JP7367467B2 - Swing analysis system, swing analysis program and swing analysis method - Google Patents

Description

The present invention relates to a swing analysis system, a swing analysis program, and a swing analysis method.

Various techniques are known as techniques for recommending (fitting) a golf club suitable for a golf player when the player purchases a golf club. The following techniques are known as the technique. That is, the response surface method is applied to swing data from a plurality of golf clubs of the player to calculate swing data from a golf club that is not swinging, and the movement of the golf club is simulated using the swing data. Then, based on the simulation results, the flight distance and direction of the golf ball are predicted (for example, see Patent Document 1).

Furthermore, fitting using a self-organizing map (SOM) is also known as this technique. Fitting using SOM involves first clustering swing data obtained when multiple players swing multiple golf clubs using SOM, thereby classifying players who make similar swings. Subsequently, swing data of a certain player using a certain golf club is clustered on the SOM. Then, refer to the swing data of similar players on the map. Thereby, even if the number of golf clubs that a certain player swings is reduced, it is possible to efficiently predict swing data from golf clubs that are not being swung (for example, see Patent Document 2).

Furthermore, as a related technique to this technique, a technique is known in which the swing of a golf club is decomposed into singular values and its motion is analyzed (for example, see Non-Patent Document 1).

International Publication No. 2014/132885 International Publication No. 2016/185894

Kenta Matsumoto et. al., Golf swing motion analysis using singular value decomposition for club design, [online], Japan Society of Design Engineers, [Retrieved March 14, 2019], Internet <URL: https:// www.jstage.jst.go.jp/article/jjsde/53/6/53_2017.2761/_pdf/-char/ja＞

When selecting a golf club, it is necessary to select a golf club that is suitable for each player's swing, but in the conventional technology described above, it is difficult to explain in an easy-to-understand manner to the player the influence that the characteristics of the golf club have on the swing motion. Therefore, there is still room for consideration.

One aspect of the present invention aims to realize a swing analysis technique that can quantitatively express the influence of the characteristics of a golf club on a swing motion.

In order to solve the above problems, a swing analysis system according to one aspect of the present invention includes a swing data acquisition unit that acquires swing data of a golf club, and a singular value decomposition process that performs singular value decomposition processing on the swing data. It has a section and a. The swing data is individual swing data for a plurality of golf clubs having a plurality of different characteristics.

Alternatively, in order to solve the above problem, a swing analysis system according to one aspect of the present invention includes a swing data acquisition unit that acquires swing data of a golf club, and a singular value decomposition process that performs singular value decomposition on the swing data. It has a section and a. The swing data is individual swing data for a plurality of golf clubs having different conditions.

Further, in order to solve the above problems, a swing analysis program according to one aspect of the present invention is a swing analysis program for causing a computer to function as the above-described swing analysis system, and the swing analysis program is a swing analysis program for causing a computer to function as the swing analysis system described above. It functions as an acquisition section and the singular value decomposition processing section. The swing data is individual swing data for a plurality of golf clubs having different conditions, or individual swing data for a plurality of golf clubs having a plurality of different characteristics.

Furthermore, in order to solve the above problems, a swing analysis method according to one aspect of the present invention includes the steps of acquiring swing data of a golf club, and performing singular value decomposition on the swing data. As the swing data, individual swing data of a plurality of golf clubs having different conditions of the golf clubs, or individual swing data of a plurality of golf clubs having a plurality of different characteristics is used.

According to one aspect of the present invention, in swing analysis, the influence of the characteristics of a golf club on a swing motion can be expressed quantitatively.

1 is a diagram schematically showing an example of a functional configuration of a swing analysis system according to an embodiment of the present invention. 2 is a flowchart illustrating an example of a swing analysis process according to an embodiment of the present invention. It is a figure showing a self-organizing map by swing analysis processing concerning one embodiment of the present invention. FIG. 6 is a diagram illustrating the probability that the coefficient of determination of the weight parameter with respect to the mode 1 contribution rate is positive for each cluster in a self-organizing map obtained by swing analysis processing according to an embodiment of the present invention. FIG. 7 is a diagram illustrating the probability that the coefficient of determination of the flex parameter with respect to the mode 1 contribution rate is negative for each cluster in a self-organizing map obtained by swing analysis processing according to an embodiment of the present invention. FIG. 7 is a diagram illustrating the probability that the coefficient of determination of the tone parameter with respect to the mode 1 contribution rate is negative for each cluster in a self-organizing map obtained by swing analysis processing according to an embodiment of the present invention. FIG. 6 is a diagram illustrating the probability that the coefficient of determination of the weight parameter with respect to the mode 2 contribution rate is negative for each cluster in a self-organizing map obtained by swing analysis processing according to an embodiment of the present invention. FIG. 7 is a diagram illustrating the probability that the coefficient of determination of the flex parameter with respect to the mode 2 contribution rate is positive for each cluster in a self-organizing map obtained by swing analysis processing according to an embodiment of the present invention. FIG. 7 is a diagram showing the probability that the coefficient of determination of the tone parameter with respect to the mode 2 contribution rate is positive for each cluster in a self-organizing map obtained by swing analysis processing according to an embodiment of the present invention.

Hereinafter, one embodiment of the present invention will be described in detail. FIG. 1 is a diagram schematically showing an example of the functional configuration of a swing analysis system according to an embodiment of the present invention.

[Example of functional configuration of swing analysis system]
As shown in FIG. 1, the swing analysis system 1 includes a control section 10, a sensor 11a, a transmission section 11b, a motion capture 12, a display section 13, a storage section 14, and an input section 15. The control unit 10 will be described later.

As the sensor 11a, the sensors described in Patent Documents 1 and 2 can be used. The sensor 11a is, for example, an inertial unit sensor. The sensor 11a may be attached anywhere on the golf club, but is preferably configured to fit inside the grip portion of the golf club shaft and to be embedded within the golf club shaft of the grip portion. By configuring the sensor 11a to be embedded in the grip portion inside the golf club shaft, the sensor 11a will not come off due to impact at the time of hitting, and the sensor 11a can be easily attached and detached.

Note that the golf club is composed of a head, a golf club shaft, and a grip. The sensor 11a measures the acceleration and angular velocity at the grip position when the player actually swings the golf club in which the sensor 11a is incorporated into the grip.

The transmitter 11b receives the data measured by the sensor 11a and transmits it to the controller 10 via wireless communication. The transmitter 11b is, for example, a transmitter for short-range wireless communication for digital devices such as Bluetooth (registered trademark), which is included in the sensor 11a. Table 1 shows an example of the specifications of the sensor 11a.

As the motion capture 12, the one described in Patent Document 3 can be used. For example, the motion capture 12 can use a motion capture system (manufactured by Motion Analysis). The sampling frequency is, for example, 500 [Hz]. The motion capture 12 measures the behavior of the player's body and the marker attached to the golf club. For example, 39 markers are placed on the body and 4 markers are placed on the club. As a coordinate system for measurement, three axes that are perpendicular to each other can be used. In the coordinate axes, for example, the X axis is in the backward direction of the fly ball line, the Y axis is in the direction from the back to the abdomen of the body, and the Z axis is in the vertical direction.

The locations on the body where markers are placed are, for example, the top of the head, the front of the head, the back of the head, the right shoulder, the left shoulder, the right upper arm, the left upper arm, the outside of the right elbow, the outside of the left elbow, the inside of the right elbow, the inside of the left elbow, the right forearm, and the left. Forearm, outside right wrist, outside left wrist, inside right wrist, inside left wrist, right fist, left fist, right shoulder blade, coccyx, right pelvis, left pelvis, right thigh, left thigh, outside right knee, outside left knee, These are the inside of the right knee, the inside of the left knee, the right calf, the left calf, the outside of the right ankle, the outside of the left ankle, the inside of the right ankle, the inside of the left ankle, the toes of the right foot, the toes of the left foot, the right heel, and the left heel. Markers are installed in the golf club at, for example, the upper edge of the toe portion on the face side, the upper edge of the heel portion on the face side, the rear end portion of the crown, and the grip end of the club head.

The swing analysis system 1 may have both the sensor 11a and the motion capture 12, or may have only one of them.

The display unit 13 is, for example, a device that displays input items, analysis results, and the like. The storage unit 14 is a device that stores, for example, swing data that becomes an input signal in swing analysis, analysis results, and a program for analysis. The input unit 15 is a device for inputting operation signals related to swing analysis. These can be constructed using known devices.

[Example of functional configuration of control unit]
The control unit 10 includes a swing data acquisition unit 100, a variance analysis processing unit 110, a singular value decomposition processing unit 120, a mode contribution rate acquisition unit 130, a multiple regression analysis processing unit 140, and a clustering processing unit 150. The control unit 10 can be constructed using known techniques as long as it has the functional structure shown in FIG.

The swing data acquisition unit 100 acquires swing data of a golf club. The swing data acquisition unit 100 includes, for example, a data reception unit 101, a measured value acquisition unit 102, and a data interpolation unit 103, and acquires swing data from the data acquired by the measured value acquisition unit 102.

The data receiving unit 101 receives data measured by the sensor 11a. The sensor 11a transmits data obtained through measurement to the transmitter 11b, and the transmitter 11b transmits the data to the data receiver 101 by wireless communication.

Note that instead of the transmitter 11b, the sensor 11a may output by directly transmitting data to the data receiver 101 via wired communication. However, transmitting data by wireless communication as described above is preferable because it does not require the cables used in wired communication and allows the golf club to be swung more naturally.

The measurement value acquisition unit 102 acquires motion data output by the sensor 11a or the motion capture 12. In this embodiment, the measured value acquisition unit 102 acquires the motion data of the sensor 11a sent from the transmission unit 11b from the data reception unit 101. The motion data is data that is obtained when a player swings a golf club and represents a swing motion.

In this embodiment, the swing data is data indicating the behavior of a golf club when the golf club is swung. The swing data is, for example, data acquired by the sensor 11a, data acquired by the motion capture 12, or predicted data obtained by supplementing the data obtained from the sensor 11a and the motion capture 12. These data are obtained as acceleration, angular velocity, position data, etc. In this embodiment, the swing data may be extracted and used at six timings during a swing, which will be described later.

The first timing is data at the address position in the swing data. It is also simply called an "address." The address timing is the point immediately before the start of the swing motion. Specifically, average processing is performed on the behavior of three markers attached to the club head. Then, if the amount of change in the Z axis has changed by 100 [mm] or more at t = 0.1 [s] using an arbitrary timing (t = 0 [s]) from the start of measurement as a reference, then t = 0 is defined as the address timing.

The second timing is data in which the golf club is at the 9 o'clock position after the address in the swing data. It is also called "Back 9 o'clock." The back nine o'clock timing is the timing when the golf club becomes parallel to the ground for the first time after address. The timing at which the average value of the Z coordinates of the three markers attached to the club head portion and the Z coordinate of the marker attached to the grip portion coincide for the first time after address is defined as the back nine o'clock timing.

The third timing is data of the position at the start of the downswing in the motion data. Also simply called "top." The top timing is the timing when the backswing changes to the downswing. Specifically, average processing is performed on the behavior of three markers attached to the club head, position data in the X-axis direction is subjected to differential processing, and velocity data in the X-axis direction is calculated. The timing at which the sign of the speed data in the X-axis direction becomes positive after 9 o'clock in the back (the start of the downswing) is defined as the top timing.

The fourth timing is data in which the golf club is at the 9 o'clock position during the downswing in the motion data. It is also called "down at 9 o'clock." The timing of 9 o'clock down is the timing when the golf club becomes parallel to the ground during the downswing. Similar to back 9 o'clock, the down 9 o'clock timing is defined as the timing when the average value of the coordinates of the three markers attached to the club head and the Z coordinate of the marker attached to the grip match for the first time after the top. do.

The fifth timing is data when the golf club comes to the position of the ball during the swing in the motion data. Also called "impact". The timing of impact is the timing when the golf club passes the ball position during the swing. Specifically, the impact is defined as the timing when the three markers attached to the club head move for the first time in the negative direction of the X-axis compared to the time of address.

The sixth timing is data when the golf club reaches the 3 o'clock position in the motion data. It is also called "Follow 3 o'clock." The timing at follow 3 is the timing when the golf club becomes parallel to the ground during the follow swing. Similar to back 9 o'clock and down 9 o'clock, follow the timing when the average value of the Z coordinates of the three markers attached to the club head and the Z coordinate of the marker attached to the grip part match for the first time after impact 3 It is defined as the timing of time.

Among the data obtained from the sensor 11a or the motion capture 12, data from the above address to follow 3 o'clock is extracted as swing data. When performing data complementation, since the extracted swing data is used, data from the address to follow 3 o'clock can be similarly obtained as swing data.

In the control unit 10, a data interpolation unit 103 interpolates swing data, which will be described later. For example, in response surface method, swing data is interpolated using a response surface. As the response surface, a known technique for analyzing golf club swing data, such as that described in Patent Document 1, can be applied. The response surface may be, for example, a single swing response surface that is a response surface related to a golfer's swing, or a combination of a swing response surface and a time response surface that is a response surface related to swing time. Furthermore, as described in Patent Document 2, swing data may be interpolated using a self-organizing map.

As described in Patent Document 1, the swing response surface represents an approximate relationship between swing data and golf club characteristics when a certain player swings a plurality of golf clubs having different characteristics, and is expressed by the following formula (1 ). In formula (1), three variables are used for illustrative purposes, but in the present invention, any number of variables may be used. The characteristics of the golf club will be described later.

In the above equation, f _m (t) represents the swing response surface and is for any golf club. This golf club may be, for example, a golf club that the player has not yet swung. The x bar, y bar, and z bar are arbitrary values of the design variables of the golf club (for example, intermediate values of the design variables). x, y, and z are characteristics of the golf club, for example, x is the weight of the shaft, y is the flex of the golf club, and z is the condition of the golf club. The x-bar, y-bar, and z-bar can be determined to values convenient for subsequent swing analysis. x _m , y _m , z _m represent the aforementioned characteristics in any golf club. That is, x _m is a variable that can take any value. a ₁ to a ₄ represent coefficients on the swing response surface, and are values corresponding to the player's skill and swing habits. t represents swing time.

As described in Patent Document 1, the time response surface is a formulation of an approximate relationship representing the swing time of a certain player and the characteristics of a golf club. The time response curve is expressed by the following equation (2).

In the above formula, g _m represents the time response surface and is for any golf club as described above. x _m , y _m , z _m and x bar, y bar, and z bar are as described above. b ₁ to b ₄ represent coefficients on the time response surface, and are values corresponding to the player's swing time. In formula (2), three variables are used for illustration, but in the present invention, any number of variables may be used.

Further, as described in Patent Document 1, a response surface based on both a swing response surface and a time response surface is expressed by the following equation (3). In the following formula, f _m '(t) is a response surface that is a composite of a swing response surface and a time response surface, and is a response surface that is a composite of a swing response surface and a time response surface. Represents swing data calculated for a club. When changing the characteristics of a golf club, the swing time of the golf club also changes. Therefore, by using a response surface that includes a swing response surface and a time response surface, the change in the swing time can be corrected. As a result, more accurate swing data than that predicted by the swing response surface can be obtained.

In the control unit 10, the clustering processing unit 150 clusters the swing data. Clustering refers to classifying swing data into groups that have similar swing data. A group of similar swing data is called a cluster. The clustering processing unit 150 classifies (clusters) swing data using a self-organizing map (SOM), for example, as described in Patent Document 2. A specific method is described in Patent Document 2.

Note that swing data may be classified by neural network, support vector machine, Bayesian network, hidden Markov model, k-means method, principal component analysis, machine learning, or multivariate analysis.

Next, define the swing change function for the players of the population data. The swing change function is a function that expresses how the swing changes when the characteristics of the golf club are changed in a plurality of golf clubs that are tried.

Let y ₁ be the swing of a certain golf club by a certain player, y ₂ be the swing of a golf club with other characteristics, and let F _{(1, 2)} be the change function of the swing from the former to the latter, then y ₂ is It is represented by the following formula (4). The change function can be obtained, for example, by applying the swing data of the prediction source to y ₁ and applying the swing data of the prediction destination to y ₂ .

By using the swing data function and SOM, it is possible to predict the swing data of a new player. First, swing data of a golf club of a player whose swing data is to be newly predicted is measured. Subsequently, the SOM is used to search for players who have swing data similar to the swing data concerned. By using the swing change functions of players determined to be highly similar, it is possible to accurately predict swing data of a golf club that has non-swinging characteristics. In this way, by using the swing data function and SOM even if the number of trial hits is at least, it is possible to predict swing data with high accuracy.

Note that the control unit 10 may further include other functional configurations as long as the effects of this embodiment can be obtained. For example, in the present embodiment, predicted swing data may be used when interpolating swing data using the response surface described above. In this case, the control unit 10 may further include a functional configuration that predicts swing data from the acquired swing data. More specifically, the control unit 10 may further include a swing data storage unit, a data classification unit, and a data prediction unit.

The data storage section stores swing data acquired by the swing data acquisition section 100. The swing data storage section stores a plurality of swing data from many players. The swing data storage section may have data other than swing data. For example, the swing data storage unit may further include data associated with the swing data. Examples of such data include head behavior data calculated based on swing data.

The data classification section classifies the swing data stored in the swing data storage section. The method by which the data classification unit classifies data is not limited, but may be, for example, classification using SOM. Other examples of such methods include neural networks, support vector machines, Bayesian networks, hidden Markov models, k-means methods, principal component analysis, machine learning, and multivariate analysis. When the swing data storage section stores other data other than swing data, for example the aforementioned head behavior data, the data classification section may classify the other data in the same manner as the swing data. .

The data prediction unit predicts swing data of a golf club having characteristics for which no swing data has been acquired by referring to the swing data classified by the data classification unit. For example, the data prediction unit extracts swing data similar to the acquired swing data. Then, based on the player's data corresponding to the extracted swing data, the swing data of the golf club that has not been swung is predicted. The data prediction unit may predict other data (such as head behavior data) classified by the data classification unit in the same manner as the swing data.

The swing data predicted by the data prediction unit can be used for swing data interpolation using the above-mentioned response surface. Further, the predicted swing data may be used as swing data in the singular value decomposition processing section 120 described above.

The prediction of swing data based on the swing data acquired by the swing data acquisition unit 100 will be described below.

The swing data storage unit stores a plurality of swing data obtained by the swing data acquisition unit 100 by a plurality of players using a plurality of golf clubs. Difference data between swing data for a certain golf club and swing data for another golf club is stored for each player.

The data classification unit classifies swing data from a plurality of golf clubs and differential data between the swing data from these golf clubs for each player.

The data prediction unit predicts swing data of the certain player with another golf club based on one new swing data of the certain player, using appropriate known swing data and difference data. This is based on the assumption that players who have similar swings will also adjust their swings in similar ways when swinging with different golf clubs. The data prediction unit predicts swing data using, for example, SOM.

First, when acquiring one new swing data of a certain player using a first golf club, the data prediction unit identifies the position of the new swing data in the SOM. For example, the data prediction unit searches all the swing data in the swing data storage unit and identifies where in the SOM the new swing data is classified.

Next, in the case of predicting swing data of a certain player using a second golf club, the data prediction unit may predict the swing data of the first golf club and the second golf club among the plurality of swing data in the classification extracted in the SOM. Difference data when all of the swing data of the first golf club and the swing data of the second golf club are combined with each other is calculated for each of the plurality of swing data regarding the golf clubs, and the average value thereof is calculated. The data prediction unit adds the calculated average value to the swing data of a certain player with the first golf club, and calculates the swing data of this player with the second golf club.

Alternatively, the data prediction unit may use the most recent player's difference data instead of the average value of the difference data. The most recent player is the player identified as being closest to a certain player with respect to the swing data of a certain golf club in the classification within the SOM extracted based on the swing data of a certain player with a certain golf club. It is.

The prediction of swing data by the data prediction unit may be based on predicted swing data instead of actually measured swing data.

For example, when another player who has swing data for a certain golf club is identified as the most recent for a certain player who has swing data for a certain golf club, Swing data of another golf club different from the golf club may also be considered to be the most recent to the unmeasured swing data of the other golf club of the certain player. Thereby, one or more unmeasured swing data can be predicted from one new swing data for a certain player based on other swing data for the other player.

By doing the above, the data prediction unit can predict, for example, swing data that can be assumed to be obtained by swinging each of a plurality of types of golf clubs by a certain player.

In the control unit 10, a variance analysis processing unit 110 performs variance analysis on the swing data acquired by the swing data acquisition unit 100. The variance analysis processing unit 110 may perform variance analysis on swing data subjected to singular value decomposition, which will be described later.

The singular value decomposition processing section 120 decomposes the swing data acquired by the swing data acquisition section 100 into singular values and develops it into modes. The mode can also be said to be a unit motion that constitutes a swing motion. As the mode, a known mode for swinging a golf club can be used, and for example, the modes shown in Table 2 are used. This mode is a mode operation verbalized by Matsumoto et al.

(Logic of singular value decomposition)
(1) In the case of a sensor When motion data is acquired by a sensor, singular value decomposition is obtained as follows. Assuming that the timing at the top is the first time point and the timing at impact is the time point N, the acceleration angular velocity vector at the nth time point is expressed by the following equation (5).

Next, a vector shown in Equation (5) is created at times 1 to N points. The observation matrix in which they are arranged row by row is expressed by the following equation (6).

Taking the average of [S] in the time direction with respect to the observation matrix of equation (6), and assuming that the matrix in which N averages are arranged per row is [S ₀ ], [S] is shown in equation (7) below. The mode can be expanded as follows. In the following formula, λ _j represents the singular value of the j-th mode, ν _j represents the left singular vector of [S]-[S ₀ ], and z _j represents the right singular vector of [S]-[S ₀ ] , and T represents a transposed matrix.

(2) In the case of motion capture When motion data is acquired by motion capture, singular value decomposition is obtained as follows. Let the timing of the address in the swing data be the first time point, and the timing at 3 o'clock of the follow be the Nth time point. In the marker attached to the body, the position vector of the i-th marker at the n-th time point is expressed by equation (8).

The components in Equation (8) are the X, Y, and Z direction components of the i-th marker at the n-th time point, respectively. Next, the position vector of equation (8) is created at times 1 to N points. A matrix [r _i ] in which they are arranged row by row is expressed by the following formula (9).

Furthermore, a matrix of equation (9) is created for each marker, for example, for the 1st to 39th markers. A matrix [R] in which they are arranged column by column is expressed by the following formula (10).

In order to bring the reference point of singular point decomposition closer to the posture data at address, the matrix of equation (10) is expanded. First, the position vector of the i-th marker at the time of address is determined using equation (8). A matrix [r ₁ (1)] is created in which 2N points are arranged in each row. All elements of each row of this matrix are r ₁ (1). A matrix [R(1)] in which these are arranged column by column for the 1st to 39th markers is expressed by the following equation (11).

Further, if the matrix obtained by reversing the matrix of equation (10) in the time series direction is [R _t ], then the observation matrix [R _α ] is expressed by the following equation (12).

Regarding the observation matrix of equation (12), take the average of [R _α ] in the time direction, and let [R ₀ ] be a matrix in which 6N of the averages are arranged for each row. [R _α ] can be modally expanded as expressed by the following formula (13). In the following formula, λ _j represents the singular value of the j-th mode, ν _j represents the left singular vector of [R _α ]−[R ₀ ], and z _j represents the right singular value of [R _α ]−[R ₀ ]. It represents a singular vector. Moreover, "n" in the following formula represents the number of marker position vectors. For example, when the number of markers is "39", the position of the marker in space is represented by three components (x, y, z), so n becomes "117".

In the control unit 10, the mode contribution rate acquisition unit 130 acquires the mode contribution rate. The mode contribution rate is the degree of contribution of each mode to [S] or [R _α ] from the singular value. The mode contribution rate is expressed, for example, by the following equation (14). In formula (14), "m" represents the number of modes. m may be any integer, for example, the number (n) of marker position vectors.

In the control unit 10, the multiple regression analysis processing unit 140 obtains a plurality of determination coefficients of the golf club. The coefficient of determination is the degree of contribution of each design variable to the mode contribution rate, and is a quantification of the influence of the design coefficient on the mode contribution rate. For example, the objective function is the contribution rate C _j of the j-th mode, the design variables (weight, flex, tone) are x ₁ , x ₂ , x ₃ , the y-intercept is β ₀ , and the coefficient of determination of each design variable is β _{1 .} , β ₂ , β ₃ , the contribution rate C _j of the j-th mode is expressed by the following equation (15).

A design variable is a variable used in optimization, and indicates a changeable parameter such as dimensions. Optimization is the process of bringing a function, program, product, etc. closer to an optimal state, and is applied to things such as car body design. When performing optimization, a problem is formulated. The formulated function is called an objective function and is usually treated as a minimization problem. Variables that can be changed to find a solution to the objective function are called design variables.

In this embodiment, the design variable is a numerical value determined according to an arbitrary level of the characteristics of the golf club. In this embodiment, the design variable is a characteristic that affects a player's swing motion among various characteristics that a golf club shaft has. In this embodiment, as described above, the weight, flex, and condition of the shaft are used as design variables, but other characteristics may be used. The above three design variables are preferable from the viewpoint of specifying golf club characteristics that have a greater influence on the player's swing motion than other characteristics and are suitable for each player.

In this embodiment, the design variable is determined from the range (0 to 1) when three levels are set for each of the above three characteristics and the range of the design variable is set from 0 to 1. This is the numerical value. In this embodiment, the design variable is 0, 0.5, or 1.

Note that the type, number, and level of design variables in the present invention are not limited to this embodiment. The number and level of design variables may be other than three (for example, five), the range of design variables may be other than 0 to 1 (for example, 0 to 100), or each characteristic It may be an absolute value according to .

In the control unit 10, the clustering processing unit 150 clusters the swing data. Clustering refers to classifying swing data into groups that have similar swing data. A group of similar swing data is called a cluster. The clustering processing unit 150 classifies (clusters) swing data using SOM, for example, as described in Patent Document 2.

Note that swing data may be classified by neural network, support vector machine, Bayesian network, hidden Markov model, k-means method, principal component analysis, machine learning, or multivariate analysis.

[Swing analysis processing example]
The swing analysis process in this embodiment is performed, for example, as follows. In the processing examples below, the golf club characteristics are shaft weight, golf club flex, and golf club tone. The flex of a golf club indicates the stiffness of the shaft and is expressed by the natural frequency of the golf club. The condition of a golf club represents the rate of change in bending stiffness in the shaft, indicating which part of the shaft flexes the most. If the point at which the shaft bends the most is closer to the grip, it is said to be in original condition, if it is closer to the head it is in first condition, and if it is in the middle, it is in middle condition. A plurality of golf clubs differ in at least one of the above three characteristics.

In addition to the above characteristics, the characteristics of the golf club in this embodiment include shaft torsional rigidity (also referred to as "torque"), shaft torsional rigidity distribution, shaft weight distribution, golf club length, and head weight. , club balance, head center of gravity depth, head center of gravity height, head center of gravity distance, grip weight, loft angle, lie angle, and face angle.

FIG. 2 is a flowchart showing an example of a swing analysis process according to an embodiment of the present invention.

In step S201, the swing data acquisition unit 100 acquires the player's swing data. For example, the sensor 11a or the motion capture 12 measures motion data when a certain player swings with a predetermined golf club. The measurement value acquisition unit 102 acquires the motion data from the sensor 11a or the motion capture 12. The swing data acquisition unit 100 extracts the aforementioned first to sixth swing data (address, back 9 o'clock, top, down 9 o'clock, impact, follow 3 o'clock) from the motion data. Characteristic parts of a player's swing motion can be acquired as swing data.

In step S202, the data interpolation unit 103 interpolates the swing data using response surface method. In interpolating the swing data, as described above, the swing data may be interpolated using predicted swing data. Prediction of swing data will be described in detail later, and below, an example of a method for interpolating the swing data acquired by the swing data acquisition unit 100 will be described.

As the response surface, one or both of the above-described swing response surface and time response surface is used. For example, assume that a certain player has acquired swing data for nine golf clubs that differ in three characteristics: shaft weight, golf club flex, and pitch. A response surface is created based on the data for these 9 swings, and based on the response surface, if the flex and tone of the golf club characteristics are varied at 5 levels, and the weight is varied at 3 levels, the response curve for 75 swings is created. Swing data is interpolated.

By interpolating such data, it is possible to reduce the number of swings by the player, shorten the time required for analysis, and reduce data fluctuations, thereby further increasing the accuracy of swing analysis. becomes possible. Note that if a sufficient number of swing data can be acquired in step S201, step S202 may be omitted.

In step S203, the variance analysis processing unit 110 performs variance analysis on the swing data. By applying variance analysis to the acquired swing data, it is possible to determine which timing of the player's swing motion is affected when the characteristics of the golf club change. Below, an example of variance analysis performed on swing data obtained when the golf club's pitch is changed among the characteristics of a golf club is shown.

Swing data of four players H, A, T, and Y swinging two types of golf clubs having different pitches was obtained by motion capture 12. The average scores of each player were 80, 85, 80, and 80, respectively. For each player's swing data, we calculated differences in golf club condition (2 degrees of freedom), trials (10 degrees of freedom), and posture at each timing (39 markers, 117 degrees of freedom depending on the three axis directions). A three-way analysis of variance was performed using factors. The results are shown in Table 3.

In Table 3, the row labels indicate player names, and the column labels indicate each of the above-mentioned timings in the swing data. The values listed in the table indicate p values for determining significant differences in posture data depending on the condition of the golf club. If the p value is <0.05, this indicates that a difference in the condition of the golf club is affecting the posture data at that timing.

A common point among the above players is that at the top timing, it can be seen that the swing changes depending on the condition of the golf club. This is thought to be because the velocity vector of the golf club reverses at the timing of the turnaround at the top, causing the shaft to flex significantly, and players unconsciously change their swings due to the difference in the way the shaft flexes. . In this way, by applying variance analysis to swing data, it can be seen where changes occur in the player's swing motion when the condition of the golf club changes.

In step S204, the singular value decomposition processing unit 120 decomposes the swing data into singular values according to each of the aforementioned modes from the first mode to the fifth mode. For example, the singular value decomposition processing unit 120 obtains the singular values in each of the five modes described above based on the aforementioned equation (13). Thus, the swing data is expanded into the aforementioned modes by singular value decomposition. By expanding the modes, it becomes possible to determine the mode contribution rate in the swing motion, as will be described later. Since the swing data is developed into the modes expressed in the aforementioned words, when explaining the swing analysis results to the player based on the mode, it becomes easy to explain the swing analysis results in an easy-to-understand manner using expressions of motion. From this point of view, swing data can be handled based on more appropriate unit movements.

In step S205, the variance analysis processing unit 110 performs variance analysis on the swing data subjected to singular value decomposition. According to this variance analysis, it can be seen which timing during the swing motion is affected for each mode when changing the golf club. Below, an example of variance analysis performed on swing data subjected to singular value decomposition when the golf club's pitch changes among the characteristics of the golf club will be shown.

For each player mentioned above, posture data was extracted from the behavior exhibited by each mode from the first mode to the fifth mode, and three-way variance analysis was performed as in the previous chapter. The results are shown in Tables 4 to 7. In the table, the row labels indicate the mode-specific movements of the subject, and the column labels indicate the above-mentioned swing timings in the swing data. The values listed in the table indicate p values for determining significant differences in posture data depending on the condition of the golf club.

As a common feature among players, it is confirmed that the impact from the top of the first mode has a p value < 0.05. That is, it was found that the behavior from the top to impact in the first mode and the behavior from address to back nine o'clock in the second mode are commonly affected by changes in the condition of the golf shaft for all players. Similarly, it has been found that the third mode 3 to the fifth mode have timings that are commonly affected by the condition of the golf shaft. More specifically, the following characteristics are confirmed.

- The behavior of the first mode (swinging up and down of the arms, constant rotation of the legs and hips) changes from the top to the impact when the condition of the golf shaft is changed.
- The behavior in the second mode (rotational behavior of the shoulder) changes from address to 9 o'clock back when the condition of the golf shaft is changed.
- The behavior of the third mode (hip, leg thrust, rotational behavior) changes from address to top when the condition of the golf shaft is changed.

In this way, by performing variance analysis on the swing data after singular value decomposition, it is possible to specify which timing in a portion of the swing motion the characteristics of the golf shaft or golf club affect. Therefore, it is possible to specify mode operations that are affected by design variables.

In step S206, the mode contribution rate acquisition unit 130 calculates the mode contribution rate in each mode based on the above-mentioned equation (13). Here, the mode operation in the golf swing is expressed by the following equation (16). In the formula below, φ represents the entire swing, α _i represents the contribution rate of the i-th mode, ψ represents the player's body movement for each mode, x represents the design variables of the golf club, and t represents the swing It represents time, and p represents personal data. α shown in equation (16) is the same as the contribution rate C described above.

In this way, the swing motion of the golf club can be expressed as the sum of the products of the mode contribution rate and the body movement for each mode. Therefore, it is possible to guide changes in the player's swing when at least one of the characteristics of the golf club is changed.

In step S207, the multiple regression analysis processing unit 140 performs multiple regression analysis on the swing data regarding golf club characteristics. The multiple regression analysis processing unit 140 obtains the degree to which each of the characteristics of the golf club contributes to the mode contribution rate in each of the first to fifth modes as the above-mentioned coefficient of determination. Therefore, it is possible to decompose the contribution rate of each mode into the proportion of each design variable. An example of multiple regression analysis performed in the first mode and the second mode using swing data of 21 players is shown below.

Nine swing data were measured for 21 players, and the swing data was interpolated using the swing response surface and time response surface described above to obtain 75 swing data for each player. The swing data was subjected to singular value decomposition as described above, the mode contribution rates of the first mode and the second mode were calculated, and multiple regression analysis was performed on the mode contribution rates. The results are shown in Table 8. In the table below, β ₀ represents the value of the y-intercept in the relational expression obtained by multiple regression analysis, and β ₁ represents the coefficient of determination of the design variable in the weight of the shaft in the golf club. Further, β ₂ represents the coefficient of determination of the design variable in the flex of the golf club, and β ₃ represents the coefficient of determination of the design variable in the condition of the golf club.

The coefficient of determination represents how the objective function changes with changes in the relevant design variables. As a concrete example, if β ₁ is positive, it can be said that as the weight parameter increases, the mode contribution also increases. An increase in mode contribution rate indicates an increase in the degree of contribution in the swing of the corresponding mode operation. Therefore, by performing multiple regression analysis on swing data regarding golf club characteristics, it is possible to describe swing changes due to changes in design variables.

Regarding the multiple regression analysis example described above, the results of calculating the average and standard deviation of the coefficient of determination for all 21 players for each mode contribution rate are shown in Tables 9 and 10 below. Table 9 shows the calculation results for the mode contribution rate of the first mode, and Table 10 shows the calculation results for the mode contribution rate of the second mode.

Create a normal distribution based on the above calculation results. In a normal distribution, the probability density is determined by the distance based on the standard deviation from the average value, and the probability that the coefficient of determination is positive or negative is determined. Table 11 shows the probability of the sign of the coefficient of determination in the first mode, and Table 12 shows the probability of the sign of the coefficient of determination in the second mode.

In this way, by referring to the results of multiple regression analysis, it is possible to obtain the probability of the sign of the coefficient of determination. When the above-mentioned clustering process is executed, the sign of the coefficient of determination cannot be predicted with sufficiently high accuracy by referring to the results of the multiple regression analysis described above (the sign of the coefficient of determination tends to be either positive or negative). It also clearly shows whether the sign is likely to be positive or negative. Therefore, it is possible to more accurately predict changes in the swing motion when changing the characteristics of the golf club in each mode.

In step S208, the clustering processing unit 150 clusters the swing data using the SOM. As a result, players who make similar swings are classified into the same cluster. FIG. 3 shows the results of classifying the acquired swing data of 21 players by SOM. In the figure, symbols C1 to C6 correspond to the first to sixth clusters, respectively. Further, the numbers 1 to 21 in the figure represent each player in the player data, and the position of the number indicates the position of the player data in each cluster.

Regarding the above clustering, similar verification was performed only on player data belonging to each cluster, and the obtained probabilities are shown in FIGS. 4 to 9. In the figure, the numbers C1 to C6, 1 to 21, and the positions where the numbers are written are the same as in FIG. 3. The probabilities shown in each figure from FIG. 4 to FIG. 9 are the probabilities that the sign that is superior to all players in the cluster, that is, the sign of the coefficient of determination of the target parameter becomes the sign listed in Tables 13 and 14. It shows.

Referring to FIG. 4, the probability that only the fifth cluster has a positive determination coefficient is close to 50%, and the other clusters have extremely high probabilities. In this case, it can be said that in the fifth cluster, the mode contribution rate is difficult to change due to a change in weight characteristics, and in the other clusters, the mode contribution rate also increases with an increase in weight. Regarding FIGS. 4 to 9, Tables 13 and 14 show the codes corresponding to those determined to be 75% or more (25% or less). In the table, the symbol "-" indicates that the influence of the parameter change on the mode contribution rate in that cluster is small or indeterminate.

As is clear from FIGS. 4 to 9, in the case of the contribution rate of the first mode, all clusters except the fifth cluster have a correlation with the weight characteristics of the shaft. Furthermore, it can be said that there is a correlation with all three characteristics of the golf club in the third cluster and the fourth cluster. In the case of the second mode contribution rate, there are few clusters that are relatively correlated with the characteristics of the golf club. However, in the fourth cluster, as in the case of the first mode, a correlation is seen in all three characteristics.

If there is a correlation between the mode contribution rate and the characteristics of the golf club, as shown in Table 2 above, changes in the movement of each mode can be explained by changes in the characteristics of the golf club. For example, explain that if you increase the weight design variable from 0 to 1 (reduce the weight), the contribution rate of the first mode will increase, thereby exciting the arm swing-up and swing-down movements. Can be done. Therefore, by clustering the swing data of a new player based on the population data, it is possible to predict changes in swing motion due to changes in the characteristics of the golf club. More specifically, through the above clustering, it is possible to infer how a new player's swing motion will change when he changes the characteristics of his golf club, depending on the cluster to which that player's data belongs. It is.

Furthermore, by acquiring the swing data of a new player and classifying it by SOM, it is possible to predict how the swing of the new player will change if each design variable changes.

Further, according to this embodiment, it is expected that it can be used to estimate a golf club that brings about a desired change in a certain player's swing motion.

[Effect]
In this embodiment, by applying singular value decomposition to the movement of the golf swing, the swing movement can be understood as a combination of linear and simple mode movements.

Furthermore, by performing variance analysis on the swing data in this embodiment, it may be possible to identify swing data that is affected by changes in the characteristics of the golf club. For example, as described above, by applying variance analysis to swing data, it can be seen that changes in the condition of the golf club affect the entire swing at the top timing of the swing motion. Furthermore, it can be seen that in the first mode 1 the swing movement is affected from the top to the impact, and in the second mode the swing movement is affected from the address to the back 9 o'clock.

Further, as described above, a golf swing can be expressed by the sum of products of a mode contribution rate and a mode operation. When the characteristics of the golf club are changed, the essential mode operation does not change significantly, but the mode contribution rate changes. That is, since the degree of contribution of the mode contribution rate to each mode operation changes, the actual movement of the body changes, and therefore the swing movement changes. According to the present embodiment, it is possible to infer changes in the swing motion from changes in the mode contribution rate due to changes in the characteristics of the golf club.

More specifically, the first mode of movement, ``arm swinging up and swinging down movements,'' is excited by reducing the weight of the shaft. Furthermore, depending on the player who belongs to a certain cluster classified by the above-mentioned SOM, the "raised and lowered arm movements" may be stimulated by stiffening the flex or adjusting the tone to the original tone side. .

In this embodiment, the correlation between the design variables of the golf club and the mode contribution rate is characterized for each cluster. As a result, clustering is performed on the newly obtained player data based on the population data. Thereby, it is possible to measure changes in the swing motion due to changes in the characteristics of the golf club based on the clusters that correspond to the clusters.

By applying this method to existing swing estimation systems, it becomes possible to explain to players changes in swing motion due to changes in the characteristics of a golf club in a more quantitative manner. It is also expected that it will become possible to perform golf club fitting for the purpose of changing swing motion.

Furthermore, since it is possible to quantitatively explain the relationship between golf club characteristics and swing motion, it is possible to more easily and accurately suggest golf clubs suitable for players, regardless of their level of fitting skill. It becomes possible.

[Modified example]
The swing analysis system according to the embodiment of the present invention may have further functional configurations within the range where the effects of the present embodiment can be obtained. For example, the swing analysis system includes an ID acquisition unit that acquires a player's ID, and a unique information reference unit that refers to information unique to the player and further reflects the information in the golf club that is selected based on the above-mentioned swing analysis results. , may further include. The ID may be any code that can identify each player on the data.

The player-specific information includes, for example, the player's past swing data, the player's age, the player's purchase history of golf clubs, the player's preferred manufacturer and brand, the player's playing frequency, and the like. These unique information can be appropriately reflected in the golf club by, for example, appropriately converting the information into numerical values and adjusting the degree of recommendation of the golf club selected from the swing analysis results.

For example, among the unique information, swing data or unique information more closely related to the swing data is weighted more heavily with respect to the degree of recommendation. For unique information that has less relevance to the swing analysis results, such as personal preference, the recommendation level is weighted less. As a result, player-specific information is appropriately added to the swing analysis results, and it is possible to further increase the satisfaction level of the player who is recommended a golf club based on the swing analysis results.

In addition, for golf club sellers, the seller's customer is the player mentioned above, and the company supplies the seller with a golf club based on the customer's swing analysis results and with the customer's unique information taken into account. It becomes possible to do so. This will further promote the purchase of golf clubs by customers at the store, which will further improve the sales performance of the store, and will further increase the satisfaction level of customers who have purchased golf clubs at the store. becomes.

Note that for a store, information specific to the store may be further added in the unique information reference section. The information specific to the store includes, for example, the store's strengths, sales history, customer base, and the like. For example, a store that is experienced in selling high-end brands will offer products from higher-end brands among the golf clubs that can be selected based on a customer's swing analysis. It becomes possible to provide golf clubs that more closely match the store's strategy, and it becomes possible to increase the store's satisfaction level.

[Example of implementation using software]
The control block (for example, the control unit 10) of the swing analysis system 1 may be realized by a logic circuit (hardware) formed in an integrated circuit (IC chip) or the like, or may be realized by software.

In the latter case, the swing analysis system 1 includes a computer that executes instructions of a program that is software that implements each function. This computer includes, for example, one or more processors and a computer-readable recording medium that stores the above program. In the computer, the processor reads the program from the recording medium and executes the program, thereby achieving the object of the present invention.

As the processor, for example, a CPU (Central Processing Unit) can be used. As the recording medium, in addition to "non-temporary tangible media" such as ROM (Read Only Memory), tapes, disks, cards, semiconductor memories, programmable logic circuits, etc. can be used. Further, the computer may further include a RAM (Random Access Memory) for expanding the above program.

Furthermore, the program may be supplied to the computer via any transmission medium (communication network, broadcast waves, etc.) that can transmit the program. Note that one aspect of the present invention can also be realized in the form of a data signal embedded in a carrier wave, in which the program is embodied by electronic transmission.

The present invention is not limited to the embodiments described above, and various modifications can be made within the scope of the claims, and embodiments obtained by appropriately combining technical means disclosed in different embodiments. are also included within the technical scope of the present invention.

[summary]
A swing analysis system (1) in an embodiment of the present invention includes a swing data acquisition unit (100) that acquires swing data of a golf club, and a singular value decomposition processing unit (120) that performs singular value decomposition on the swing data. . The swing data is individual swing data for a plurality of golf clubs having a plurality of different characteristics.

According to this configuration, in swing analysis of a golf club, it becomes possible to quantitatively express the influence of each of a plurality of characteristics of a golf club or a combination of a plurality of characteristics on a swing motion.

Alternatively, a swing analysis system according to an embodiment of the present invention includes the above-described swing data acquisition section and singular value decomposition processing section, and the swing data is individual swing data of a plurality of golf clubs having different golf club conditions. be.

According to this configuration, in the swing analysis of the golf club, it is possible to quantitatively express the influence of the golf club on the swing motion in the motion in which the shaft flexes in the swing motion.

In an embodiment of the present invention, the singular value decomposition processing unit may decompose the swing data into singular values corresponding to a mode representing a unit motion that constitutes a swing motion of a golf club.

According to this configuration, it is possible to process the swing motion data based on a more appropriate unit, which is even more effective from the viewpoint of increasing the accuracy of swing analysis.

In an embodiment of the present invention, it may further include a mode contribution rate acquisition unit (130) that acquires a mode contribution rate that is a contribution rate to a mode in swing data.

This configuration is effective from the standpoint of explaining in more detail the changes in swing motion due to changes in the characteristics of the golf club.

The embodiment of the present invention further includes a regression analysis processing unit (multiple regression analysis processing unit 140) that performs regression analysis processing on the mode contribution rate and obtains the degree of contribution of the characteristics of the golf club to the mode contribution rate. Good too.

According to this configuration, it is possible to obtain the degree of contribution of the characteristics of the golf club in the mode contribution rate, which is effective from the viewpoint of explaining in more detail changes in the swing motion due to changes in the characteristics of the golf club.

In an embodiment of the present invention, the regression analysis processing unit may further calculate the sign of the coefficient of determination and the probability that the coefficient of determination is positive or negative.

According to this configuration, it is possible to make the tendency of the sign of the coefficient of determination more clear, and therefore, it is possible to more accurately measure the swing motion.

In an embodiment of the present invention, it may further include a clustering processing unit (150) that clusters swing data using SOM.

This configuration is even more effective from the viewpoint of predicting swing data of different golf clubs by a player with high accuracy from a small amount of swing data of known golf clubs by the player.

In an embodiment of the present invention, the swing data acquisition section may further include a swing data interpolation section (data interpolation section 103) that interpolates swing data using a response surface in response surface method.

According to this configuration, it is possible to increase the amount of swing data that can be applied to swing analysis with fewer player swings, which is even more effective in terms of shortening swing analysis time and reducing the burden on players. .

In embodiments of the present invention, the response surface may be a swing response surface, or may be a swing response surface and a time response surface.

This configuration is even more effective from the perspective of increasing appropriate swing data that reflects the player's characteristics.

In an embodiment of the present invention, the swing analysis system includes a swing data storage unit that stores swing data acquired by the swing data acquisition unit, a data classification unit that classifies the swing data stored in the swing data storage unit, and a data classification unit that classifies the swing data stored in the swing data storage unit. The golf club may further include a data prediction unit that predicts swing data of a golf club having characteristics for which no swing data has been acquired by referring to the swing data classified by the classification unit.

According to this configuration, it is possible to predict the swing data of a golf club that the new player has not actually measured, based on the swing data that the new player has actually measured and other known swing data. Therefore, it is even more effective from the viewpoint of increasing appropriate swing data with a small number of trial hits.

In an embodiment of the present invention, it may further include a variance analysis processing unit (110) that performs variance analysis of swing data.

According to this configuration, it is possible to estimate where the swing action will be affected when the player changes the golf club.

In embodiments of the invention, a golf club may have three different characteristics, more specifically, shaft weight, golf club flex, and golf club tone. It's fine.

According to this configuration, it is possible to analyze the swing according to the characteristics of the golf club that are likely to be influenced by changes in the swing motion.

In an embodiment of the present invention, the swing data acquired by the swing data acquisition unit is either motion data when the golf club is swung, or an output value of a sensor attached to the golf club when the golf club is swung. It may be extracted from both.

This configuration is more effective from the standpoint of easily acquiring the player's swing motion with the sensor, and more effective from the standpoint of accurately acquiring the player's swing motion as motion data.

A swing analysis program according to an embodiment of the present invention is a swing analysis program for causing a computer to function as the above-described swing analysis system, and causes the computer to function as at least a swing data acquisition section and a singular value decomposition processing section. The swing data is individual swing data for a plurality of golf clubs having different golf club conditions, or individual swing data for a plurality of golf clubs having a plurality of different characteristics.

Further, the swing analysis method according to the embodiment of the present invention includes the steps of acquiring swing data of a golf club and performing singular value decomposition on the swing data. As the swing data, individual swing data of a plurality of golf clubs having different golf club conditions or individual swing data of a plurality of golf clubs having a plurality of different characteristics is used.

According to these configurations, in golf club swing analysis, it becomes possible to quantitatively express the influence that the characteristics of the golf club have on the swing motion.

1 Swing analysis system 10 Control unit 11a Sensor 11b Transmission unit 12 Motion capture 13 Display unit 14 Storage unit 15 Input unit 100 Swing data acquisition unit 101 Data reception unit 102 Measured value acquisition unit 103 Data interpolation unit 110 Variance analysis processing unit 120 Singular value Decomposition processing unit 130 Mode contribution rate acquisition unit 140 Multiple regression analysis processing unit 150 Clustering processing unit C1 First cluster C2 Second cluster C3 Third cluster C4 Fourth cluster C5 Fifth cluster C6 Sixth cluster

Claims (11)

a swing data acquisition unit that acquires individual swing data of a plurality of golf clubs having a plurality of different characteristics ;
a singular value decomposition processing unit that performs singular value decomposition processing on the swing data to decompose the swing data into singular values corresponding to a mode representing a unit motion that constitutes a swing motion of a golf club;
a mode contribution rate acquisition unit that acquires a mode contribution rate that is a contribution rate to the mode in the swing data;
a clustering processing unit that clusters the swing data using a self-organizing map;
The mode contribution rate is subjected to regression analysis processing, and the sign of the coefficient of determination, which is the degree of contribution of the characteristics of the golf club to the mode contribution rate, is positive or negative and the probability that the sign becomes positive or negative is obtained according to the cluster of the self-organizing map. a regression analysis processing unit,
has,
Swing analysis system. The swing analysis system according to claim 1, wherein the swing data acquisition unit further includes a swing data interpolation unit that interpolates the swing data using a response surface in response surface method. The swing analysis system according to claim 2, wherein the response curved surface is a swing response curved surface. The swing analysis system according to claim 2, wherein the response surface is a swing response surface and a time response surface. a swing data storage unit that stores swing data acquired by the swing data acquisition unit;
a data classification unit that classifies the swing data stored in the swing data storage unit;
a data prediction unit that predicts swing data of a golf club having characteristics for which no swing data has been acquired by referring to the swing data classified by the data classification unit;
The swing analysis system according to any one of claims 2 to 4, further comprising: The swing analysis system according to claim 1, further comprising a variance analysis processing unit that performs variance analysis of the swing data. The swing analysis system according to claim 1, wherein the golf club has three different characteristics. 8. The swing analysis system according to claim 7, wherein the golf club has three types of characteristics: weight of a shaft in the golf club, flex of the golf club, and condition of the golf club. The swing data acquired by the swing data acquisition unit is derived from one or both of motion data when the golf club is swung, and an output value of a sensor attached to the golf club when the golf club is swung. The swing analysis system according to any one of claims 1 to 8, wherein the swing analysis system is extracted. A swing analysis program for causing a computer to function as the swing analysis system according to claim 1,
The swing analysis program causes the computer to function as at least the swing data acquisition section, the singular value decomposition processing section, the mode contribution rate acquisition section, the clustering processing section, and the regression analysis processing section.
Swing analysis program. obtaining individual swing data for a plurality of golf clubs having a plurality of different characteristics ;
performing singular value decomposition on the swing data to decompose the swing data into singular values corresponding to a mode representing a unit motion constituting a swing motion of the golf club;
obtaining a mode contribution rate that is a contribution rate for the mode in the swing data;
clustering the swing data using a self-organizing map;
The mode contribution rate is subjected to regression analysis processing, and the sign of the coefficient of determination, which is the degree of contribution of the characteristics of the golf club to the mode contribution rate, is positive or negative and the probability that the sign becomes positive or negative is obtained according to the cluster of the self-organizing map. the step of
including,
Swing analysis method.

Images