Electric training device for identifying fatigue level of user performing isokinetic exercise on basis of machine learning model, control method therefor, and computer program

Abstract

Provided are an electric training device, a control method therefor, and a computer program. The control method according to an embodiment of the present disclosure comprises the steps of: generating a force profile for a user by detecting the force of the user through a handle part gripped by the user performing an isokinetic exercise, while controlling a driving part to generate a load according to the isokinetic exercise mode; extracting feature information from the force profile, inputting the extracted feature information into a pretrained machine learning model, and identifying the fatigue level of the user performing the isokinetic exercise; and adjusting the magnitude of the load on the basis of the identified fatigue level.

Description

Electric training device for identifying the fatigue level of a user during isokinetic exercise based on a machine learning model, the control method thereof, and computer program

[0001] The present disclosure relates to an electric training device and a control method thereof, and more specifically, to an electric training device that identifies the fatigue level of a user during isokinetic exercise based on a machine learning model, a control method thereof, and a computer program.

[0002] With the recent increase in social interest in wellness and health management, various exercise equipment aimed at improving individual physical fitness and function is being developed, and the demand for such equipment is continuously expanding. These devices are widely utilized in gyms, rehabilitation facilities, and home environments, and there is a demand for functional advancements to enhance user convenience and exercise efficiency. Conventional exercise equipment typically relied on fixed weights, requiring users to manually adjust exercise intensity or weight. This approach made it difficult to adequately reflect changes in the user's fitness level or physical condition during exercise. Furthermore, the inconvenience of having to pause movement to adjust intensity led to issues that degraded the exercise flow and user experience. Additionally, there is a possibility that improper intensity settings could reduce exercise effectiveness or increase the risk of injury.

[0003] To overcome these limitations, electric exercise equipment incorporating electronic control technology has recently been proposed. Electric exercise equipment offers the advantage of automatically generating loads through motors and control units, or performing exercises according to preset conditions. However, most conventional electric exercise equipment also relies on pre-set parameters or simple control rules, which limits its ability to adequately reflect the user's physical condition as it changes in real time during exercise. In particular, user fatigue accumulated during repetitive exercise significantly impacts performance ability and safety; however, conventional technology has been restricted in objectively assessing this fatigue state or utilizing it for the control of the exercise equipment. In most cases, exercise intensity is adjusted based on the user's subjective judgment, which entails problems such as significant variability among users and a lack of consistency. Therefore, there is a continuous demand for technology that can automate exercise intensity control and improve user convenience by more effectively utilizing user-related information generated during exercise.

[0004] The present disclosure aims to provide an electric training device that identifies the fatigue level of a user during isokinetic exercise based on a machine learning model, a control method thereof, and a computer program.

[0005] However, the problems to be solved in this disclosure are not limited to those mentioned above, and other unmentioned problems may be clearly understood based on the description below.

[0006] A control method for an electric training device that identifies the fatigue level of a user performing isokinetic exercise based on a machine learning model according to an embodiment of the present disclosure for realizing the aforementioned objectives comprises the steps of: generating a force profile of the user by detecting the force of the user through a handle portion grasped by the user performing the isokinetic exercise while controlling a drive unit to generate a load according to an isokinetic exercise mode; extracting feature information from the force profile and inputting the extracted feature information into a pre-trained machine learning model to identify the fatigue level of the user performing the isokinetic exercise; and adjusting the magnitude of the load based on the identified fatigue level.

[0007] Alternatively, the method includes a step of setting the number of sets and the number of repetitions for each set according to the isokinetic exercise mode, and the step of identifying the user's fatigue includes extracting feature information corresponding to each set from the user's force profile and inputting the extracted feature information into a pre-trained machine learning model to identify the user's fatigue corresponding to each set.

[0008] Alternatively, the step of identifying the fatigue corresponding to each of the above sets includes inputting the extracted feature information into the above-mentioned pre-trained machine learning model to identify the user's expected number of sets and identifying the fatigue based on the expected number of sets.

[0009] Alternatively, the step of adjusting the size of the load based on the identified fatigue level includes comparing the expected number of sets with the actual number of sets performed by the user to determine whether to adjust the size of the load.

[0010] Alternatively, the feature information includes at least one of work, mean force, peak force, minimum force, median force, rate of force development, time to peak force, variance of the force distribution, standard deviation, skewness, kurtosis, and root mean square (RMS).

[0011] Alternatively, the step of generating the user's force profile includes generating a set-specific force profile of the user performing the isometric exercise, excluding intervals where a force of less than a preset size is detected.

[0012] Alternatively, the step of generating the user's force profile includes detecting the user's force through a handle portion grasped by the user performing the isometric exercise, normalizing the user's force based on the user's body weight, and generating the user's force profile.

[0013] Alternatively, the method includes the step of selecting a machine learning model corresponding to the set number of sets and the number of repetitions for each set among a plurality of machine learning models trained according to the number of sets of the isokinetic exercise mode and the number of repetitions for each set.

[0014] Alternatively, the step of adjusting the size of the load may determine to increase the size of the load if the estimated number of sets is smaller than the actual number of sets performed by the user, to maintain the size of the load if the estimated number of sets matches the actual number of sets performed by the user, and to decrease the size of the load if the estimated number of sets is larger than the actual number of sets performed by the user.

[0015] Alternatively, in an electric training device that identifies the fatigue level of a user during isokinetic exercise based on a machine learning model,

[0016] An electric training device for identifying the fatigue level of a user during isokinetic exercise based on a machine learning model according to one embodiment of the present disclosure for realizing the aforementioned task comprises: one or more pairs of floor frames and a vertical frame; A strength training rack comprising a horizontal frame and a pair of vertical frames, an electric exercise assistance unit installed on each of the pair of vertical frames and generating an exercise load for the user, a main fitness unit connected to each of the electric exercise assistance units and configured to perform a predetermined fitness exercise, a sub fitness unit configured to be detachably connected to each of the main fitness units and configured to perform a predetermined fitness exercise, and a computing device comprising one or more processors for controlling the electric exercise assistance units, wherein while the computing device controls the drive unit of the electric exercise assistance unit to generate a load according to an isokinetic exercise mode, the device detects the force of the user through the handle of the main fitness unit held by the user performing the isokinetic exercise to generate a force profile of the user, extracts feature information from the force profile, inputs the extracted feature information into a pre-trained machine learning model to identify the fatigue level of the user performing the isokinetic exercise, and adjusts the magnitude of the load based on the identified fatigue level.

[0017] A computer program stored on a computer-readable storage medium according to one embodiment of the present disclosure for realizing the aforementioned objectives, wherein the computer program, when executed on one or more processors, performs a control operation of an electric training device that identifies the fatigue level of a user performing isometric exercise based on a machine learning model, wherein the operation includes, while controlling a drive unit to generate a load according to an isometric exercise mode, an operation of generating a force profile of the user by detecting the force of the user through a handle unit grasped by the user performing the isometric exercise, an operation of extracting feature information from the force profile and inputting the extracted feature information into a pre-trained machine learning model to identify the fatigue level of the user performing the isometric exercise, and an operation of adjusting the magnitude of the load based on the identified fatigue level.

[0018] According to one embodiment of the present disclosure, the load of an electric training device can be controlled in response to changes in the user's condition during exercise, thereby improving the user's exercise performance efficiency and safety. In particular, by more objectively identifying the user's fatigue state based on user-related data generated during exercise, the limitations of conventional methods that relied on the user's subjective judgment can be improved. Furthermore, by adjusting the magnitude of the load to reflect the fatigue state accumulated during the exercise process, the risk of injury caused by excessive load can be reduced, and the continuity of exercise and user convenience can be improved. Accordingly, a more stable exercise environment can be provided that takes into account differences in physical fitness levels or conditions among users.

[0019] FIG. 1 is a perspective view showing the overall configuration of an electric training device according to one embodiment of the present disclosure.

[0020] FIG. 2 is a front view showing the overall configuration of an electric training device according to one embodiment of the present disclosure.

[0021] FIG. 3 is a side view showing the overall configuration of an electric training device according to one embodiment of the present disclosure.

[0022] FIG. 4 is a plan view showing the overall configuration of an electric training device according to one embodiment of the present disclosure.

[0023] FIG. 5 is an attempt to illustrate the internal configuration of some components in an electric control type complex fitness machine according to the present invention.

[0024] Figure 6 is a drawing showing part "A" of Figure 5.

[0025] FIG. 7 is a perspective view showing the upper side of an electric training device according to one embodiment of the present disclosure.

[0026] FIG. 8 is a perspective view showing the lower side of an electric training device according to one embodiment of the present disclosure.

[0027] FIG. 9 is a block diagram showing a configuration that performs the function of an electric training device according to one embodiment of the present disclosure.

[0028] FIG. 10 is a flowchart of a control method for an electric training device that identifies the fatigue level of a user during isokinetic exercise based on a machine learning model according to one embodiment of the present disclosure.

[0029] FIG. 11 is an example diagram of training a machine learning model according to one embodiment of the present disclosure.

[0030] FIG. 12 is an example diagram of adjusting the size of the load of an electric training device based on fatigue identified through a machine learning model according to one embodiment of the present disclosure.

[0031] FIG. 13 is an example diagram of selecting a machine learning model according to a preset number of sets and the number of repetitions of the sets according to one embodiment of the present disclosure.

[0032] Embodiments of the present disclosure are described below with reference to the attached drawings so that those skilled in the art (hereinafter, those skilled in the art) can easily implement them. The embodiments presented in the present disclosure are provided to enable those skilled in the art to use or implement the contents of the present disclosure. Accordingly, various modifications to the embodiments of the present disclosure will be apparent to those skilled in the art. That is, the present disclosure may be embodied in various different forms and is not limited to the embodiments below.

[0033] Throughout the specification of the present disclosure, identical or similar reference numerals refer to identical or similar components. Additionally, to clearly explain the present disclosure, reference numerals in the drawings that are unrelated to the description of the present disclosure may be omitted.

[0034] The term “or” as used in this disclosure is intended to mean an implicit “or” rather than an exclusive “or.” That is, unless otherwise specified in this disclosure or its meaning is unclear from the context, “X uses A or B” should be understood to mean one of the natural implicit substitutions. For example, unless otherwise specified in this disclosure or its meaning is unclear from the context, “X uses A or B” may be interpreted as X using A, X using B, or X using both A and B.

[0035] The term “and / or” as used in this disclosure should be understood to refer to and include all possible combinations of one or more of the enumerated related concepts.

[0036] The terms “comprising” and / or “comprising” as used in this disclosure should be understood to mean the presence of certain features and / or components. However, the terms “comprising” and / or “comprising” should be understood not to exclude the presence or addition of one or more other features, other components and / or combinations thereof.

[0037] Where not otherwise specified in the present disclosure or where the context does not make it clear that the singular form is indicated, the singular should generally be interpreted as including "one or more."

[0038] The term "the N (N is a natural number)" used in this disclosure may be understood as an expression used to distinguish the components of this disclosure from one another according to certain criteria, such as functional perspectives, structural perspectives, or convenience of explanation. For example, components performing different functional roles in this disclosure may be distinguished as a first component or a second component. However, components that are substantially identical within the technical scope of this disclosure but need to be distinguished for the convenience of explanation may also be distinguished as a first component or a second component.

[0039] The term “acquisition” as used in this disclosure can be understood to mean not only receiving data through a wired or wireless communication network with an external device or system, but also generating data in an on-device form.

[0040] Meanwhile, the terms "module" or "unit" as used in this disclosure may be understood as referring to an independent functional unit that processes computing resources, such as a computer-related entity, firmware, software or a part thereof, hardware or a part thereof, or a combination of software and hardware. In this case, "module" or "unit" may be a unit composed of a single element, or a unit expressed as a combination or set of multiple elements. For example, in a narrow sense, "module" or "unit" may refer to a hardware element of a computing device or a set thereof, an application program that performs a specific function of software, a procedure implemented through software execution, or a set of instructions for program execution. Furthermore, in a broad sense, "module" or "unit" may refer to the computing device itself that constitutes the system, or an application running on the computing device. However, since the above-described concept is merely an example, the concepts of "module" or "part" may be defined in various ways within the scope understandable to those skilled in the art based on the contents of this disclosure.

[0041] As used in this disclosure, the term "model" may be understood as a system implemented using mathematical concepts and language to solve a specific problem, a set of software units to solve a specific problem, or an abstract model regarding a processing process to solve a specific problem. For example, a neural network "model" may refer to an overall system implemented as a neural network that possesses problem-solving capabilities through learning. In this case, the neural network may possess problem-solving capabilities by optimizing parameters connecting nodes or neurons through learning. A neural network "model" may include a single neural network or a set of neural networks composed of multiple neural networks.

[0042] The term "data" as used in this disclosure may include "image," "signal," etc. The term "image" as used in this disclosure may refer to multidimensional data composed of discrete image elements. In other words, "image" may be understood as a term referring to a digital representation of an object visible to the human eye. For example, "image" may refer to multidimensional data composed of elements corresponding to pixels in a two-dimensional image. "Image" may refer to multidimensional data composed of elements corresponding to voxels in a three-dimensional image.

[0043] The explanation of the foregoing terms is intended to aid in understanding the present disclosure. Accordingly, it should be noted that unless a foregoing term is explicitly stated as a matter limiting the content of the present disclosure, it is not to be used in the sense of limiting the technical concept of the content of the present disclosure.

[0044] FIG. 1 is a perspective view showing the overall configuration of an electric training device according to one embodiment of the present disclosure.

[0045] FIG. 2 is a front view showing the overall configuration of an electric training device according to one embodiment of the present disclosure.

[0046] FIG. 3 is a side view showing the overall configuration of an electric training device according to one embodiment of the present disclosure.

[0047] FIG. 4 is a plan view showing the overall configuration of an electric training device according to one embodiment of the present disclosure.

[0048] FIG. 5 is an attempt to illustrate the internal configuration of some components in an electric control type complex fitness machine according to the present invention.

[0049] Figure 6 is a drawing showing part "A" of Figure 5.

[0050] FIG. 7 is a perspective view showing the upper side of an electric training device according to one embodiment of the present disclosure.

[0051] FIG. 8 is a perspective view showing the lower side of an electric training device according to one embodiment of the present disclosure.

[0052] An electric training device (1000) according to one embodiment of the present disclosure, as shown in FIGS. 1 to 8, comprises a rack for strength training that forms the shape and skeleton of fitness equipment and is formed by including a pair of floor frames (110), a vertical frame (120), and a horizontal frame (130); an electric exercise assist unit (200) that generates an exercise load and is installed on each of the parallel pair of vertical frames (120) that form the rack for strength training; a main health unit (300) connected to each of the electric exercise assist units (200); and one or more sub health units that are detachably provided to each of the main health units (300).

[0053] A strength training rack comprises a bottom frame (110) which is a pair of first-direction frames extending in a first direction (D1), a pair of front vertical frames and a pair of rear vertical frames (120) which are second-direction frames extending in a second direction (D2) perpendicular to the first direction (D1), and a horizontal frame (130) which is a third-direction frame extending along a third direction (D3) intersecting the second direction (D2) and provided between the tops of a pair of front second-direction frames (120).

[0054] The electric exercise assist unit (200) may include a housing (210) that is fixed parallel to each of the vertical frames (120), has an internal space, is formed on a surface facing the main health unit (300), and is connected to the main health unit (300) to guide it, and a driving unit that is positioned at one end (upper part in the drawing) of the housing (210) and generates power.

[0055] The drive unit includes a motor (220) configured to transmit an encoder signal, a ball screw (230) disposed in the internal space of the housing (210) and connected to the motor (220) to transmit power generated from the motor (220) and extending in an up-and-down direction (a second direction (D2) or a direction opposite to the second direction (D2)), each ball screw nut block (240) coupled to the threads of the ball screw (230) and movable in an up-and-down direction, a computing device (not shown) that calculates the torque to be generated by the motor (220) and transmits a control signal to the motor (220), and a display (250) capable of inputting exercise information such as the type of exercise and the degree of exercise intensity, and displaying the exercise situation and results.

[0056] The housing (210) includes a coupling plate member (211) comprising fixing members (not shown) to which a motor (220) can be fixed, a "C-shaped" rectangular member (212) having a guide slot (211a) extending along a second direction (D2) to guide the movement of the main health unit (300), and a support cover member (213) that rotatably supports a ball screw (230) at the upper and lower ends, respectively (see FIG. 5 and FIG. 6).

[0057] The motor (220) is covered by a motor housing and can transmit its power to the ball screw (230) through a gear train that receives and transmits the power to the ball screw (230). The motor (220) transmits an encoder signal and is driven by a control signal transmitted from a computing device.

[0058] The ball screw nut block (240) may be positioned on the threads of each ball screw (230) to be movable in the second direction (D2) or the opposite direction of the second direction (D2), and may be connected to each main health unit (300) to move the main health unit (300). For example, each ball screw nut block (240) may move in the second direction (D2) or the opposite direction of the second direction (D2) in a predetermined distance unit (e.g., 1 mm) according to the operation of each motor (220). Meanwhile, not limited thereto, the drive unit may include, in addition to the ball screw (230) and the ball screw nut block (240), a belt (not shown) that transmits power generated from the motor (220) and a pulley (not shown) that supports and rotates the belt.

[0059] The computing device receives exercise information such as user information, exercise type and intensity, and receives encoder signals transmitted from the motor (220) to control the motor (220), thereby enabling the user to selectively or in combination perform multiple modes of exercise.

[0060] Specifically, the computing device controls the execution of an isotonic exercise implementation mode, an isotonic and isokinetic exercise implementation mode, a bidirectional isokinetic exercise implementation mode, and an elastic exercise implementation mode. The isotonic exercise implementation mode is a mode that allows for weight exercises such as those using a standard weight, and controls the weight using the torque of a motor. The isotonic and isokinetic mode may be a mode that controls the speed isokinetically when the user grips and lifts the bar (or handle) of the main health unit (300) to increase the load above a set value and maintain a constant exercise speed, and switches to controlling the exercise weight isotonically when lowering. The isotonic and isokinetic mode is intended to enhance the muscle exercise effect by performing pushing or pulling movements with muscle strength during concentric contraction, and lowering or raising movements while the muscle resists during eccentric contraction. The bidirectional isokinetic exercise implementation mode is a mode that controls both pushing and pulling movements isokinetically, and is an exercise mode that allows training of muscles in two parts simultaneously. Such exercise modes can be performed by conventional hydraulic mechanisms, but the user may feel a sense of unfamiliarity during exercise due to the compressible material contained in the hydraulic mechanism. In contrast, the electric training device (1000) according to one embodiment of the present disclosure has the advantage of having no sense of unfamiliarity during pushing and pulling exercise, as it has low backlash and high responsiveness to control movements using a ball screw. The elastic exercise implementation mode is an exercise mode that controls the motor torque so that the load increases according to distance, such as with a spring or a rubber band. This is an exercise method mainly used by athletes and can be implemented by setting it according to the exercise style. Each exercise mode can be implemented selectively or in combination. Meanwhile, the exercise mode can be set based on control commands obtained through a display or user input interface.

[0061] Meanwhile, in the control process executed by the computing device, information regarding the torque applied to the main health unit (300), which will be described later, and / or the movement speed of the main health unit (300) may be obtained to control the exercise. Information regarding the movement speed may be obtained, for example, based on the encoder signal of the motor (220). Here, information regarding the movement speed may be obtained by considering the movement section. That is, if the movement section of the main health unit (300) is classified as a single section, the movement speed may be obtained based on the main health unit (300) completing one round trip. Alternatively, if the movement section of the main health unit (300) is set as multiple sections, the movement speed for each section may be obtained. For example, the movement section may be divided into three sections—initial, middle, and final—with respect to the one-way section, and information regarding the movement speed of each section may be obtained. At this time, the initial section may be defined as a section where acceleration is generated in the direction of movement of the main health unit (300), the middle section as a section where the main health unit (300) moves at a constant speed in the direction of movement, and the later section as a section where acceleration of the main health unit (300) is generated in the opposite direction to the movement, with different speed profiles. As the exercise section is subdivided, the user's exercise ability status can be determined in a more detailed manner.

[0062] The computing device determines whether the movement speed of the main health unit (300) is below a predetermined speed. The computing device can determine whether the movement speed of the main health unit (300) is below a predetermined speed based on the movement speed information obtained during the process. This may be to determine whether the user's muscle strength is depleted. More specifically, when the user pushes the main health unit (300) upward (isokinetic exercise) or pulls it (isotonic exercise), if strength is depleted after a predetermined number of repetitions, this can be determined through the movement speed of the main health unit (300). Accordingly, the computing device can determine whether the user needs assistance.

[0063] At this time, the predetermined speed can be determined by various criteria. For example, the predetermined speed can take into account the user's previous exercise information. That is, the predetermined speed can be set to a percentage or less of the movement speed of the main health unit in the previous exercise (initial exercise). For example, the predetermined speed can be 50% or less of the movement speed of the main health unit (300) in the previous exercise. Also, for example, the predetermined speed can be a specific value. Additionally, the predetermined speed can be determined according to the exercise segment of the main health unit (300). For example, if the exercise segment of the main health unit consists of three segments, the predetermined speed can be set for each segment. More specifically, when considering the general movement of the main health unit (300), the predetermined speed can have a profile of increasing speed in the initial segment, constant speed in the middle segment, and decreasing speed in the later segment. When the computing device determines that the movement speed of the main health unit (300) is lower than a predetermined speed standard, it operates the motor (220) to perform exercise assistance control. That is, when the computing device determines that exercise assistance is needed for the user, it increases the generated torque of the motor (220) to reduce the user's weight resistance. At this time, the computing device can determine how much exercise assistance to provide through the motor (220). For example, the computing device can increase the degree of exercise assistance (increase the generated torque) as the weight resistance felt by the user is relatively greater, and conversely, if the weight resistance felt by the user is relatively smaller, it can decrease the degree of exercise assistance (increase the generated torque). In addition, the computing device can determine exercise assistance information by considering the user's maximum muscle strength. For example, for a user with high maximum muscle strength, the degree of exercise assistance (generated torque) can be reduced, and for a user with low maximum muscle strength, the degree of exercise assistance can be increased.Thus, the electric training device (1000) according to one embodiment of the present disclosure can perform exercise control optimized for the user's condition. Additionally, the computing device can determine the degree of control based on the user's exercise history information. For example, the computing device may decide not to provide assistance through the motor (220) for exercises that the user achieved in a previous exercise. More specifically, the computing device may determine that it is not necessary to provide assistance for exercises of 3 sets of 15 repetitions each with a weight of 10 kg that a specific user previously achieved.

[0064] The display, combined with a touch panel, can receive and store user information (e.g., height, weight, etc.) and exercise information (exercise mode, applied exercise weight, whether both main fitness units are used simultaneously or independently, etc.) before exercise, and display exercise status (repetitions, repetition / set alarm, etc.) and exercise results.

[0065] The main health unit (300) includes a coupling block (310) that is fixed to each ball screw nut block (320) through a guide slot (211a) of the exercise assistance unit (200), a weight block (320) that is configured to extend in a direction facing each other from each coupling block (310) and contains a weight having a predetermined weight, and a bar (330) that is fixed to each coupling block (310) parallel to the weight block (320) or connected and fixed to two coupling blocks (310) or a plate that is tiltably coupled.

[0066] Using the main health unit (300), isokinetic and isotonic exercises for various body muscles, including bench press, incline chest exercise machine, decline chest exercise machine, shoulder press, Smith machine, power leg press, squat, etc., can be performed.

[0067] According to one embodiment of the present disclosure, the main health unit (300) may include at least one sensor for measuring the user's strength. For example, the sensor may be implemented as a load cell.

[0068] The sub-health unit is described with reference to FIGS. 7 and 8. The sub-health unit may be configured as a cable fitness tool (a type of cable machine) that can perform fitness exercises such as deadlifts, preacher curls, and cable crossovers in front of a strength training rack through a cable member that is connected to the main fitness unit (300) as a first sub-health unit and guided forward. Specifically, the sub-health unit is a first sub-health unit and comprises a plurality of pulley members (410) rotatably provided on the inner side of each of a pair of floor frames (110) of a strength training rack, a cable member (first cable member) (420) which is detachably fixed to one side of the main health unit (300) (specifically, the lower surface of the coupling block (310)) and extends forward of the floor frame (110) through the plurality of pulley members (410), a bar member (430) detachably coupled to the other end of the cable member (420), and detachable means for detachably coupling the one end of the cable member (420) to the main health unit (300) and the other end of the cable member (420) to the bar member (430).

[0069] A plurality of pulley members (410) include a first pulley member (411) provided on a floor frame (110) located directly below the main health unit (300) (specifically, directly below the coupling block (310)), and a second pulley member (412) provided at the front end of the floor frame (110). Although the drawing illustrates the configuration of two pulley members (411, 412), it is not limited thereto. The bar member (430) is illustrated as being composed of a single bar member (430) in which the other end of each cable member (420) is detachably connected to both ends, but it may be composed of separate bar members in which the other end of each cable member (420) is detachably connected to the central part. Accordingly, when the bar member (430) in the sub-health unit (first sub-health unit) of the present invention is connected to the cable member (420) respectively, the bar member can be held in one hand and exercised. When the bar member (430) is implemented as a single bar member, it is connected to each cable member together, so the weight is doubled, allowing the type of exercise to be changed. That is, while using the main health unit (300), it is possible to perform exercises that can achieve different effects simply by changing the configuration of the bar member (430).

[0070] The detachable means may consist of an annular ring formed on one end and the other end of the cable member (420) and one of the main health unit (300) and the bar member (430), and a hook (hook-type ring) formed on the other end of the cable member (420) and one of the main health unit (300) and the bar member (430). Additionally, the detachable means may further include a hook (510) (specifically, one side of the vertical frame (120)) (see FIG. 7) provided on one side of the strength training rack so that when the sub-health unit is not in use, each end of the cable member (420) can be detached from the main health unit (300) and stored. Here, it is preferable that the hook (510) be formed on the other end of an elastic rubber band or spring, one end of which is fixed to the vertical frame (120).

[0071] Additionally, a second sub-health unit may be further included as a sub-health unit according to one embodiment of the present disclosure. The second sub-health unit may be configured as a cable health tool (a type of cable machine) capable of performing health exercises such as butterfly and high pulley from the upper part of a strength training rack by guiding a cable member connected to the main health unit (300) to the rear upper part. Specifically, the sub-health unit is a second sub-health unit and comprises a plurality of pulley members (411, 413) rotatably provided on the inner side of each of a pair of floor frames (110) of the strength training rack, end pulley members (not shown) provided at each end of the horizontal frame (130) of the strength training rack, a pair of central pulley members (414) provided in the central part of the horizontal frame (130) of the strength training rack, an extension frame (440) provided extending forward from the front of the central part of the horizontal frame (130) of the strength training rack, an extension end pulley member (415) provided at the free end of the extension frame (440), and one end is respectively detachably fixed to one side of the main health unit (300) (specifically, the lower surface of the coupling block (310)), and extends forward through the plurality of pulley members (411, 413) and the end pulley members, or the plurality of pulley members (411, 413) and a cable member (second cable member) (421) extending forward through a pair of central pulley members (414) and an extension end pulley member (415), a bar member (431) detachably coupled to the other end of the cable member (second cable member) (421), and a detachable means for detachably coupling one end of the cable member (421) to the main health unit (300) and the other end of the cable member (421) to the bar member (431).

[0072] Additionally, the second sub-health unit may have an end extension frame having the same configuration as the extension frame (440) having an extension end pulley member (415) at the end, each provided at both ends of the horizontal frame (130). Accordingly, the cable member (421) may be configured to be guided to each end extension frame via a plurality of pulley members (411, 413) and end pulley members, and extended forward.

[0073] A plurality of pulley members (411, 413) include a first pulley member (411) provided on a floor frame (110) located directly below the main health unit (300) (specifically, directly below the coupling block (310)), and a third pulley member (413) provided on a floor frame (110) on the rear side of the first pulley member (411). Two pulley members (411, 413) are shown, but are not limited thereto.

[0074] The bar member (431) of the second sub-health unit may be formed as a single bar member and implemented so that the other end of each cable member (second cable member) (421) is detachably connected to the central part or both ends. Here, when the other end of the cable member (second cable member (421)) is detachably connected to the central part of the bar member (431), it is connected via the extension frame (440) on the central side. In addition, when the other end of the cable member (second cable member (421)) is detachably connected to both ends of the bar member (431), it is connected via the end pulley members at both ends of the horizontal frame (130) (or via the end extension frame if an end extension frame is configured). Here, when the other end of the cable member (second cable member (421)) is guided through both ends of the horizontal frame (130), the bar member (431) may be composed of separate bar members (a pair of bar members) in which the other end of each cable member (second cable member) (421) is separately and detachably connected to the central part. Accordingly, when the bar member (431) in the second sub-health unit is composed of a pair of bar members and connected to the other end of the cable member (421), one can exercise by holding the pair of bar members one by one in each hand, and when the bar member (431) is composed of a single bar member, the weight is doubled by connecting it to each cable member, allowing one to change the type of exercise. That is, while using the main health unit (300), one can perform exercises that achieve different effects simply by changing the configuration of the bar member (431).

[0075] The detachable means may consist of an annular ring formed on one end and the other end of a cable member (second cable member) (421) and one of the main health unit (300) and the bar member (431), and a hook (hook-type ring) formed on the other end of a cable member (second cable member) (421) and one of the main health unit (300) and the bar member (431). Additionally, the detachable means may further include a hook (510) (specifically, one side of the vertical frame (120)) (see FIG. 7) provided on one side of the strength training rack so that when the sub-health unit is not in use, each end of the cable member (second cable member) (421) can be detached from the main health unit (300) and stored. Here, it is preferable that the hook (510) be formed on the other end of an elastic rubber band or spring, one end of which is fixed to the vertical frame (120). A second sub-health unit may be configured in place of a first sub-health unit or together with the first sub-health unit. The drawings of the present invention illustrate the latter case.

[0076] FIG. 9 is a block diagram showing a configuration that performs the function of an electric training device (1000) according to one embodiment of the present disclosure. However, since FIG. 2 is merely an example, the training device (1000) may include other configurations for implementing the training function, or only some of the configurations disclosed in FIG. 9 may be included in the training device (1000).

[0077] Referring to FIG. 9, a training device (1000) according to one embodiment of the present disclosure may be a block diagram showing some of the functional configurations of the training device (1000) of FIG. 1. An electric training device (1000) according to one embodiment of the present disclosure may include a computing device (610), a sensor (620), a driving unit (630), and a display (640). Since the display (640) of FIG. 9 may correspond to the display (150) of FIG. 1, a detailed description is omitted.

[0078] The computing device (610) may be built into a case placed on the back of the display (250) of the electric training device (1000) of FIG. 1 or placed on a bottom frame. However, it is not limited thereto, and the computing device (610) may be placed outside the electric training device (1000) by being network-connected through a communication interface rather than inside.

[0079] A computing device (610) according to one embodiment of the present disclosure may be a hardware device or part of a hardware device that performs comprehensive processing and computation of data, or it may be a software-based computing environment connected to a communication network. The computing device (610) may be structurally included in a training device (1000), or it may be implemented as an external electronic device that performs data communication by being connected to the training device (1000) via wired or wireless means.

[0080] The computing device (610) may be a server or a client that performs intensive data processing functions and shares resources through communication with the training device (1000) described above. Additionally, the computing device (610) may be a cloud system connected to the training device (1000) described above, enabling multiple servers and clients to process data comprehensively. Since the description above is merely one example regarding the type of computing device (610), the type of computing device (610) may be configured in various ways within a range understandable to those skilled in the art based on the contents of this disclosure.

[0081] Referring to FIG. 9, a computing device (610) according to one embodiment of the present disclosure may include a processor (611), a memory (612), and a network unit (613). However, since FIG. 9 is merely an example, the computing device (610) may include other components for implementing a computing environment. Additionally, only some of the disclosed components may be included in the computing device (610).

[0082] A processor (611) according to one embodiment of the present disclosure is electrically connected to one or more sensors (620), a driving unit (630), and a display (150), including a memory (612) and a communication interface (613), so as to control the overall operation of the training device (1000).

[0083] A processor (611) according to one embodiment of the present disclosure may be understood as a constituent unit comprising hardware and / or software for performing computing operations. For example, the processor (611) may process instructions generated as a result of user interaction through a user interface. A processor (611) for performing such data processing and operations may include a central processing unit (CPU), a general purpose graphics processing unit (GPGPU), a tensor processing unit (TPU), an application specific integrated circuit (ASIC), or a field programmable gate array (FPGA). Since the above-described type of processor (611) is merely an example, the type of processor (611) may be configured in various ways within a range understandable to those skilled in the art based on the contents of the present disclosure.

[0084] A memory (612) according to one embodiment of the present disclosure may be understood as a configuration unit comprising hardware and / or software for storing and managing data processed by a computing device (610). That is, the memory (612) may store data of any form generated or determined by a processor (611) and data of any form received by a communication interface (613). For example, the memory (612) may include at least one type of storage medium among a flash memory type, a hard disk type, a multimedia card micro type, a card type memory, RAM (random access memory), SRAM (static random access memory), ROM (read-only memory), EEPROM (electrically erasable programmable read-only memory), PROM (programmable read-only memory), magnetic memory, a magnetic disk, or an optical disk. Additionally, the memory (612) may include a database system that controls and manages data in a predetermined system. Since the above-described type of memory (612) is merely an example, the type of memory (612) can be configured in various ways within a range understandable to those skilled in the art based on the contents of the present disclosure.

[0085] Memory (612) can manage data, combinations of data, and program code executable by the processor (611) by structuring and organizing them for the processor (611) to perform operations. For example, memory (612) can store program code that enables the processor (611) to process images, program code that enables the processor (611) to process commands based on user input through a user interface, and various data generated as the program code is executed.

[0086] The memory (612) may include at least one exercise mode information of the training device (1000), information on the type and size of the load according to the exercise mode, user information, a pre-trained machine learning model, etc.

[0087] A communication interface (613) according to one embodiment of the present disclosure may be understood as a configuration unit that transmits and receives data through any known form of wired or wireless communication system. For example, the communication interface (613) may perform data transmission and reception using wired or wireless communication systems such as a local area network (LAN), wideband code division multiple access (WCDMA), long term evolution (LTE), wireless broadband internet (WiBro), 5th generation mobile communication (5G), ultrawide-band communication, ZigBee, radio frequency (RF) communication, wireless LAN, wireless fidelity, near field communication (NFC), or Bluetooth. Since the communication systems described above are merely examples, wired or wireless communication systems for data transmission and reception of the communication interface (613) may be applied in various ways other than those described above.

[0088] The communication interface (613) can receive data necessary for the processor (611) to perform calculations through wired or wireless communication with any system or any client, etc. Additionally, the communication interface (613) can transmit data generated through the calculations of the processor (611) through wired or wireless communication with any system or any client, etc. The communication interface (613) can perform wired or wireless communication with a server managing a training device or another training device outside the training device (1000). Alternatively, if the computing device (610) is configured outside the training device (1000), the communication interface (613) can transmit and receive data to control the operation of the training device (1000) and implement a user interface through communication with the training device (1000).

[0089] One or more sensors (hereinafter, sensors) (620) can measure position, velocity, and acceleration, which are information generated by the user while exercising, and the force provided to the training device (1000) by the user operating the training device (1000). Specifically, the sensors (620) can measure the magnitude of the load transmitted from the user through the bar of the main health tool unit (i.e., the user's force), etc. The sensors (620) may include an encoder, a load cell, and a force sensor.

[0090] Additionally, the sensor (620) may detect the position of the main health tool unit or the movement speed of the main health tool unit. Specifically, for this purpose, the sensor (620) may include an IMU sensor, an accelerometer, a laser sensor, etc.

[0091] The drive unit (630) may include one or more motors (hereinafter referred to as motors) for providing exercise load to a user and a controller for controlling the motors. The controller may control the motors according to control commands transmitted from the processor (611). However, it is not limited thereto, and the processor (611) and the controller of the drive unit (630) may be integrated into a single processor or a single control module. The drive unit (630) may include cables, wires, etc., connecting the motors to the main health unit (300).

[0092] FIG. 10 is a flowchart of a control method for an electric training device that identifies the fatigue level of a user during isokinetic exercise based on a machine learning model according to one embodiment of the present disclosure.

[0093] According to one embodiment of the present disclosure, while the processor (611) controls the drive unit (630) to generate a load according to the constant speed exercise mode, it can generate a force profile of the user by detecting the user's force through the handle unit (e.g., the bar (330) of the main health unit (300)) that the user is gripping while performing the constant speed exercise (S1010).

[0094] The isometric exercise mode may include the aforementioned isotonic and isometric exercise implementation modes and bidirectional isometric exercise implementation modes. When the isometric exercise mode is set among the plurality of exercise modes of the electric training device, the processor (611) can control the drive unit to generate a load according to the isometric exercise mode. The isometric exercise mode may be set according to a control command input through the user interface or display (640) of the electric training device.

[0095] Meanwhile, the processor (611) may receive target weight information through the user interface or display (640) of the electric training device. At this time, the processor (611) may set a target weight based on the target weight information and control the drive unit (630) to generate resistance corresponding to the target weight. Resistance may be provided in a direction opposite to the direction in which the user pushes or pulls the main health unit (300). Specifically, the processor (611) may calculate a reference resistance corresponding to the set target weight and generate the reference resistance by controlling the torque applied to the motor included in the drive unit (630). At this time, the processor (611) may identify the movement speed and direction of movement of the main health unit (300) in real time based on the encoder signal of the motor, and provide resistance corresponding to the preset target weight by correcting the torque applied to the motor using a feedback control method according to the direction and change of the movement speed. Meanwhile, the feedback control method may include, but is not limited to, a proportional-integral-derivative (PID) control method. Accordingly, changes in speed that occur while the user performs exercise against resistance corresponding to the target weight are mitigated by feedback control, and as a result, the movement speed of the main health unit (300) can be maintained within a constant range. That is, the processor (611) can provide an isokinetic exercise mode through feedback control of the resistance while generating resistance corresponding to the target weight.

[0096] The processor (611) can continuously acquire the force acting on the handle portion (e.g., bar (330) or handle) held by the user during the performance of constant-speed motion through the sensor (620) over time. The processor (611) can sense the force transmitted from the user through the sensor (620), which includes a load cell or a force sensor, at a preset sampling period (e.g., 50 Hz). At this time, the processor (611) can generate the user's force-profile using the time-series data of the force acquired through the sensor (620).

[0097] Here, a force profile can be defined as a set of force values ​​arranged continuously along a time axis while an isometric exercise is being performed. A force profile can be generated in units of repetitions and / or sets. For example, the processor (611) may identify each repetition interval based on changes in the position of the main health unit (300), velocity information, or start and end events of the exercise motion, and group force profiles corresponding to multiple repetitions into a single set to generate a force profile per set, or the processor (611) may generate a force profile for each repetition interval. In addition, the processor (611) may obtain not only force data but also position data or velocity data of the main health unit (300) from the sensor (620). The processor (611) may also generate a velocity profile per set or a position profile per set using the velocity or position data. According to one embodiment of the present disclosure, the processor (611) may generate a set-specific force profile of a user performing an isometric exercise, excluding a section in which a force of less than a preset size is detected. Specifically, to generate the user's force profile, the processor (611) may exclude a section in which the force detected through the sensor (620) is less than a preset size. For example, the processor (611) may determine the point in time when the user's force is detected to be greater than or equal to a preset size as the effective start point of the isometric exercise, and determine the point in time when the force decreases again to less than a preset size as the end point of the corresponding exercise.

[0098] That is, the processor (611) can determine the section where the detected force is maintained above a preset size as an effective motion section where the user actually exerts force against the load, and determine the section where the force decreases below a preset size as a relaxation section following the end of the repetitive motion and exclude it. Accordingly, the processor (611) can generate a force profile using only the force data corresponding to the section where the user actually exerts force against the load.

[0099] Additionally, according to one embodiment of the present disclosure, the processor (611) can detect the user's force through a handle portion held by a user performing an isometric motion, normalize the user's force based on the user's body weight, and generate the user's force-profile.

[0100] Specifically, the processor (611) can normalize the user's force detected through the handle portion held by the user performing isometric exercise based on the user's body weight. The processor (611) can correct the force value obtained from the sensor (620) to the body weight based on the user's body weight information input through the display (640) or user input interface or stored in memory (612). The processor (611) can generate a force profile by normalizing the sensed force based on the following first equation. Then, the processor (611) can generate the user's force profile by arranging the normalized force values ​​in chronological order.

[0101] (Equation 1)

[0102] Here, F is the force measured through the sensor (620), m is the user's weight, and b is an allometric parameter, which can be, for example, 0.67. The processor (611) normalizes the force A user's force profile can be generated by arranging them in chronological order. Through this normalization process, the influence of differences in absolute force magnitude between users with different body weights is reduced, and a force profile reflecting the relative force expression characteristics of each user can be generated.

[0103] Meanwhile, the processor (611) can set the number of sets and the number of repetitions for each set according to the isokinetic exercise mode. Specifically, the processor (611) can set the number of sets to be performed by the user and the number of repetitions included in each set based on exercise program information input through the user interface or display (640) of the electric training device. While isokinetic exercise is being performed, the processor (611) can identify each repetition segment based on the force profile obtained through the sensor (620), the position information of the main health unit (300), movement speed information, or start and end events of the exercise motion. Furthermore, the processor (611) can generate a set-specific force profile by grouping the force profiles corresponding to multiple repetition segments into sets. Accordingly, the processor (611) can analyze the changes in the user's force output characteristics according to the progress of each set by chronologically separating them.

[0104] And, the processor (611) can extract feature information from the force-profile and input the extracted feature information into a pre-trained machine learning model to identify the fatigue level of a user performing isometric exercise (S1020).

[0105] The processor (611) can analyze time-force time-series data included in the force-profile for each set to produce one or more feature information that reflects the user's muscle strength expression state and change pattern. The feature information may be values, indicators, etc. that quantitatively express the magnitude, distribution, and temporal change characteristics of the force-profile. The processor (611) can acquire the extracted feature information as a single feature vector and input the extracted feature information into a pre-trained machine learning model. At this time, the machine learning model can identify the user's fatigue level corresponding to the set or the progress state of the set based on the input feature information.

[0106] In particular, according to one embodiment of the present disclosure, the feature information may include at least one of work, mean force, peak force, minimum force, median force, rate of force development, time to peak force, variance of the force distribution, standard deviation, skewness, kurtosis, and root mean square (RMS).

[0107] The processor (611) can analyze time-force time-series data included in the generated set-specific force-profile to produce one or more feature information that quantitatively represents the user's muscle strength expression state and fatigue accumulation pattern. Specifically, the processor (611) can produce feature information by performing statistical operations, temporal change analysis, and accumulation operations on multiple force values ​​included in the set-specific force-profile. For example, the processor (611) can calculate the workload by integrating the force values ​​included in the force-profile corresponding to a specific set along the time axis, and calculate the average force by calculating the average of the force values. Additionally, the processor (611) can calculate the maximum force and minimum force by detecting the maximum and minimum values, respectively, in the force-profile corresponding to a specific set, and calculate the median force by sorting the force values ​​in order of magnitude and selecting the value located in the center. The processor (611) can calculate the time to peak force by calculating the time from the start of the iteration interval included in a specific set until the time the force reaches its maximum value, and can calculate the rate of force development based on the amount of force change relative to time in the section where the force increases. Meanwhile, the processor (611) can calculate the variance and standard deviation by analyzing the distribution characteristics of the force values ​​included in the force-profile corresponding to the specific set, and can calculate the skewness and kurtosis, respectively, which indicate the degree of asymmetry and concentration of the force value distribution. In addition, the processor (611) can calculate the root mean square (RMS) value by calculating the square root of the mean square of the force values. In this way, the processor (611) can construct a feature vector per set or per iteration by combining one or more calculated feature information, and can provide the feature vector as input data for a machine learning model in a subsequent step.

[0108] FIG. 11 is an example diagram of training a machine learning model according to one embodiment of the present disclosure.

[0109] Meanwhile, the machine learning model may include, as an example, a Random Forest model, but is not limited thereto. For example, the machine learning model may be a decision tree-based model, a support vector machine, a neural network model, or a combination thereof. To train the machine learning model, the processor (611) may prepare training data based on force profiles collected from multiple users performing isokinetic exercise. Referring to FIG. 11, the training data may include feature information extracted from a set-by-set force profile (or a repetition-by-repetition force profile) obtained during the performance of isokinetic exercise, and label data corresponding to the feature information. Here, the label data may be information indicating the set number, the set progression stage, or the fatigue state in the isokinetic exercise. The processor (611) may train the machine learning model using the training data to estimate the set number corresponding to the feature information based on the input feature information. Accordingly, when feature information extracted from a set-by-set force profile during actual exercise performance is input, the trained machine learning model may output which set the feature information corresponds to during the user's exercise process.

[0110] As described above, the training data may include a force profile with weight-based normalization applied and force data from which force intervals below a preset size have been excluded. Accordingly, the machine learning model can learn the relationship between feature information and fatigue status while minimizing the influence of weight differences between users and invalid intervals. The machine learning model trained in this way can take feature information extracted during actual exercise performance as input and identify the fatigue level of a user performing isokinetic exercise in real time.

[0111] Meanwhile, the processor (611) may train a machine learning model using not only the force-profile but also the aforementioned velocity-profile or position-profile. That is, the processor (611) may train a machine learning model by configuring training data with feature information extracted from the set-specific velocity-profile (or position-profile) and label data assigned a set number. In this regard, the above description applies equally, so a detailed explanation is omitted.

[0112] In particular, according to one embodiment of the present disclosure, the processor (611) can extract feature information corresponding to each set from the user's force-profile and input the extracted feature information into a pre-trained machine learning model to identify the fatigue level corresponding to each set of the user.

[0113] Specifically, the processor (611) can analyze time-force time series data included in the force-profile generated corresponding to each set to extract feature information reflecting the user's muscle strength expression state and degree of fatigue accumulation in the corresponding set. The processor (611) can produce one or more feature information corresponding to the set by performing statistical operations, temporal change analysis, or cumulative operations on all or part of the repetition intervals included in each set.

[0114] At this time, the processor (611) can reflect in the feature information aspects such as a decrease in force magnitude, a change in the speed of force expression, and a deformation of the force distribution that occur as the number of repetitions progresses within the same set. For example, the processor (611) can calculate the rate of decrease in average force, the amount of change in maximum force, or the degree of decrease in the force expression rate by comparing the force profiles at the beginning and end of the set, and these values ​​can be used as indicators representing the accumulation of fatigue in the set.

[0115] The processor (611) can identify the fatigue level of a user corresponding to a set by configuring the feature information calculated in this way into a feature vector in sets and providing the feature vector as input to a pre-trained machine learning model. Accordingly, the processor (611) can precisely distinguish and determine the fatigue state of a user that appears differently for each set, even if it is the same exercise.

[0116] And, the processor (611) inputs the extracted feature information into a pre-trained machine learning model to identify the user's expected number of sets and can identify the fatigue level based on the expected number of sets.

[0117] Here, the estimated set number is a value obtained as the output of a machine learning model when the processor (611) inputs a force profile (or feature information extracted therefrom) generated in correspondence with each set into a pre-trained machine learning model, and may represent an estimated value indicating which set (set number, Set index) the force-profile corresponds to during the user's exercise process. For example, the estimated set number may be output as an integer (e.g., 1, 2, 3, 1) or a real number, and if it is output as a real number, the processor (611) may convert it into a set number through rounding, truncation, or interval mapping.

[0118] Based on the fact that accumulated fatigue is reflected in changes to the feature information of the force-profile as the set number increases (i.e., as the isometric exercise time continues), the processor (611) can use the difference between the predicted number of sets and the actual number of sets, which is the output of the machine learning model, as an indicator of the degree of accumulated fatigue of the user. To this end, whenever a set is completed, the processor (611) can extract feature information from the force profile corresponding to the set and input it into the machine learning model.

[0119] The processor (611) can determine the user's fatigue level by comparing the number of sets actually performed by the user with the number of expected sets identified through a machine learning model. The processor (611) can identify the user's fatigue level by comparing the actual number of sets identified by the electric training device counting (currently ongoing set number) with the number of expected sets obtained as the output of the machine learning model (set number estimate).

[0120] Specifically, the processor (611) can use whether the predicted number of sets, which is the output of the machine learning model, is greater or smaller than the actual number of sets as a basis for determining the user's fatigue state, based on the fact that the force profile under the same exercise conditions shows a pattern according to the progress of sets (e.g., decrease in average force, change in maximum force, decrease in force expression rate, etc.) as the force profile changes to a form closer to the later sets.

[0121] For example, if the actual number of sets is less than the expected number of sets, the processor (611) may determine that the user is currently in a state of accumulated fatigue. On the other hand, if the actual number of sets is greater than the expected number of sets, the processor (611) may determine that the user's fatigue level is relatively low. Additionally, if the actual number of sets matches the expected number of sets, the processor (611) may determine that the user is exercising appropriately. In this way, the processor (611) can quantitatively identify the user's fatigue level based on the relationship between the actual number of sets and the expected number of sets.

[0122] FIG. 12 is an example diagram of adjusting the size of the load of an electric training device based on fatigue identified through a machine learning model according to one embodiment of the present disclosure.

[0123] And, the processor (611) can adjust the size of the load based on the identified fatigue (S1030).

[0124] In particular, the processor (611) can determine whether to adjust the load size by comparing the estimated number of sets with the actual number of sets performed by the user. Specifically, the processor (611) can determine whether the currently set exercise load is suitable for the user's performance ability and accumulated fatigue state based on the user's fatigue level identified through a machine learning model. The processor (611) can determine whether the fatigue level falls within a preset standard range, or whether the fatigue level is trending upward or downward, and determine whether it is necessary to adjust the load size based on the result of the determination. At this time, the processor (611) can determine whether to adjust the load by determining the user's fatigue level as a direct numerical value, or by using the estimated number of sets, which is the output value of the machine learning model, as an intermediate indicator corresponding to the fatigue level. Accordingly, the processor (611) can perform a load adjustment determination by quantitatively reflecting the user's current exercise performance state.

[0125] For example, referring to FIG. 12, the processor (611) may decide to increase the load size if the estimated number of sets is less than the actual number of sets performed by the user, decide to maintain the load size if the estimated number of sets matches the actual number of sets performed by the user, and decide to decrease the load size if the estimated number of sets is greater than the actual number of sets performed by the user. Specifically, if the estimated number of sets is less than the actual number of sets performed by the user, the processor (611) may decide to increase the load size by determining that the user is exhibiting a relatively high performance ability relative to the set exercise conditions. Conversely, if the estimated number of sets matches the actual number of sets performed by the user, the processor may decide to maintain the load size by determining that the current load is suitable for the user's performance ability and fatigue state. Additionally, if the estimated number of sets is greater than the actual number of sets performed by the user, the processor may decide to decrease the load size by determining that the user's fatigue level is relatively high. In this way, the processor (611) can make a load adjustment decision that reflects the user's fatigue level based on the relationship between the estimated number of sets and the actual number of sets.

[0126] Additionally, the processor (611) can adjust the exercise plan for the user's next exercise schedule using the adjusted exercise load and fatigue information identified through the machine learning model. For example, the processor (611) can adjust the exercise load for the next day based on the results of the day's exercise performance and load adjustment. If the processor (611) determines that the user's fatigue has increased, it can modify the exercise plan to change the exercise items to be performed the next day or to exclude specific exercise items. For example, if high fatigue is determined in an exercise using a specific muscle group, the exercise plan for the next day can be adjusted to exclude the exercise items using that muscle group or replace them with other exercise items.

[0127] Additionally, the processor (611) may store and manage exercise plan data structures to manage exercise plans for each user. For example, the exercise plan data may include at least one of an exercise type, target load, target number of repetitions, target number of sets, scheduled date of execution, and rest schedule. The processor (611) may dynamically modify the exercise plan data based on the user's exercise performance results and fatigue analysis results. For example, the processor (611) may change the target load, target number of sets, or exercise type of the next exercise schedule by reflecting the fatigue and load adjustment results calculated after the day's exercise performance in the exercise plan data.

[0128] Additionally, the processor (611) may adjust the exercise schedule based on fatigue information. For example, if it is determined that the user's fatigue level is higher than a preset threshold value, the processor (611) may modify the exercise plan to change the previously scheduled exercise schedule for the next day into a rest schedule. At this time, the processor (611) may automatically set a rest period of a certain duration (e.g., 1 to 2 days) considering the user's recovery status, and then reconstruct the exercise plan so that exercise resumes after the rest period. This prevents the accumulation of fatigue in the user and provides a customized exercise plan suitable for the individual's recovery status.

[0129] Additionally, the processor (611) can determine the optimal load, number of sets, or exercise type for the exercise schedule to be performed by the user based on at least one of the fatigue, exercise performance pattern, past exercise history, and load adjustment results estimated through a machine learning model during the exercise plan modification process. Accordingly, the processor (611) can learn the user's long-term exercise performance pattern and continuously update the personalized exercise plan.

[0130] FIG. 13 is an example diagram of selecting a machine learning model according to a preset number of sets and the number of repetitions of the sets according to one embodiment of the present disclosure.

[0131] According to one embodiment of the present disclosure, a machine learning model can be trained based on exercise data performed under conditions where the type of isokinetic exercise mode, the number of sets, and the number of repetitions per set are the same. That is, the processor (611) can train a machine learning model specialized for the corresponding exercise condition by using feature information of the force-profile per set obtained from multiple users who have performed the same type of isokinetic exercise and the same number of sets (and the number of repetitions per set). In this way, since the machine learning model learns the distribution characteristics of the force-profile and the pattern of change of feature information according to the progress of sets depending on the exercise condition, the accuracy of fatigue identification may be reduced when a single common model is applied to different exercise conditions. Accordingly, according to one embodiment of the present disclosure, the processor (611) can select a machine learning model corresponding to the exercise condition set for the current user from among a plurality of machine learning models trained according to the type of isokinetic exercise mode, the number of sets, and the number of repetitions per set. Specifically, the processor (611) identifies information regarding the type of isokinetic exercise mode, number of sets, and number of repetitions set through the user interface or display (640), and can select a machine learning model having the same learning conditions as the identified exercise conditions from among a plurality of machine learning models stored in memory (612). For example, if machine learning models learned for different number of sets or number of repetitions are stored, the processor (611) can select a machine learning model corresponding to the currently set number of sets and number of repetitions and use it for fatigue identification. Accordingly, the processor (611) can more accurately reflect the change in the distribution of feature information due to differences in exercise conditions and the pattern of fatigue accumulation according to the progress of sets, and as a result, can improve the reliability and accuracy of fatigue identification and load adjustment accordingly.

[0132] Meanwhile, a non-transitory computer-readable medium may be provided that stores a program for sequentially performing a control method of an electric training device that identifies the fatigue level of a user during isometric exercise based on a machine learning model according to one embodiment of the present disclosure.

[0133] A non-transient readable medium refers to a medium that stores data semi-permanently and can be read by a device, rather than a medium that stores data for a short moment, such as a register, cache, or memory. Specifically, the various applications or programs described above may be stored and provided on non-transient readable media such as CDs, DVDs, hard disks, Blu-ray discs, USBs, memory cards, and ROMs.

[0134] The various embodiments of the present disclosure described above may be combined with additional embodiments and modified to the extent understandable to those skilled in the art in light of the detailed description above. The embodiments of the present disclosure are illustrative in all respects and should be understood as not restrictive. For example, each component described as a single unit may be implemented in a distributed manner, and components described as distributed may likewise be implemented in a combined form. Accordingly, all modifications or variations derived from the meaning, scope, and equivalents of the claims of the present disclosure should be interpreted as being included within the scope of the present disclosure.

Claims

1. A control method for an electric training device that identifies the fatigue level of a user during isokinetic exercise based on a machine learning model, wherein A step of generating a force profile of the user by detecting the user's force through a handle part grasped by the user performing the constant-speed exercise while controlling a drive unit to generate a load according to the constant-speed exercise mode; A step of extracting feature information from the force profile and inputting the extracted feature information into a pre-trained machine learning model to identify the fatigue level of the user performing the isometric exercise; and A step of adjusting the magnitude of the load based on the identified fatigue level; comprising method.

2. In Paragraph 1, The method includes the step of setting the number of sets and the number of repetitions for each set according to the above isokinetic exercise mode; The step of identifying the fatigue level of the above-mentioned user is, A step comprising: extracting feature information corresponding to each set from the user’s force profile, inputting the extracted feature information into a pre-trained machine learning model to identify the fatigue level of the user corresponding to each set; method.

3. In Paragraph 2, The step of identifying the fatigue corresponding to each of the above sets is, The method comprises the step of inputting the extracted feature information into the previously trained machine learning model to identify the user's expected number of sets and identifying the fatigue level based on the expected number of sets. method.

4. In Paragraph 3, The step of adjusting the magnitude of the load based on the identified fatigue level is, A step of determining whether to adjust the size of the load by comparing the above-mentioned expected number of sets with the actual number of sets performed by the user; method.

5. In Paragraph 1, The above feature information is, Comprising at least one of work, mean force, peak force, minimum force, median force, rate of force development, time to peak force, variance of the force distribution, standard deviation, skewness, kurtosis, and root mean square (RMS). method.

6. In Paragraph 2, The step of generating the power profile of the above user is, The method comprises the step of generating a set-specific force profile of a user performing the above-mentioned constant velocity motion, excluding sections where a force smaller than a preset size is detected. method.

7. In Paragraph 2, The step of generating the power profile of the above user is, A step comprising: detecting the force of the user through a handle portion grasped by the user performing the above isometric exercise, normalizing the user's force based on the user's body weight, and generating the user's force profile; method.

8. In Paragraph 2, A step of selecting a machine learning model corresponding to the set number of sets and the number of repetitions of each set among a plurality of machine learning models trained according to the number of sets of the above isokinetic exercise mode and the number of repetitions of each set; comprising method.

9. In Paragraph 4, The step of adjusting the size of the above load is, If the estimated number of sets is smaller than the actual number of sets performed by the user, it is decided to increase the size of the load; if the estimated number of sets matches the actual number of sets performed by the user, it is decided to maintain the size of the load; and if the estimated number of sets is larger than the actual number of sets performed by the user, it is decided to decrease the size of the load. method.

10. An electric training device that identifies the fatigue level of a user during isokinetic exercise based on a machine learning model, A strength training rack comprising one or more pairs of floor frames and vertical frames; and horizontal frames; Electric exercise assistance units installed on each of the above pair of vertical frames and generating an exercise load for the user; A main fitness unit configured to be connected to each of the above-mentioned electric exercise assistance units to enable the execution of a predetermined fitness exercise; Sub-health units configured to be detachably connected to each of the above-mentioned main health units so as to enable the execution of predetermined health exercises; and A computing device comprising one or more processors for controlling the above-mentioned electric motion assistance unit, and The above computing device is, While controlling the drive unit of the electric exercise assistance unit to generate a load according to the isometric exercise mode, the force of the user is detected through the handle of the main health unit held by the user performing the isometric exercise to generate the user's force profile, feature information is extracted from the force profile, and the extracted feature information is input into a pre-trained machine learning model to identify the fatigue level of the user performing the isometric exercise, and the magnitude of the load is adjusted based on the identified fatigue level. Electric training device.

11. A computer program stored on a computer-readable storage medium, wherein the computer program, when executed on one or more processors, performs control operations of an electric training device that identifies the fatigue level of a user performing isokinetic exercise based on a machine learning model, and The above operation is, An operation of generating a force profile of the user by detecting the user's force through a handle part grasped by the user performing the constant motion while controlling a drive unit to generate a load according to the constant motion mode; An operation to extract feature information from the force profile and input the extracted feature information into a pre-trained machine learning model to identify the fatigue level of the user performing the isometric exercise; and Operation of adjusting the magnitude of the load based on the identified fatigue level; including Computer program.

Images