019] As shown in Figure 1, in this embodiment, an optical motion capture system is used as the measurement device 2 for swing measurement. This system uses, for example, multiple reflective markers 2C, multiple cameras 2A, and a computer 2B. The reflective markers 2C are attached in advance to predetermined positions on the golfer's 5 clothing and golf club 30. When the golfer 5 swings, the movement is captured by the multiple cameras 2A.

[0020] The motion capture system's computer 2B collects video data captured by each camera 2A, analyzes it using the principle of triangulation, and calculates the three-dimensional coordinates of each reflective marker 2C. This allows the three-dimensional coordinates of the required positions to be obtained as measurement data. A golfer's swing typically includes address, top, impact, and finish in that order. In this embodiment, the swing is measured over the entire swing from address to finish at a predetermined short sampling period (e.g., 500 Hz).

[0021] In this embodiment, the position of each reflective marker 2C is determined in a predetermined absolute coordinate system (XYZ). As shown in Figure 1, the absolute coordinate system is a Cartesian coordinate system, with the origin O being the center of the ball 6 being struck. The X-axis is defined as the direction from the golfer's stomach to their back (this direction is positive). The Y-axis is defined as the direction of the ball's flight path (target flight path) (this direction is positive). The Z-axis is defined as the vertical upward direction (this direction is positive).

[0022] As shown in Figure 3, the golf club 30 of this embodiment has, as an example, four reflective markers 2C1 to 2C4 attached to it. Of the four reflective markers 2C1 to 2C4, three reflective markers 2C1 to 2C3 are attached to the grip 33 of the golf club 30, for example, via a jig 2D. The jig 2D comprises two rod-shaped members 2E that protrude from a ring-shaped body in a direction perpendicular to the shaft axis direction. The two rod-shaped members 2E are also perpendicular to each other. Spherical reflective markers 2C2 and 2C3 are fixed to the tips of the two rod-shaped members 2E, respectively. In addition, a reflective marker 2C1 is attached to the base position of the two rod-shaped members 2E.

[0023] Reflective markers 2C1 and 2C2 are positioned such that the vector from reflective marker 2C2 to reflective marker 2C1 aligns with direction a, which is from the heel H to the toe T of the club head 32. Reflective markers 2C3 and 2C1 are positioned such that the vector from reflective marker 2C3 to reflective marker 2C1 aligns with direction b, which is perpendicular to the face 32a of the golf club head and points in the direction of the ball flight. Furthermore, a reflective marker 2C4 is attached to the club head 32 end of the shaft 31.

[0024] Furthermore, in this embodiment, one reflective marker 2C5 is attached to the left shoulder of golfer 5 (see Figure 1).

[0025] [Calculation of time-series data of torque (Step S2)] Next, time-series data of the torque exerted by golfer 5 during the swing motion is calculated. The method for calculating the time-series data of torque from the above-mentioned measurement data is not particularly limited, but in this embodiment, the method described in Non-Patent Document 1 is adopted. Specifically, a three-dimensional rigid two-link model, as shown in Figure 4, is created from the arm of golfer 5 and the golf club 30, and the torque is calculated from this model. The three-dimensional rigid two-link model is composed of a vector from the left shoulder of golfer 5 toward the grip end 33a (a link corresponding to the arm of golfer 5, with the pivot point being the left shoulder of golfer 5) and a vector from the grip end 33a toward the center of gravity of the golf club (a link corresponding to the golf club 30, with the pivot point being the grip end 33a), based on the position coordinates of each measured reflective marker 2C.

[0026] To create a 3D rigid 2-link model, a grip coordinate system and an arm coordinate system are defined as local coordinate systems for the grip end 33a and the left shoulder of the golfer 5, respectively.

[0027] [Grip Coordinate System] As shown in Figures 3 and 4, the grip coordinate system [e→ grip ](Note: The notation "e→" is intended to be the notation "e vector" as shown in Figure 4, and the same applies below.) is a coordinate system with grip end 33a as the origin, and has an x ​​axis (x) with a vector pointing from reflective marker 2C2 to reflective marker 2C1. grip ) has. Also, the grip coordinate system has a y-axis (y) with a vector from reflective marker 2C3 to reflective marker 2C1. grip Furthermore, the grip coordinate system has its x-axis (x grip ) and y axis (y grip The z-axis (z) has a vector obtained by the cross product with ). grip ) has.

[0028] [Arm coordinate system] Arm coordinate system [e→ armis a coordinate system with the reflection marker 2C5 on the left shoulder of the golfer 5 as the origin, and has a z-axis (z arm ) having a vector from the grip end 33a to the reflection marker 2C5 on the left shoulder of the golfer 5. Also, the arm coordinate system has a y-axis (y grip ) having a vector obtained by the cross product of the z-axis (z arm ) of the grip coordinate system and the z-axis (z arm ) of the arm coordinate system. Further, the arm coordinate system has an x-axis (x arm ) having a vector obtained by the cross product of its y-axis (y arm ) and z-axis (z arm ).

[0029] Next, the measured position coordinates of each reflection marker 2C are input into the 3D rigid body 2-link model shown in FIG. 5. Thereby, the arm coordinate system and the grip coordinate system are constructed. Also, the posture at each coordinate is expressed by a quaternion, the angular velocity around each axis is obtained from the change in posture, and further by differentiating these, the angular acceleration is calculated. By inputting these results into the inverse dynamics calculation formula, the torque exerted during the swing is calculated. This calculation procedure is as described in detail in the above-mentioned Non-Patent Document 1, and the content of the above-mentioned Non-Patent Document 1 is incorporated herein by reference.

[0030] And the torque around the arm center of gravity and the torque around the club center of gravity can be calculated by the following formulas (1) and (2), respectively. That is, by inputting the swing data measured in step S1 into the above formulas (1) and (2), the time series data of each torque exerted during the swing operation is calculated.

[0031]

Equation

[0032] In this embodiment, for the classification of the swing, the analysis is performed without considering the reaction term of the torque around the club center of gravity shown in formula (1).

[0033] [Swing phases] To evaluate the swing from the swing motion, the phases of the swing are defined as follows. First, the timing when the golfer 5 starts the motion from the moment they set up with the golf club 30 and come to a standstill is called the address timing. Next, the motion of the grip 33 is projected onto the YZ plane of the global coordinate system, and the resulting grip coordinate system -z axis (z grip The timing at which the angle between the ball 6 and the overall coordinate system's Z-axis is maximized is defined as the top of timing (Top). After the Top timing, the timing at which the ball 6 collides with the club head 32 and moves is defined as the impact timing. In this embodiment, the time-series data of the calculated torque is expressed with the impact timing as time 0(s).

[0034] The processing described above is pre-processing to gather the data necessary for stratifying the swing using the swing analysis device 1 of this embodiment. Then, for each golfer 5, the time-series data of the torque during the swing motion calculated is provided to the swing analysis device 1 via a communication line, wireless communication, a storage device, etc. (all not shown).

[0035] [Swing analysis device] The swing analysis device 1 may be configured using, for example, a general-purpose computer, as illustrated in Figure 1. The computer includes, for example, a desktop computer, a notebook computer, a tablet computer, and a smartphone. The swing analysis device 1 is configured by installing the swing analysis program 13a of this embodiment on one of these computer devices. Therefore, the swing analysis program 13a can perform predetermined processing described later on the swing analysis device 1 based on the measurement data provided by the measurement device 2.

[0036] Figure 5 is a block diagram of the swing analysis device 1. As shown in Figure 5, the swing analysis device 1 comprises a display unit 11, an input unit 12, a storage unit 13, a control unit 14, and a communication unit 15. These units are connected via a bus line 16 so that they can communicate with each other.

[0037] The display unit 11 is composed of, for example, a display or the like. The display unit 11 displays swing analysis results and other information to the user as visual information. Users include golfers whose swings have been measured, instructors, golf equipment developers, and anyone who needs the results of a golf swing analysis.

[0038] The input unit 12 consists of a mouse, keyboard, touch panel, voice input means, etc. The swing analysis device 1 receives operations, instructions, etc. from the user via the input unit 12.

[0039] The memory unit 13 is composed of non-volatile memory such as a hard disk or SSD. The memory unit 13 receives input such as the swing analysis program 13a and time-series torque data 13b received from the measuring device 2.

[0040] The control unit 14 is composed of, for example, a CPU, ROM, and RAM. The control unit 14 reads the swing analysis program 13a from the storage unit 13 and executes predetermined processing. As a result, the control unit 14 virtually functions as an acquisition unit 14a, an expansion unit 14b, a determination unit 14c, and a stratification unit 14d, etc. (details of each unit will be described later). The communication unit 15 is an interface for sending and receiving data with external devices such as the measuring device 2.

[0041] [Acquisition of time-series torque data and mode expansion (Step S3)] In step S3, first, the acquisition unit 14a acquires the time-series data of each golfer 5's torque obtained in step S2 from the storage unit 13. Next, the expansion unit 14b unifies the length of the time-series data for each torque and creates an observation matrix. In this embodiment, data including the downswing period is extracted as the observation matrix. As an example, when the impact timing is taken as the reference (time 0(s)), data in the interval from -0.8(s) to 0(s) is extracted. The interval extracted as the observation matrix is ​​not limited to this example and can be changed in various ways depending on the purpose of the analysis.

[0042] Figures 6 and 7 show the torque data (observation matrix) for 55 golfers obtained as described above. The graphs in Figures 6(a), (b), and (c) show the x, y, and z axis components of the torque around the center of gravity of the arm, respectively (as described above, each component of the torque here has the reaction term (T) of the torque around the center of gravity of the club in equation (1). club ) has been removed. ) and the relationship with time are shown. Graphs 7(a), (b), and (c) show the relationship between the x, y, and z axis components of torque around the club's center of gravity and time, respectively. As shown in Figures 6 and 7, although there are differences in the magnitude of torque among the golfers 5, it can be confirmed that all waveforms show almost the same trend.

[0043] Next, the expansion unit 14b uses the data shown in Figures 6 and 7 to perform mode expansion. Mode expansion extracts the main data structures and common patterns from the observation matrix, which is the torque data. In this embodiment, the expansion unit 14b performs mode expansion by singular value decomposition. Singular value decomposition is performed by the following equation (3).

number

[0044] Here, the right singular vector V (n) This shows the magnitude of the torque data generated by golfer 5. Also, the left singular vector U (n) This indicates time information. Furthermore, the singular value Γ (n) This represents the contribution of each mode to the observation matrix [Ra].

[0045] [Determining the hiring mode (Step S4)] In step S3, the observation matrix, which contains torque data, is expanded into multiple modes. In step S4, the determination unit 14c determines the mode to be adopted for swing classification. The adopted mode will become the data for classifying the swing in the next step S5, so the mode that best represents the overall structure of the original observation matrix is ​​adopted. Such an adopted mode may be determined, for example, based on the contribution rate of each mode in the observation matrix. In one embodiment, the determination unit 14c calculates the contribution rate of each singular value, which is the ratio of each singular value to all singular values ​​(step S4a), and determines one or more modes whose cumulative contribution rates exceed a predetermined threshold as the adopted mode (step S4b). In this embodiment, the contribution rate CR of the nth mode to the observation matrix [Ra] (n) This can be calculated using the following equation (4). In this embodiment, the observation matrix [Ra] is a matrix (number of rows > number of columns) in which -0.8 to 0 (s) are arranged vertically (rows) and the golfers (1 to M) who are subjects are arranged horizontally (columns). Therefore, the number of modes (number of singular values) is equal to the number of subjects M.

[0046]

number

[0047] Figures 8 and 9 show the contribution rates of the 1st to 10th modes when the observation matrix shown in Figures 6 and 7 is subjected to singular value decomposition. In this embodiment, the threshold for the cumulative contribution rate is set to 95% in order to extract modes that adequately represent the observation matrix while reducing noise in the observation matrix. Therefore, in this embodiment, when the cumulative contribution rate is calculated sequentially from the 1st mode to the nth mode, one or more modes whose cumulative contribution rate exceeds 95% are extracted as the adopted modes. Table 1 summarizes the relationship between each torque and the adopted mode.

[0048] [Table 1]

[0049] As is clear from Figures 8 and 9 and Table 1, the x, y, and z components of each torque have a cumulative contribution rate exceeding 95% for the first to second mode, or the first to third mode, and these modes can be considered to adequately represent the entire observation matrix.

[0050] [Swing stratification (Step S5)] In step S4, the adopted mode with a cumulative contribution rate exceeding 95% was identified. In step S5, the stratification unit 14d stratifies each swing based on the adopted mode and the results of the mode development.

[0051] Here, the right singular vector V(n) obtained by singular value decomposition of the observation matrix [Ra] represents the intensity of the torque data for each golfer 5 in each mode. This embodiment focuses on the right singular vector that represents the intensity of such torque data as a result of mode expansion, and stratifies multiple swings (i.e., the swings of 55 golfers 5) into several groups by analyzing the right singular vector in the adopted mode.

[0052] The stratification method is not particularly limited, but in this embodiment, cluster analysis is employed. In the cluster analysis, the right singular vectors in the adopted modes shown in Table 1 above were used to calculate the Euclidean distance for each golfer, and clusters were created using Ward's method. Figure 10 shows a graph with golfers on the horizontal axis and the values ​​of the right singular vectors of torque data around the arm's center of gravity on the vertical axis. Figure 11 shows a dendrogram (tree diagram) created using the Euclidean distances of the right singular vectors of torque around the arm's center of gravity as an example of the cluster analysis results.

[0053] As shown in Figure 11, by performing cluster analysis using the right singular vector obtained from the torque data around the arm's center of gravity, multiple swings can be classified into several groups. In this embodiment, by setting cutoff points as shown by the dashed lines, 55 types of swings can be stratified into three clusters (groups 1 to 3). Table 2 shows the results. The number of clusters to stratify can be set as appropriate depending on the purpose.

[0054] [Table 2]

[0055] The differences in torque were examined for each of the stratified groups 1 to 3 in Table 3. Figures 12(a), (b), and (c) show graphs with the x, y, and z components of the torque around the arm's center of gravity on the vertical axis and time on the horizontal axis. In Figures 12(a), (b), and (c), the solid line represents the swing classified as Group 1 in Table 2, the dashed line represents the swing classified as Group 2 in Table 2, and the dashed line represents the swing classified as Group 3 in Table 2. In Figures 12(a), (b), and (c), the area enclosed by the dashed line is an enlarged view of the target portion (around the impact timing). For example, in Figure 12(b), the range of approximately -0.06 to 0 (s) was extracted, and in Figure 12(c), the range of approximately -0.04 to 0 (s) was extracted. Furthermore, Figure 13 shows a further enlarged view of the area indicated by the dashed lines in Figures 12(a), (b), and (c).

[0056] As shown in Figures 12 and 13, differences can be observed in the x, y, and z-axis components of the torque among groups 1 to 3. More specifically, Figures 13(a) and (b) are magnified views of the x and y-axis components of the torque around the arm's center of gravity near the impact timing, where they are arranged from top to bottom in roughly the order of groups 1, 2, and 3. Figure 13(c) is a magnified view of the z-axis component of the torque around the arm's center of gravity near the impact timing, where it is arranged from top to bottom in roughly the order of groups 2, 1, and 3.

[0057] As described above, the swing analysis device 1 of this embodiment makes it possible to easily classify the swings of multiple golfers 5 using time-series data of the torque exerted during the swing motion.

[0058] Furthermore, in this embodiment, the differences in swing trajectories for each of the classified groups 1 to 3 were also confirmed. Figure 14 shows the results of comparing the trajectories of the reflective marker 2C1 during the swing motion and the reflective marker 2C4 attached to the tip of the shaft 31 in the classified groups 1 to 3. Figure 14(a) shows the trajectories of the reflective markers 2C1 and 2C4 on the absolute coordinate system YZ plane, corresponding to the swing motion viewed from the front (ventral side) of the golfer 5. Figure 14(b) shows the trajectories of the reflective markers 2C1 and 2C4 on the absolute coordinate system XZ plane, corresponding to the swing motion viewed from the rear of the golfer 5. Furthermore, in Figures 14(a) and (b), the solid lines represent swings classified as group 1 in Table 2, the dashed lines represent swings classified as group 2 in Table 2, and the dotted lines represent swings classified as group 3 in Table 2. From Figure 14, it can be confirmed that the stratification of the time-series torque data is reflected in the differences in swing trajectories.

[0059] Furthermore, in the XZ plane shown in Figure 14(b), the swing plane angle was defined as the angle between the line segment from the point where the tip position of the golf club shaft 31 takes maximum value on the Z axis to the point where it takes minimum value, and the XY plane. The differences in swing plane angles among the groups were then examined. Figure 15 shows the results of comparing the mean values ​​of the swing plane angles (°) among the groups. In Figure 15, the error bars indicate the standard deviation. To confirm the differences between the groups, a one-way ANOVA was performed with a significance level of α = 0.05. Since the p-value from the one-way ANOVA was less than 0.05, multiple comparison tests were performed to confirm the details of the differences between the groups. As a result, significant differences were observed between Group 1 and Group 2, and between Group 1 and Group 3.

[0060] Furthermore, Figure 15 shows that the swings in Group 1 have a smaller swing plane angle, while the swings in Group 3 have a larger swing plane angle. These differences in swings are thought to have resulted in differences in the torque data generated during the swing motion.

[0061] In the above embodiment, the results of cluster analysis using torque data around the arm's center of gravity were illustrated, but it goes without saying that cluster analysis can also be performed using torque data around the club's center of gravity. In addition to cluster analysis, other methods such as decision tree analysis, k-approximation, and support vector machines may also be used to stratify the swing.

[0062] Although embodiments of the present invention have been described in detail above, the present invention is not limited to the specific disclosures described above, and can be implemented with various modifications within the scope of the technical idea described in the claims.

[0063] [Note] The present invention includes the following embodiments.

[0064] [Invention 1] A swing analysis device for classifying the swings of multiple players, The acquisition unit acquires time-series data of the torque exerted during the swing of the hitting tool for the aforementioned multiple players, A development unit that performs mode expansion on the time-series data of the torque, A determination unit that determines the mode to be adopted from the results obtained from the aforementioned development unit, A stratification unit that stratifies the swings of the plurality of players using the determined adoption mode and the results obtained by the development unit, A swing analysis device, including a swing analysis device. [2nd Invention] The swing analysis apparatus according to the present invention, wherein the mode expansion is singular value decomposition. [Invention 3] The swing analysis apparatus according to the present invention 2, wherein the determination unit determines one or more modes as the adopted modes, in which the cumulative contribution rate of modes calculated using the singular values ​​obtained in the singular value decomposition exceeds a predetermined threshold. [4th Invention] The result obtained by the aforementioned expansion unit is a right singular vector, according to the swing analysis device according to the present invention, as described in the third invention. [5th ​​Invention] The stratification unit performs the stratification by cluster analysis, as described in any one of inventions 1 to 4. [Invention 6] The swing analysis device according to any one of inventions 1 to 5, wherein the torque is the torque around the center of gravity of the player's arm, or the torque around the center of gravity of the striking tool. [7th Invention] A swing analysis method for classifying the swings of multiple players, The steps include acquiring time-series data of the torque exerted during the swing of the club for multiple players, and The steps include: a step of performing mode expansion on the time-series data of the torque, The step of determining the adopted mode from the results obtained by the aforementioned mode deployment, A step of stratifying the swing using the determined adoption mode and the results obtained from the mode development, A swing analysis method that includes this. [8th Invention] A swing analysis program for classifying the swings of multiple players, The steps include acquiring time-series data of the torque exerted during the swing of the club for multiple players, and The steps include: a step of performing mode expansion on the time-series data of the torque, The step of determining the adopted mode from the results obtained by the aforementioned mode deployment, A step of stratifying the players using the determined adoption mode and the results obtained from the mode deployment, A swing analysis program that is run on a computer. [Explanation of symbols]

[0065] 1. Swing analysis device 3 Batting tools 5 Golfers 14a Acquisition part 14b Expansion section 14c Decision section 14d Stratification Department 30 golf clubs

Claims

1. A swing analysis device for classifying the swings of multiple players, The acquisition unit acquires time-series data of the torque exerted during the swing of the hitting tool for the aforementioned multiple players, A development unit that performs mode expansion on the time-series data of the torque, A determination unit that determines the mode to be adopted from the results obtained from the aforementioned development unit, A stratification unit that stratifies the swings of the plurality of players using the determined adoption mode and the results obtained by the development unit, A swing analysis device, including a swing analysis device.

2. The swing analysis apparatus according to claim 1, wherein the mode expansion is singular value decomposition.

3. The swing analysis apparatus according to claim 2, wherein the determination unit determines one or more modes as the adopted modes, in which the cumulative contribution rate of modes calculated using the singular values ​​obtained in the singular value decomposition exceeds a predetermined threshold.

4. The swing analysis apparatus according to claim 3, wherein the result obtained by the expansion unit is a right singular vector.

5. The stratification unit performs the stratification by cluster analysis, as described in any one of claims 1 to 4.

6. The swing analysis device according to any one of claims 1 to 4, wherein the torque is the torque around the center of gravity of the player's arm, or the torque around the center of gravity of the striking tool.

7. A swing analysis method for classifying the swings of multiple players, The steps include acquiring time-series data of the torque exerted during the swing of the club for multiple players, and The steps include: a step of performing mode expansion on the time-series data of the torque, The step of determining the adopted mode from the results obtained by the aforementioned mode deployment, A step of stratifying the swing using the determined adoption mode and the results obtained from the mode development, A swing analysis method that includes this.

8. A swing analysis program for classifying the swings of multiple players, The steps include acquiring time-series data of the torque exerted during the swing of the club for multiple players, and The steps include: a step of performing mode expansion on the time-series data of the torque, The step of determining the adopted mode from the results obtained by the aforementioned mode deployment, A step of stratifying the players using the determined adoption mode and the results obtained from the mode deployment, A swing analysis program that is run on a computer.

Images