KR20220106448A - Personal exercise management system using artificial intelligence and electromyographic signal - Google Patents

Abstract

The present invention relates to a personal exercise managing system for combining artificial intelligence and wireless electromyogram signal processing, and more specifically, to a personal exercise managing system for combining artificial intelligence and wireless electromyogram signal processing capable of evaluating a movable range of fitness and taekwondo movements by an angle as a value; and inducing motor muscle (nerve) training based on an evaluated result. In addition, the present invention can improve muscle power of only sole muscle, muscle endurance, or flexibility to improve functional motor ability in a lack direction of balance sense of a human body depending on a test result. When performing a functional movement, by improving co-ordination or co-contraction of the nerve and the muscle for a muscle group related to a motion, the present invention prevents injury which occurs during functional activity of the human body. The present invention induces accurate movements. The present invention comprises: a control server; a signal processing module; and an artificial intelligence module.

Description

Artificial intelligence motion recognition evaluation and electromyography measurement system for customized exercise prescription {Personal exercise management system using artificial intelligence and electromyographic signal}

The present invention relates to an artificial intelligence motion recognition evaluation and electromyography measurement system for customized exercise prescription. Neuron) training (Sensory-Motor Training), and in order to improve functional exercise ability in the direction lacking the sense of balance of the human body according to the test result, the strength of a single muscle and muscular endurance In addition to improving (muscle endurance) or flexibility (flexibility), in performing any functional action, the coordination ability of the nerve and the muscle with respect to the muscle group related to the action (Co- Artificial intelligence motion recognition for customized exercise prescription, characterized in that it improves ordination or co-contraction to prevent injury that may occur during functional activity of the human body and induces accurate movement It relates to an evaluation and electromyography measurement system.

The body is inherently unstable until the balance sensory control system is activated because about 2/3 of the height of the mass is concentrated from the ground. In the response of the muscle to a sudden load from the outside, the muscle responds to stabilize the body at a high speed for postural stability and balance maintenance. In the homeostasis of maintaining joints in the movement of the human body, sensory information transmits the motion and position sensed by external receptors on the skin, joints and muscles to the central nervous system (CNS), and through the vestibular organs and vision, the central nervous system (CNS) is affected. Sensory information is organized and reformed in the subcortical center and actually begins to affect the periphery where movement is required.

Therefore, sensory-motor training is a process of integrating all components including sensory nerves and motor muscles (nerves) into the central nervous system for functional joint stability.

Meanwhile, smart fitness is a field that combines the core ICT technologies of the 4th industrial revolution, the Internet of Things (IoT), cloud computing, big data, and artificial intelligence (AI) with healthcare. Looking at the basic industrial structure, data specialists collect and analyze data generated by consumers in their daily lives or at specialized institutions such as medical institutions, and medical and health management companies use it again to provide advice and treatment to consumers.

Globally, the demand for 'cost-effective health care' is increasing in the context of a large increase in medical expenses due to the deepening of the aging population. The convergence of ICT technology and healthcare, that is, 'smart fitness', which is leading the 4th industrial revolution, is emerging as an alternative. It is expected. As a result, the trend of digitization of medical records worldwide is accelerating, and its usefulness is expected to be further improved.

Governments in each country are actively promoting policy support and regulatory reform for the smart healthcare industry in order to secure competitiveness and preoccupy the market. In order to nurture the smart healthcare industry as a new growth engine, Korea will also need to establish a regulatory infrastructure suitable for the new industry by referring to cases of other countries. In addition, in order to develop technologies and services of high public value, although it is difficult for individual companies to access them alone, government support measures are required for public big data and AI solutions using them.

An object of the present invention is to provide a personal exercise management system that combines artificial intelligence and wireless electromyography signal processing so that when muscle activity is monitored in real time during exercise, you can exercise more efficiently with objective and quantitative data and analysis results. .

In addition, there was no device for measuring individual exercise volume and efficiency and giving objective feedback because there were no exercise experts or non-professionals or accurate data on muscle use according to exercise type, time, and intensity. An object of the present invention is to provide a personal exercise management system that can prevent side effects such as injury and reduced exercise ability due to the failure of pace control.

In addition, using an EMG band with an EMG sensor attached to it, we measure the amount of muscle used in the chest, arm, and leg to exercise, measure the efficiency of the exercise with a signal processing module, collect and monitor data with a smart device, and provide a total solution to our customers. It aims to provide a personal exercise management system that provides

As a means for solving the above problems,

The present invention provides a control server for receiving exercise information by interworking with a monitoring module in which an exercise management application is installed through a wired/wireless communication network, and providing analysis information about the user's exercise; a signal processing module that receives detection signals from a plurality of EMG sensors attached to the user's body, analyzes the detection signals to calculate muscle activity, and provides them to the monitoring module; It provides a personal exercise management system that combines artificial intelligence and wireless electromyography signal processing, including; an artificial intelligence module for deep learning of big data including the detection signal to manage exercise.

The personal exercise management system of the present invention fuses artificial intelligence and wireless EMG signal processing to monitor muscle activity in real time during exercise to enable more efficient exercise with objective and quantitative data and analysis results.

1 is an overall configuration diagram of a personal exercise management system in which artificial intelligence and wireless EMG signal processing are fused according to an embodiment of the present invention.
2 is a communication structure diagram of a personal exercise management system that combines artificial intelligence and wireless EMG signal processing according to an embodiment of the present invention.
Figure 3 is an exemplary diagram of the evaluation joint position display for each position of the body tilt and sensor during the knee push-up.
Figure 4 is a view showing an example of the human body tilt and angle and posture maintenance operation in the preparation process of the push-up from the side.
5 is a view showing an example of the correct operation of the push-up preparation process from the front.
Figure 6 is an exemplary view displaying the exercise results and motion evaluation.
7 is a detailed exemplary view of exercise results.
8 is an exercise guide image and internal screen display.
Figure 9 is an example of the knee push-up preparation stage normal posture.
Figure 10 is an exemplary view of the knee push-up implementation stage normal posture.
11 is an exemplary view of the posture of the push-up preparation stage.
Figure 12 is an exemplary view of the push-up implementation stage normal posture.
13 is a conceptual diagram of a technical application and system of the present invention.
14 is a technical application and system flow diagram of the present invention.

Hereinafter, the principle of operation of the preferred embodiment of the present invention will be described in detail with reference to the accompanying drawings and description. However, the drawings shown below and the description to be given below relate to a preferred implementation method among various methods for effectively explaining the characteristics of the present invention, and the present invention is not limited only to the following drawings and description.

In addition, in the following description of the present invention, if it is determined that a detailed description of a related well-known function or configuration may unnecessarily obscure the gist of the present invention, the detailed description thereof will be omitted. And the terms to be described later are terms defined in consideration of functions in the present invention, which may vary according to intentions or customs of users and operators. Therefore, the definition should be made based on the content throughout the present invention.

In addition, preferred embodiments of the present invention to be implemented below are already provided in each system functional configuration in order to efficiently describe the technical components constituting the present invention, or system functions normally provided in the technical field to which the present invention pertains. The configuration is omitted as much as possible, and the functional configuration to be additionally provided for the present invention will be mainly described.

If a person of ordinary skill in the art to which the present invention pertains, will be able to easily understand the functions of the components already used in the prior art among the functional configurations omitted without being shown below, and the configuration omitted as described above Relationships between elements and components added for purposes of the present invention will also be clearly understood.

In addition, the following examples will be used with appropriate modifications to terms so that those of ordinary skill in the art to which the present invention pertains can clearly understand in order to effectively describe the key technical features of the present invention, thereby It is never limited.

As a result, the technical spirit of the present invention is determined by the claims, and the following embodiments are one means for efficiently explaining the technical spirit of the present invention to those of ordinary skill in the art to which the present invention belongs. it's just

Moreover, it is to be understood that all detailed description reciting the principles, aspects, and embodiments of the invention, as well as specific embodiments, are intended to cover structural and functional equivalents of such matters. It is also to be understood that such equivalents include not only currently known equivalents, but also equivalents developed in the future, i.e., all devices invented to perform the same function, regardless of structure.

Thus, for example, the block diagrams herein are to be understood as representing conceptual views of illustrative circuitry embodying the principles of the present invention. Similarly, all flowcharts, state transition diagrams, pseudo code, etc. may be tangibly embodied on computer-readable media and be understood to represent various processes performed by a computer or processor, whether or not a computer or processor is explicitly shown. should be

The functions of the various elements shown in the figures including a processor or functional blocks represented by similar concepts may be provided by the use of dedicated hardware as well as hardware having the ability to execute software in association with appropriate software. When provided by a processor, the functionality may be provided by a single dedicated processor, a single shared processor, or a plurality of separate processors, some of which may be shared.

In addition, the clear use of terms presented as processor, control, or similar concepts should not be construed as exclusively referring to hardware having the ability to execute software, and without limitation, digital signal processor (DSP) hardware, ROM for storing software. It should be understood to implicitly include (ROM), RAM (RAM) and non-volatile memory. Other common hardware may also be included.

In the claims of the present specification, a component expressed as a means for performing the function described in the detailed description includes, for example, any form of software including a combination of circuit elements or firmware/microcode that performs the above function. It is intended to include all methods of performing the functions of the device, coupled with suitable circuitry for executing the software to perform the functions. Since the present invention defined by these claims is combined with the functions provided by the various enumerated means and in a manner required by the claims, any means capable of providing the functions are equivalent to those contemplated from the present specification. should be understood as

The above-described objects, features, and advantages will become more apparent through the following detailed description in relation to the accompanying drawings, and accordingly, those of ordinary skill in the art to which the present invention pertains can easily implement the technical idea of the present invention. There will be. In addition, in the description of the present invention, when it is determined that a detailed description of a known technology related to the present invention may unnecessarily obscure the gist of the present invention, the detailed description thereof will be omitted.

Before describing various embodiments of the present invention in detail, it will be understood that the application is not limited to the details of the construction and arrangement of components described in the following detailed description or shown in the drawings. The invention is capable of being embodied and practiced in other embodiments and of being carried out in various ways. Also, device or element orientation (eg, “front”, “back”, “up”, “down”, “top”, “bottom”) The expressions and predicates used herein with respect to terms such as ", "left", "right", "lateral", etc. are used merely to simplify the description of the invention, and the associated apparatus Or it will be appreciated that it does not simply indicate or imply that an element must have a particular orientation.

1 is an overall configuration diagram of a personal exercise management system in which artificial intelligence and wireless EMG signal processing are fused according to an embodiment of the present invention.

2 is a communication structure diagram of a personal exercise management system that combines artificial intelligence and wireless EMG signal processing according to an embodiment of the present invention.

Figure 3 is an exemplary diagram of the evaluation joint position display for each position of the body tilt and sensor during the knee push-up.

Figure 4 is a view showing an example of the body tilt and angle and posture maintenance operation in the push-up preparation process from the side.

5 is a view showing an example of the correct operation of the push-up preparation process from the front.

6 is an exemplary view showing the exercise results and motion evaluation.

7 is a detailed exemplary view of exercise results.

8 is an exercise guide image and internal screen display.

Figure 9 is an example of the knee push-up preparation stage normal posture.

Figure 10 is an exemplary view of the knee push-up implementation stage normal posture.

11 is an exemplary view of the posture of the push-up preparation stage.

Figure 12 is an exemplary view of the push-up implementation stage normal posture.

13 is a conceptual diagram of a technical application and system of the present invention.

14 is a technical application and system flow diagram of the present invention,

The basic concept of the present invention is as follows.

1. Motion evaluation using two Kinect 4 sensors

(1) Maintaining an angle of 45 to 65 degrees on the front and side of the user's location, each fixed and mounted

(2) Score judgment by measuring the movement and posture maintenance of joints that can be evaluated by each position sensor

(3) Motion joint position (angle) range of motion measured (evaluated) by two sensors for every motion

<Evaluation criteria/method>

2. Evaluate the range of motion (joint) angle value for each fitness and taekwondo movement

(1) The point deduction step is applied according to the error of the movable range (angle value)

(2) For the evaluation criteria of Taekwondo trainees, the range of accuracy error is differentially applied according to the five-level grade.

Ex) Level 6~10: Very low accuracy for introductory course, Level 1~5: Low accuracy for beginner course

Grade 1~2: Intermediate course accuracy is normal, Grade 3~4: Advanced course accuracy is high

Athlete: Master course with very high accuracy, beyond the margin of error: No count

※ Evaluate by applying the course (grade) according to the member's exercise experience, experience, and exercise amount

3. The number of incorrect movements, the difference in movement speed, and the percentage of muscle use (applicable only when wearing an EMG sensor)

4. For all motions (except Poomsae), motion measurement (evaluation) methods and numerical standards for each joint position are confirmed

5. Prepare a method for professional fitness trainers to directly input evaluation criteria for each movement

<Evaluation analysis method>

6. After the motion evaluation, it is necessary to search the replay time for each section in detail at intervals of 0.05 seconds or less.

(1) Currently divided into 10 stages based on 60 seconds

(2) Present the wrong number of times by dividing the evaluation deduction items “moving range (angle), posture, and movement speed”

(3) Install the result analysis button on the screen to easily check the points deduction (operation error)

* Display the deduction button on the total exercise time graph to make it easier to check the deduction action

* Replay video by pressing the point deduction button: Marker color classification on the deduction part performer's avatar

(4) For Taekwondo movements, each hand/foot score is displayed + 3 points are displayed

(5) Only 3 points deducted for fitness movements (postures) are displayed

(6) When the action is finished, the evaluation score and result are presented within 2 seconds

* Accuracy evaluation sample action presentation is shortened to within 3 to 5 seconds

* After presenting the sample motion, “3, 2, 1 run” is displayed on the screen and motion evaluation proceeds

* Motion evaluation is performed at least 11 times (No evaluation is applied to the first motion)

(7) Comparison with standard exercise immediately after Taekwondo movement or fitness/Pilates/Yoga movement

* Evaluation by overlapping standard actions and user actions in the screen division area

* Overlap by dividing standard action avatars and users into size or real and transparent avatars

* The current screen is divided into two, but when analyzing the overlap, one screen is analyzed

* Select one of the screen areas to overlap the avatar + the user's real

(8) When evaluating motion, three screen sizes are equally divided (standard, member, EMG avatar)

(9) The EMG attachment avatar rotates in place as a 3D model and expresses the part/use of muscle used

(10) Points to be deducted from the use of electromyography muscles. Points displayed on the body (presented as a percentage of use of the prime mover and cooperative muscles)

<Artificial Intelligence Smart Health Coach Service>

7. When the last two specified number of moves Ex) “Fighting for the last two, do your best for the last one!”

Voice service (3 types of expression voice support)

8. Real-time (immediate) voice support in case of 2 incorrect movements Example) “Straighten your back, bend your knees,

Motion correction voice service such as “Raise your arm.” (Fix a specific joint according to the movement movement)

9. Detect user movement and guide exercise performance with command (count) voice (one, two, three...)

10. Differential application of exercise frequency command (voice count) speed

(1) Very fast in the beginning → Fast in the middle → Normal in the middle → Slow → Very slow in the second half → Very slow

(2) Service can be provided by calculating the command speed for each exercise as a time ratio (mathematical calculation formula applied)

11. When using an EMG sensor, express muscle usage by body part as a percentage

(1) 3D avatar part-specific color classification to express usage

(2) Expressed in the total muscle usage avatar

<Smartphone standard operation description and individual motion video replay Unity app>

12. App for user smartphone Unity behavior evaluation and analysis

(1) Fitness (basic) + Taekwondo, golf, basketball, yoga, school physical education class, K-pop dance development respectively

* Only the fitness (basic) + Taekwondo app is developed first (the rest of the sports are implemented with the same function)

(2) Detailed explanation of the operation method and posture, points to be noted, and necessity according to the viewing angle

(3) The posture and range of motion shown according to the up/down/front/back/left and right positions are explained by angle values.

(4) Let a professional fitness trainer directly input the implementation method

<Motion evaluation function>

1. Admin action evaluation screen

(1) Accuracy evaluation items for each operation

* Posture - fixed angle of each joint for performing the motion (starting point of the motion)

* Range of motion - The range of movement of the joint around the prime mover during movement (straight line, curved line)

* Movement - the range of space moved by moving your feet

* Timing (speed) - the speed when the joint is moving

* Expressiveness (skill level)

* Center of gravity shift - the degree of body sway when there is movement

(2) Real-time evaluation screen

* Separate each joint

: Shoulder/elbow/wrist/pelvis/knee/ankle left and right 12 joints and waist, neck

: Display the accuracy evaluation graph for each joint of individual movement separately from standard exercise (color)

* Step-by-step classification of joint motion range and posture angle standard setting (algorithm) error range for each motion

* Guidance of movement through the standard motion avatar's range of motion (implemented on the user's avatar screen)

* Display the number of times of exercise and real time

* Presenting the overall evaluation score

(3) 3D human body avatar placed on the screen

* Present the score for each joint

* EMG sensor measurement value muscle/joint range of motion angle value display (real-time graph implementation)

(4) Analysis graph

* Present scores using diagrams, bar and curve graphs

* Cumulative data and comparative analysis: suggest the degree of change/improvement

* Total learning cumulative data analysis: type, number, time, growth change graph provided

2. Evaluation Criteria

(1) Based on 100 points for each item

Composite score is presented as an average value (compare the highest cumulative data record, average record and current record)

3. Solution

(1)Dance content

* Provide feedback after automatic extraction of incorrect parts

* Present standard motions in front/back/top/bottom/left/right split screen

* Repeat the 5 sec/10 sec motion before and after the wrong part 3 times with detailed explanation and slow motion

* Artificial intelligence choreographer's motion modification know-how provided: physical strength/strength exercise prescription

(2) Taekwondo, basketball, golf sports content

(3)Yoga, fitness movement health contents

(4) Contents of school physical education class

4. Wear the accelerometer

(1) Calculation of range of motion such as distance, height, flight time, and trajectory value for each movement

Calculate distance by calculating motion start and end times

Count the motion by calculating the range of motion of the gyro

5. Interlocking heart rate measurement smart band

(1) Display heart rate change data on the measurement screen during exercise

(2) Smart band Bluetooth interlock

<Smartphone personal exercise record Unity (game engine) App >

Transfer the entire personal exercise data obtained from the motion accuracy evaluation of the above function to the motion evaluation user's personal smart phone and check the personal exercise video in the app

(1) Comparison of sports events (taekwondo, golf, basketball), health (yoga, fitness), and K-pop dance standard movements

Personal video records can be kept for 1 year or only 24 video record data

Individual motion accuracy Compare and analyze accumulated data to provide differences

Smartphone Unity App Features

(2) 2~16 times slower or faster speed can be adjusted

(3) Classification action: When an action and an action are connected, the next action is connected after a pause (0.5 seconds to 2 seconds)

(4) Video play is possible by selecting the avatar operation position from the front to the rear up/down/left/right

(5) If you click the avatar after pause, posture/movement angle is provided according to the up/down/front/back/left/right position, multilingual voice support in Korean, Chinese, English, etc. and subtitle explanation are provided

(6) Simultaneously play the standard motion motion capture video and individual motion data to compare and analyze motion, compare and evaluate motion accuracy such as speed/posture/joint angle by playing at the same time. notice

(7) When an EMG sensor is used for motion evaluation, the standard and muscle usage of the prime mover and synergist muscles of the corresponding body part are analyzed to provide muscle use accuracy results, and the same function as the motion evaluation function is applied.

※ All analysis of the above functions provides results through the server's analysis algorithm.

※ Live class using the above motion accuracy evaluation sensor and EMG sensor

(8) It is possible to conduct sports events and fitness movements as real-time practical classes through real-time video conferencing software such as Zoom.

(8) Real-time evaluation and feedback possible by applying O2O (online 2 offline) coaching system

The present invention is an exercise training device for inducing a passive exercise that can move the entire body in eight directions, forward, backward, left and right, diagonally, and a 3-axis inclination sensor attached to the center of the shoulder to measure the movement of the upper body. It is possible to evaluate the range of motion of the spine by measuring the range of movement of the body according to the movement of the device without disturbing the posture by measuring the movement of the body.

After attaching a 4-channel EMG module to the left and right rectus abdominis muscles and the left and right erector spinae muscles, if the preset range is exceeded in the initial state of the EMG, the direction is stopped and returned to the origin again. It is to measure the angle according to the state of the muscles involved in spinal stabilization by measuring the angle according to the

As a method to measure the movement of the device, the movement of the left and right and front and back angles is measured through the encoders attached to the left and right and front and back sides of the device. Thus, when the spine moves forward, backward, left, and right, the tilt of the upper body occurs, and the degree of inclination is converted into an angle and the movement of the upper body is measured as an angle.

In addition, the entire system including a personal exercise management system that combines artificial intelligence and wireless EMG signal processing according to an embodiment of the present invention includes a monitoring module, an exercise management server, an EMG sensor, and a plurality of exercise equipment.

The monitoring module is a terminal that allows a user to access the exercise management server and download and install an exercise management application from the exercise management server, and includes a smart phone, a laptop computer, or a tablet PC including a display window.

Such a monitoring module interworks with the exercise management server through a wired or wireless Internet, and in this case, the wireless Internet may be wifi, Bluetooth, or the like.

The monitoring module may install an exercise management application for the exercise management server, transmit various information to the exercise management server by driving the application, and receive various information from the exercise management server.

The EMG sensor includes a plurality of sensor modules, and each sensor module is implemented as a wearable device.

That is, each EMG sensor module is manufactured in a band-type structure so that it can be directly attached to the user's body.

The EMG sensor module 110 may sense an EMG of movement according to the user's exercise and transmit a detection signal.

The EMG sensor module is provided with a communication unit for wireless communication with the signal processing module to transmit a detection signal generated according to a user's movement to the signal processing module.

The plurality of sensor modules of the EMG sensor may be attached to various body parts of the user and transmit respective detection signals at the same time.

That is, it can be freely attached to the user's arms, legs, chest, and buttocks, and the user can sense the exercise effect on the target muscle by attaching it to the target muscle position during exercise.

Each EMG sensor module has a unique serial number, and this serial number is transmitted to the signal processing module together with the generated detection signal so that the signal processing module can identify each sensor module.

Each EMG sensor module may include a sensor unit, an A/D converter, a communication unit, and a battery.

The sensor unit is an EMG sensor and measures the surface EMG by sensing biosignals accompanying muscle activity detected through electrodes attached around the muscle. The EMG sensor attaches two electrodes, a reference electrode and a measurement electrode, to the human body to measure the amount of voltage, current, and frequency flowing around the muscle.

At this time, the potential difference formed between the two electrodes is amplified by the amplifier of the sensor, and the power supply noise of 60Hz can be removed by the filter. In addition, the EMG signal is detected by removing the noise of the high-frequency component by the low-pass filter.

The A/D converter digitizes and outputs the EMG signal of the sensor unit, and the communication unit transmits the digital signal to the signal processing module through a wired or wireless communication network, and the communication unit transmits each EMG sensor serial number together.

In addition, each of the EMG sensor modules includes a battery, and the battery may be a rechargeable battery.

Meanwhile, the exercise management server may include a signal processing module and a control server, and the signal processing module and the control server may be physically separated from each other, but may be functionally separated from one PC.

The signal processing module receives various sensing signals from the EMG sensor through a wired or wireless communication network, processes and reads them, and calculates muscle activity, which is an effective characteristic value.

In more detail, the signal processing module may include a synchronization and filtering unit, a signal analysis unit, and a feature extraction unit.

The synchronization and filtering unit synchronizes the plurality of detection signals received from the respective EMG sensor modules for each channel and performs noise filtering.

The signal analysis unit includes a first analysis unit and a second analysis unit for obtaining an effective feature value from the detection signal.

The first analyzer decomposes the filtered detection signal into several intrinsic mode functions (IMFs) using empirical mode decomposition (EMD), obtains a spectrum value for each IMF, and calculates the IMFs value above the threshold value from the harmonic characteristics and the power ratio. can be saved

The second analyzer decomposes the filtered detection signal into a plurality of subbands using discrete wavelet transform (DWT), calculates the average, variance, skewness, and kurtosis of each band, among the rate of change of values obtained from each subband for each frame. A maximum rate of change subband having the largest rate of change may be selected.

In this way, the IMFs value and the maximum rate of change subband are defined as valid feature values.

Meanwhile, the feature extractor calculates the muscle activity from the selected effective feature value. In detail, the RMS is obtained from the selected IMFs and the selected subbands to calculate the degree of muscle contraction muscle tone, and the degree of muscle fatigue is calculated from the median frequency. In addition, the feature extractor analyzes the muscle contraction timing using a cross-correlation function between channels.

In this way, the feature extraction unit extracts the muscle contraction degree of muscle tension, fatigue, and muscle contraction timing and transmits it as the activity of the muscle.

On the other hand, the control server checks whether the user is an exercise management service subscriber through a wired or wireless communication network, and if the user is an exercise management service subscriber, receives body information and exercise information of the exercise management service subscriber, analyzes it, and provides a customized exercise program. You can make suggestions, and you can provide an improvement plan for the current exercise method.

In addition, through archive analysis, the exercise information of various subscribers is accumulated and stored, and it is analyzed by time, age, gender, and region. Draw up the city's problems and improvement plans.

An exercise management server including such a signal processing module and a control server can be installed in the monitoring module to display improvement plans and feedback during such exercise, and can transmit start information to each EMG sensor module, etc. Exercise management application to provide.

In such an exercise management system, a user installs an exercise management application on a monitoring module, for example, a user's smartphone, and performs an operation in a state in which the plurality of EMG sensor modules are attached to a body part to be exercised.

Hereinafter, the operation of the system according to an embodiment of the present invention will be described.

First, a user selects an exercise motion while having a monitoring module installed with an exercise management application, for example, a smartphone, and selects an exercise device when there is an exercise device to be used. If a separate device is not required, the selection of exercise equipment can be omitted.

Next, the user drives the exercise management application of the smartphone to describe the current exercise time and the physiological state of the user exercising. The physiological state may be gender, height, weight, age, and abdominal obesity, and the physiological state information may be measured through various measuring instruments, for example, a scale, a tape measure, InBody, and the like.

In addition, such body information may be transmitted to the exercise management server through a wired/wireless communication network.

Next, the exercise management server requests the attachment location information of each sensor unit of the sensor module from the EMG sensor, and receives the corresponding location information. At this time, the location information is also transmitted to the monitoring module.

When the monitoring module receives the location information, it initializes the device and starts exercising.

In this case, the monitoring module may transmit the corresponding exercise information, that is, information such as time, device, physiological state, etc., to the exercise management server through the application.

When the exercise starts, the EMG sensor generates a detection signal and transmits it to the signal processing module of the exercise management server.

Next, the signal processing module calculates the muscle activity for each motion from the detection signal and transmits it to the monitoring module.

The process of calculating such muscle activity is as follows.

In detail, first, a sensing signal is received, and the sensing signal is decomposed into several intrinsic mode functions (IMF) using empirical mode decomposition (EMD).

Next, spectral values for each IMF are obtained among the corresponding IMFs, and if the overtone characteristics and power ratio are above the threshold value, IMFs are selected.

Meanwhile, the filtered detection signal is decomposed into a plurality of subbands using discrete wavelet transform (DWT). Next, the average, variance, skewness, and kurtosis of each band are obtained, and the maximum rate of change subband having the largest rate of change among the rates of change of values obtained from each subband for each frame is selected.

In this case, the IMFs value and the maximum rate of change subband are defined as effective feature values, and muscle activity is calculated from the effective feature values. In detail, the RMS is obtained from the selected IMFs and the selected subbands, the degree of muscle contraction muscle tone is calculated, and the muscle fatigue degree is calculated from the median frequency.

Next, the muscle contraction timing is analyzed using a cross-correlation function between channels, that is, between each sensor module.

In this way, the muscle contraction degree of muscle tension, fatigue, and muscle contraction timing are extracted and transmitted to the monitoring module as muscle activity.

The monitoring module receives the activity of the muscle and displays it (S170). In this case, the activity through the exercise management application is displayed in the form of a body map so that the user can easily and effectively recognize it.

On the other hand, the control server of the exercise management server analyzes the muscle activity and exercise information from the signal processing module to determine the exercise state of the user, derives a plan to improve the exercise state, and transmits it to the monitoring module.

The monitoring module receives it through the application, displays it as an exercise prescription, gives feedback to the user, and terminates the application.

The control server updates the database by analyzing archives including exercise prescriptions.

In this way, by attaching the wearable EMG sensor to the user's exercise part, and reading and showing the muscle activity in real time while the exercise is in progress, it is possible to provide an accuracy and improvement plan of the exercise, so that an efficient exercise is possible.

Next, the software installed in the personal exercise management system will be described.

1. Web Development (Service)

go. It uses the MVW (Model - View - Whatever) pattern.

This makes the application modularity efficient, increasing code reusability and enabling efficient maintenance.

me. AMD (Asynchronous Module Definition) module is used.

It presents an API design with dynamic loading, dependency management, and modularity intertwined beautifully like cogs.

(1) Dynamic Loading, Lazy Loading

When loading an HTML file, the <script> tag slows down page rendering. This is because the browser does not perform any other actions while HTTP request, download, parsing, and execution of the <script> tag are taking place. At the scale of a web app with a complex screen and dotted with AJAX, this time increases significantly. Web apps often have multiple views that are converted to AJAX. For further optimization, we need a gradual method of loading only the JavaScript necessary for the first rendering and interaction first, and then loading the rest according to the user's reaction. For this reason, PencUp uses dynamic loading, which allows you to load only the files you need without interfering with page rendering.

(2) Dependency Management

Dependency management is essential to facilitate maintenance. However, in JavaScript, it is difficult to understand the dependencies between scripts. This is because, at the language level, explicit and mandatory keywords such as #include and package/import, and packaging policies are not supported. To solve this problem, the scripts of specific functions are grouped into a unit that can be named, and dependency management is thoroughly managed.

(3) Modularization

Variables and functions that are only used inside the script do not need to be placed in the global space and should not be placed. Overuse of global variables and their conflicts have a huge impact on maintenance, and modularization of scripts prevents this problem.

In the basic module pattern, only variables and functions to be accessed from the outside can be selectively exposed, and variables and functions that do not need to be exposed to the outside can be accessed without locating them in the global space using closures.

All. A mechanism for forcing uniform styling is pursued.

(1) We present a model that forces front and server-side programming to have uniform styling. Therefore, there is no great difficulty in maintaining (modifying and upgrading) the previous program. It also pursues a mechanism that enforces uniform styling between developer A and developer B.

la. Improving the input/output speed between the user (PC, mobile device) and the server

(1) Structure and conceptual diagram of software program

- Basic structure

- Software program structure structure applied to pinyfiny program

- Conceptual diagram of program control

- Front server side communication program structure diagram

server side architecture

(2) As can be seen from the above structural diagram and conceptual diagram, the speed of the web screen is improved by dividing the page into left, top and bottom, rather than moving pages in the caching system and front, compared to the existing software design.

mind. Receiver and EMG sensor (channel) automatic update structure diagram

(1) Used in case of EMG sensor failure or replacement

(2) After registering and changing the EMG sensor in the operation server, the version information (Integer) information is changed.

(3) When starting exercise with the receiver, the receiver setting information is updated if the requested version information is large compared to the previously registered version information.

bar. Receiver program remote update and patch structure diagram

(1) Receiver program update and patch can be performed remotely.

(2) If the program update is executed on a specific receiver, it can be updated by sharing the patch file.

(3) Interactive update and patching is possible.

Next, an example of an analysis method of a signal from which noise has been removed by EMG signal processing will be described.

1. Basic analysis

1.1 Time-based analysis and spectrum analysis are performed on the signal that has been removed from SW-based noise in units of 1 second. Using this, RMS, power spectrum, median frequency, mean frequency, RMS-probability distribution, RMS cross-correlation, and MVIC normalization values are calculated.

1.2 First, using MVIC Noramlization, find the maximum value of muscle activity for each person based on the maximum value of each person's muscle. Based on this, the RMS, Median Frequency, and Mean Frequency values are monitored. Using RMS, the degree of muscle contraction ('A' value) is shown in a graph, and muscle fatigue is measured by calculating the median frequency and mean frequency values ('B' value) according to the time timing. Therefore, if the values of A and B decrease significantly and last for more than 5 seconds, it is transmitted to the server and an alarm is given to the trainer.

2. Signal processing 1: The process of obtaining significant values (features) from the EMG data measured by time

2.1 Acquisition Signal Decomposition1

: Using ICEEMDAN (Improved Complete EEMD (Ensemble Empirical Mode Decomposition) with Adaptive Noise ) to decompose raw EMG data into multiple intrinsic mode functions (IMF) -> Obtain the spectrum value for each IMF -> From the spectrum of each IMF Obtain overtone characteristics and power ratio -> Select IMFs above the threshold.

2.2 Acquisition Signal Decomposition 2

: Using discrete wavelet transform to decompose raw EMG data into multiple subbands (fitting for delta wave, theta wave, alpha wave, beta wave sections) -> Obtain the average, variance, skewness, and kurtosis of each band -> Set the rate of change of the value obtained from each subband for each frame and select the subband with the most change.

3. Signal processing 2: Calculate the feature vector from the decomposed EMG signal

2.1 Obtain the AR coefficient from the AR (auto-regressive) model in the IMF selected from 1.1 and the wavelet subband obtained from 1.2. In particular, among the AR models, the EIV (Error in Variable) model, which is robust against noise, is applied.

4. Motion sensor signal processing

: Based on the threshold set by the expert group for each exercise, if the user deviates from the corresponding angle during each exercise, it is integrated with the value analyzed through 2 and 3 per set for monitoring and transmitted to the server.

In order to increase the accuracy of the AI department, sensor information of several athletes is collected and stored in the database, the analysis information according to each situation is stored in the database, and data statistics are analyzed according to the time, age, gender, and region of the worker. can be stored in the database.

An example of a method for obtaining an AR coefficient using EIV in the artificial intelligence unit may include the following method.

차 AR 모델로 표현되는 신호를 나타낸 것이다.In a system identification problem, an EIV model that represents both input and output errors and noise is called an EIV model. Figure 2 shows the EIV technique for estimating AR coefficients.

The signal represented by the car AR model is shown.

(Equation 1)

는 AR 모델의 계수를 나타내고,

는 평균이

이고 분산이

인 가산성 백색 잡음(additive white Gaussian nosie; AWGN)이다. 신호

는 평균이

이고 분산이

인 AWGN

가 더해져서 관측이 된다.here

represents the coefficient of the AR model,

is the average

and the dispersion

Phosphorus is additive white Gaussian nosie (AWGN). signal

is the average

and the dispersion

in AWGN

is added and observed.

(Equation 2)

In the present invention, the variance of the AR coefficient and input and measurement noise is measured using EIV having the following characteristics.

i) Using the property of the solution set of the Frisch method

ii) using the shift property of a time-invariant dynamic system

iii) Simultaneously estimate the AR coefficients that satisfy the positive-definite matrix of the covariance matrix, and the variance of input and measurement noise

Using the Frisch technique and the shift property, it is possible to find the point at which the noise dispersion in the noise plane is minimized. That is, it is possible to minimize the error of AR coefficients predicted in a noisy environment, so that a feature vector more robust to noise can be extracted.

개의 유한 데이터에서 EIV에 기반한 AR 계수 예측 알고리즘에 대해 간략히 기술한 것이다.next

This is a brief description of the AR coefficient prediction algorithm based on EIV on finite data.

에 대한 공분산 행렬

은 다음과 같이 정의된다. Step 1:

covariance matrix for

is defined as

(Equation 3)

는

개의 관측된 신호

로 구성되는

Hankel 행렬이고 다음과 같다.here

Is

observed signal

consisting of

It is a Hankel matrix and is

(Equation 4)

은 유한하므로, AR 계수를 추정함에 있어 공분산 행렬에 대한 추정치를 계산하여 이용해야 한다. AR 차수

,

에 대해 다음과 같이 공분산의 추정을 계산한다.Number of actual observed signals

Since is finite, an estimate for the covariance matrix must be calculated and used in estimating the AR coefficient. AR order

,

Calculate the estimate of the covariance as follows.

(Equation 5)

(Equation 6)

에서 식 (7)을 만족하는 점

에 대해 연속된 convex 곡선의 집합을

라 하자.Step 2: Noise plane

point that satisfies Equation (7) in

A set of continuous convex curves for

let's say

(Equation 7)

를 지나는 직선이

와 만나는 점

는 다음과 같이 구할 수 있다.Using Equation (8), the origin and point

a straight line passing through

meeting point with

can be obtained as

(Equation 8)

이다. 이와 같은 방식으로 원점과 점

를 지나는 직선이

와 만나는 점

또한 구할 수 있다. here

to be. In this way the origin and point

a straight line passing through

meeting point with

Also available.

,

,

를 계산한다.Step 3: Using the relation

,

,

to calculate

(Equation 9)

(Equation 10)

(Equation 11)

는 점

를 이용하여 구한 AR 계수

를 의미한다.here

the point

AR coefficient obtained using

means

일 때, 식 (12)에 주어진 SR 비용함수(shifted relation cost function)를 계산한다.Step 4:

, calculate the shifted relation cost function given in Equation (12).

(Equation 12)

를 최소화 하는 분산

를 탐색한다.Step 5: SR cost function

variance to minimize

explore

와 일치하는 AR 계수 벡터

를 구한다.Step 6:

AR coefficient vector matching with

save

Claims (1)

a control server for receiving exercise information by interworking with a monitoring module in which an exercise management application is installed through a wired/wireless communication network, and providing analysis information about the user's exercise; a signal processing module that receives detection signals from a plurality of EMG sensors attached to the user's body, analyzes the detection signals to calculate muscle activity, and provides them to the monitoring module; AI motion recognition evaluation and electromyography measurement system for customized exercise prescription, characterized in that it comprises; an artificial intelligence module that manages the exercise by deep learning the big data including the detection signal.

Images