A real-time feedback training system and method based on swimming phase monitoring

Abstract

The present invention discloses a real-time feedback training system and method based on swimming phase monitoring, including: an intelligent wearable device, which is used for a swimmer to wear during training, collects the swimming data of the wearer, and uploads it to the phase real-time monitoring module and the data storage module; the phase real-time monitoring module, which is used to monitor the swimming phase of different swimming strokes at different swimming frequencies, and obtain the stroke category and continuous phase change; the interaction module, which is used for the user to select the current required feedback mode and input personal information, so as to realize the personalized customization of the user and expand the functions of the system; the communication module, which is used for the communication inside or between the data acquisition system, the phase real-time monitoring module, the interaction module and the real-time feedback device. The flexible capacitive sensor in the present invention is waterproofed, so that it can reliably collect the changes in the muscle deformation signals during swimming, not only can it monitor the swimming phase in real time, but also can establish the muscle coordination relationship.

Description

Technical Field

[0001] The present invention relates to the technical field of swimming training, and specifically provides a real-time feedback training system and method based on swimming phase monitoring. Background Art

[0002] Currently, the improvement of swimming techniques and training effects has always been the focus of attention in the field of swimming. However, traditional swimming training methods usually cannot provide immediate and personalized feedback, which limits the improvement of swimming techniques by athletes during training.

[0003] Existing swimming detection methods, such as video analysis and inertial measurement units, although can be used to evaluate and analyze swimming techniques to a certain extent, also have some disadvantages and limitations. Video analysis requires recording the video of the swimmer in advance and performing time-consuming offline analysis and processing later, which means that real-time feedback and immediate technical adjustment cannot be provided. In addition, factors such as water refraction, water splash, and water surface fluctuation may affect the image quality, thus affecting the accuracy of the results. While the inertial measurement unit can provide real-time data, its accuracy is limited by the inertial sensor itself, and data drift is likely to occur during long-term use, resulting in cumulative errors. In addition, the phase division methods are all discrete and only applicable to a single swimming frequency. Therefore, the obtained swimming motion information is very limited; real-time feedback on the current swimming phase cannot be provided, and training feedback can only be given through video analysis after the training ends, resulting in low training efficiency.

[0004] In summary, although video analysis and inertial measurement units have certain application values in swimming technique evaluation, they have some limitations and deficiencies in terms of real-time performance, accuracy, etc. Summary of the Invention

[0005] Aiming at the above-mentioned technical deficiencies, the purpose of the present invention is to provide a real-time feedback training system and method based on swimming phase monitoring, which can monitor the swimming phase of the swimmer in real time and accurately, and provide personalized feedback to help the trainer improve swimming techniques.

[0006] To solve the above technical problems, the present invention adopts the following technical solutions:

[0007] The present invention provides a real-time feedback training system based on swimming phase monitoring, including:

[0008] An intelligent wearable device, which is used for the swimmer to wear during training, collects the swimming data of the wearer, and uploads it to the phase real-time monitoring module and the data storage module;

[0009] The swimming data includes muscle deformation, current time, muscle balance, muscle strength, and muscle fatigue.

[0010] A phase real-time monitoring module, which is used to monitor the swimming phases of different swimming strokes at different swimming frequencies, so as to obtain the stroke categories and continuous phase changes;

[0011] An interaction module, which is used for the user to select the current required feedback mode and input personal information, so as to achieve personalized customization of the user and expand the functions of the system;

[0012] A communication module, which is used for communication inside or between the data acquisition system, the phase real-time monitoring module, the interaction module and the real-time feedback device;

[0013] A real-time feedback device, which is used to give the user corresponding real-time feedback information according to the preset feedback mode and the output of the phase real-time monitoring module, and improve the swimming efficiency by optimizing the swimmer's stroke;

[0014] A data storage module, which is used to store the personal information and swimming data of the wearer; the storage method of the data storage module is one of local storage, cloud storage and memory storage; the personal information includes height, weight, age and gender; the swimming data includes muscle deformation, current time, muscle balance, muscle strength and muscle fatigue;

[0015] A technical analysis system, which is used to analyze the swimming data in the data storage module, understand the technical characteristics of the swimmer, quantify the training effect, and put forward improvement suggestions and swimming training plans;

[0016] The phase real-time monitoring module includes a swimming stroke classifier in the first stage and a swimming phase regressor in the second stage. The swimming stroke classifier can be one of a neural network, a support vector machine, a decision tree, and a linear discriminant analysis. The swimming stroke classifier classifies the data of four swimming strokes, namely freestyle, breaststroke, backstroke, and butterfly stroke, based on the muscle deformation signal to obtain the swimming stroke category of the current swimmer; the swimming phase regressor includes a freestyle phase regressor, a breaststroke phase regressor, a backstroke phase regressor, and a butterfly stroke phase regressor. The swimming phase regressor can be a linear regressor, a support vector regressor, a decision tree regressor, a random forest regressor, a gradient boosting regressor, and a neural network regressor. The phase real-time monitoring module inputs the current muscle deformation signal into the phase regressor corresponding to the swimming stroke according to the result of the first-stage swimming stroke classification, and monitors the swimming phases of different swimming strokes at different swimming frequencies to obtain continuous phase changes.

[0017] Preferably, the intelligent wearable device includes:

[0018] Flexible capacitive sensing system, including a plurality of flexible capacitive sensors, which are used to measure muscle deformation signals during swimming. It includes a waterproof elastic layer, a fixing frame, a housing, an upper electrode, a dielectric layer, a lower electrode and an acquisition circuit. The housing includes an upper end cover, a lower end cover and a sealing cover. The housing is made of waterproof material. Assembly holes and wire routing holes for assembly and wiring are installed on the housing. Externally, it is used to fix the flexible capacitive sensing unit and provide a hard boundary for its deformation. The flexible capacitive sensing unit includes an upper electrode, a dielectric layer and a lower electrode; the dielectric layer is arranged between the upper electrode and the lower electrode. The flexible capacitive sensor is fixed at the belly of the muscle to be measured. When the muscle deforms, the dielectric layer is deformed by the pressure, thereby changing the capacitance value. The waterproof elastic layer completely wraps the flexible capacitive sensor and fixes its periphery to the groove of the upper end cover through the fixing frame. Then, Ecoflex is poured into the groove for sealing, avoiding the contact between the flexible capacitive sensing unit and water, thereby reducing crosstalk noise. The acquisition circuit is arranged inside the housing assembled by the upper end cover and the lower end cover. The wire is connected to the electrode through the wire groove of the housing, which is used to acquire multi-channel capacitance signals and send all the data to the host computer after synchronization. Waterproof silicone is poured into the housing through the opening of the lower end cover to completely immerse the acquisition circuit, and then sealed with the sealing cover to prevent water from contacting the acquisition circuit and causing a short circuit phenomenon.

[0019] Swimming tight suit, including swimming tight pants, swimming tight clothes and swimming flippers. Nylon fasteners are sewn on the swimming tight suit, and the nylon fasteners are used for fixing the flexible capacitive sensors.

[0020] Preferably, the interaction mode can be one of voice, touch, gesture and facial forms.

[0021] Preferably, the communication mode of the communication module can be one of acoustic wave communication, electromagnetic wave communication, underwater acoustic optical fiber communication and wired cable communication.

[0022] Preferably, the phase real-time monitoring module has undergone offline training before actual use and determined the optimal hyperparameters through intelligent optimization algorithms. Among them, the intelligent optimization algorithms can be genetic algorithms, ant colony algorithms, particle swarm algorithms, etc. The specific steps are as follows: First, a large amount of swimming data is obtained through swimming experiments and preprocessed and feature extracted. Then, the real phase labels are obtained according to the synchronized experimental videos. Within a single swimming cycle, the real phase labels 0-100% are generated by linear interpolation. To avoid discontinuity when switching between adjacent cycles, that is, 99% directly changes to 0%, the error between the predicted phase and the real phase of the phase real-time monitoring module is represented by the included angle between two vectors. The specific steps are as follows. First, through formula (1), the swimming phase percentage is converted into an angle between 0 and 2π θ, and through Formulas (2) and (3), it is further mapped to two coordinate variables on the unit circle x and y . As shown in Formula (4), the output Y p of the phase real-time monitoring module x p and y p are also two coordinate variables. To ensure that it can be mapped to the unit circle, it is normalized through Formula (5), and finally the phase percentage error Y error is obtained through Formula (6);

[0023] (1);

[0024] (2);

[0025] (3);

[0026] (4);

[0027] (5);

[0028] (6);

[0029] Among them, Y p is the output of the phase real-time monitoring module, Y pnor is the output after normalization, Y error is the phase percentage error of the prediction result of the phase real-time monitoring module, and The two vectors are respectively composed of the true phase point P on the unit circle, the predicted phase point P p and the origin O of the coordinate system.

[0030] All data is divided into a training set and a validation set according to a custom ratio, and the type of intelligent optimization algorithm, the initial values of the hyperparameters of the phase real-time monitoring module, and the objective function to be optimized are determined. Then, the training set and the validation set are input into the phase real-time monitoring module. After repeated iterations, the optimal hyperparameters of the phase real-time monitoring module are obtained.

[0031] Preferably, the flexible capacitive sensor has excellent waterproof performance and environmental robustness, and can still work reliably at a water depth of 100 meters. In a strongly time-varying water flow environment, the sensor can still obtain muscle deformation signals with high signal-to-noise ratio.

[0032] Preferably, the real-time feedback device includes a signal receiver, a processing unit, and a feedback device;

[0033] The signal receiver receives the output from the phase real-time monitoring module and transmits the data to the subsequent processing unit.

[0034] The processing unit processes and analyzes the data, and generates corresponding real-time feedback signals according to preset algorithms and logics; the processing unit can be an embedded processor, a microcontroller, or a computer.

[0035] The feedback device gives the user corresponding real-time feedback information according to the preset feedback mode in the interaction module and the analysis result of the processing unit, and improves the swimming efficiency by optimizing the swimmer's swimming posture. The feedback device can be in the form of audition, vision, or touch.

[0036] Preferably, the technical analysis system is used to analyze the swimming data in the data storage module, understand the technical characteristics of the swimmer, quantify the training effect, and put forward improvement suggestions and swimming training plans. The technical analysis system can be intelligently generated by a big data model, designed by an expert coach, or customized by the user.

[0037] The present invention also provides a real-time feedback training method based on swimming phase monitoring. Applying the real-time feedback training system, the method includes the following steps:

[0038] S1: The swimmer puts on the intelligent wearable device, selects the required feedback mode in the interaction module, and inputs personal information, and then conducts swimming training;

[0039] S2: The intelligent wearable device collects the swimming data during the swimming process and uploads it to the phase real-time monitoring module and the data storage module;

[0040] S3: The phase real-time monitoring module classifies the swimming postures and continuously monitors the phases according to the muscle deformation signals;

[0041] S4: The real-time feedback device gives the user corresponding real-time feedback information according to the preset mode and the output of the phase real-time monitoring module, and improves the swimming efficiency by optimizing the swimmer's swimming posture;

[0042] S5: After the training is over, according to the swimming data in the data storage module, the technical analysis system evaluates the swimming technique of the wearer and puts forward improvement suggestions and swimming training plans.

[0043] The beneficial effects of the present invention are as follows:

[0044] 1. A waterproof process for the flexible capacitive sensor is designed, and the swimming strokes are classified and the phase is monitored through the physical quantity of muscle deformation. Moreover, the capacitive sensing unit and the acquisition circuit are waterproofed respectively by using waterproof silicone, so that it can be used normally underwater and a muscle deformation signal with high signal-to-noise ratio can be obtained. On the other hand, muscle deformation is the fundamental power source of the limb. Understanding the changes of muscle deformation signals during swimming can not only monitor the swimming phase in real time, but also establish the muscle coordination relationship, providing a promising new signal source for underwater motion monitoring;

[0045] 2. A two-stage machine learning model of classification-regression is proposed, which can be used for different swimming frequencies. While classifying the swimming strokes, it monitors the swimming phase under different swimming strokes, and the obtained is a continuous phase, with more complete information;

[0046] 3. An interaction module and a real-time feedback device are built, which helps to improve the swimming training efficiency, realize the personalized customization of users and expand the system functions. The interaction module is used for the user to input the current feedback mode, and the feedback device gives the corresponding real-time feedback information to the user according to the input feedback mode and the output of the phase real-time monitoring module, so as to improve the swimming efficiency by optimizing the swimming strokes of the swimmer;

[0047] 4. A technical analysis system is established, which helps the coach to analyze the technical characteristics of the swimmer in detail after the training, quantify the training effect, and put forward improvement suggestions and swimming training plans. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0049] Figure 1 It is a schematic technical route diagram of a real-time feedback training system and method based on swimming phase monitoring provided by an embodiment of the present invention;

[0050] Figure 2 It is the manufacturing process, exploded view and physical diagram of a flexible capacitive sensor of a real-time feedback training system and method based on swimming phase monitoring provided by an embodiment of the present invention;

[0051] Figure 3 It is the final structure diagram of a flexible capacitive sensor of a real-time feedback training system and method based on swimming phase monitoring provided by an embodiment of the present invention;

[0052] Figure 4 Flexible capacitance sensor performance diagram of a real-time feedback training system and method based on swimming phase monitoring provided by an embodiment of the present invention;

[0053] Figure 5 Schematic diagram of the measurement principle of muscle deformation;

[0054] Figure 6 Device wearing diagram of the swimming experiment;

[0055] Figure 7 Signal diagram of muscle deformation measured by the flexible sensor provided by an embodiment of the present invention;

[0056] Figure 8 Schematic diagram of converting the swimming phase percentage into coordinates on the unit circle.

[0057] Explanation of reference numerals: 1, fixed frame; 2, waterproof elastic layer; 3, upper electrode; 4, dielectric layer; 5, lower electrode; 6, upper end cap; 7, acquisition circuit; 8, lower end cap; 9, sealing cap; 10, flexible capacitance sensor. Detailed implementation manners

[0058] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0059] Embodiment 1, a real-time feedback training system based on swimming phase monitoring, includes:

[0060] An intelligent wearable device, which is used for a swimmer to wear during training, collect the swimming data of the wearer, and upload it to the phase real-time monitoring module and the data storage module;

[0061] The swimming data includes muscle deformation, current time, muscle balance, muscle strength and muscle fatigue;

[0062] A phase real-time monitoring module, which is used to monitor the swimming phase of different swimming strokes at different swimming frequencies to obtain the stroke category and continuous phase change;

[0063] An interaction module, which is used for the user to select the current required feedback mode and input personal information, so as to achieve personalized customization of the user and expand the functions of the system;

[0064] A communication module, which is used for communication inside or between the data acquisition system, the phase real-time monitoring module, the interaction module and the real-time feedback device;

[0065] A real-time feedback device for monitoring the output of a phase real-time monitoring module according to a preset feedback mode and phase, giving corresponding real-time feedback information to the user, and improving swimming efficiency by optimizing the swimming posture of the swimmer;

[0066] A data storage module for storing personal information and swimming data of the wearer; the storage method of the data storage module is one of local storage, cloud storage and memory storage; the personal information includes height, weight, age and gender; the swimming data includes muscle deformation, current time, muscle balance, muscle strength and muscle fatigue;

[0067] A technical analysis system for analyzing the swimming data in the data storage module, understanding the technical characteristics of the swimmer, quantifying the training effect, and putting forward improvement suggestions and swimming training plans;

[0068] The phase real-time monitoring module includes a swimming posture classifier in the first stage and a swimming phase regressor in the second stage. The swimming posture classifier can be a neural network, a support vector machine, a decision tree, or a linear discriminant analysis. The swimming posture classifier classifies the data of four swimming strokes, namely freestyle, breaststroke, backstroke, and butterfly stroke, based on the muscle deformation signal to obtain the swimming posture category of the current swimmer. The swimming phase regressor includes a freestyle phase regressor, a breaststroke phase regressor, a backstroke phase regressor, and a butterfly stroke phase regressor. The swimming phase regressor can be a linear regressor, a support vector regressor, a decision tree regressor, a random forest regressor, a gradient boosting regressor, and a neural network regressor. The phase real-time monitoring module inputs the current muscle deformation signal into the phase regressor corresponding to the swimming stroke according to the result of the first-stage swimming posture classification, and monitors the swimming phase of different swimming strokes at different swimming frequencies to obtain the continuous change of the phase.

[0069] Furthermore, the intelligent wearable device includes:

[0070] Flexible capacitive sensing system, including a plurality of flexible capacitive sensors, which are used to measure muscle deformation signals during swimming. It includes a waterproof elastic layer, a fixing frame, a housing, an upper electrode, a dielectric layer, a lower electrode and an acquisition circuit. The housing includes an upper end cover, a lower end cover and a sealing cover. The housing is made of waterproof material. Assembly holes and wiring holes for assembly and wiring are installed on the housing. Externally, it is used to fix the flexible capacitive sensing unit and provide a hard boundary for its deformation. The flexible capacitive sensing unit includes an upper electrode, a dielectric layer and a lower electrode; the dielectric layer is arranged between the upper electrode and the lower electrode. The flexible capacitive sensor is fixed at the belly of the muscle to be measured. When the muscle deforms, the dielectric layer is stressed and deformed, thereby changing the capacitance value. The waterproof elastic layer completely wraps the flexible capacitive sensor and fixes its periphery to the groove of the upper end cover through a fixing frame. Then, Ecoflex is poured into the groove for sealing, avoiding contact between the flexible capacitive sensing unit and water, thereby reducing crosstalk noise. The acquisition circuit is arranged inside the housing assembled by the upper end cover and the lower end cover. Wires are connected to the electrodes through the wire groove of the housing, which is used to acquire multi-channel capacitance signals and send all data to the host computer after synchronization. Waterproof silicone is poured into the housing through the opening of the lower end cover to completely immerse the acquisition circuit, and then sealed with a sealing cover to prevent water from contacting the acquisition circuit and causing a short circuit phenomenon.

[0071] Swimming tight suit, including swimming tight pants, swimming tight clothes and swimming fins. Nylon fasteners are sewn on the swimming tight suit, and the nylon fasteners are used for fixing the flexible capacitive sensors.

[0072] Furthermore, the interaction mode can be one of voice, touch, gesture and face forms.

[0073] Furthermore, the communication method of the communication module can be one of acoustic wave communication, electromagnetic wave communication, underwater acoustic optical fiber communication and wired cable communication.

[0074] Furthermore, the phase real-time monitoring module has undergone offline training before actual use, and the optimal hyperparameters are determined through intelligent optimization algorithms. Among them, the intelligent optimization algorithms can be genetic algorithms, ant colony algorithms, particle swarm algorithms, etc. The specific steps are as follows: First, a large amount of swimming data is obtained through swimming experiments, and preprocessing and feature extraction are carried out. Then, the real phase labels are obtained according to the synchronized experimental videos. Within a single swimming cycle, the real phase labels 0-100% are generated by linear interpolation. To avoid discontinuity when switching between adjacent cycles, that is, 99% directly changes to 0%, the error between the predicted phase and the real phase of the phase real-time monitoring module is expressed as the angle between two vectors; the specific steps are as follows. First, through formula (1), the swimming phase percentage is converted into an angle between 0 and 2π θ (Figure 8 ), and through formulas (2) and (3), it is further mapped to two coordinate variables on the unit circle x and y . As shown in formula (4), the output of the phase real-time monitoring module Y p is also two coordinate variables x p and y p . To ensure that it can be mapped to the unit circle, it is normalized through formula (5), and finally the phase percentage error Y error is obtained through formula (6);

[0075] (1);

[0076] (2);

[0077] (3);

[0078] (4);

[0079] (5);

[0080] (6);

[0081] wherein, Y p is the output of the phase real-time monitoring module, Y pnor is the output after normalization, Y error is the phase percentage error of the prediction result of the phase real-time monitoring module, and the two vectors are respectively composed of the true phase point P on the unit circle, the predicted phase point P p and the coordinate system origin O .

[0082] All data is divided into a training set and a validation set according to a custom ratio, and the type of intelligent optimization algorithm, the initial values of the hyperparameters of the phase real-time monitoring module, and the objective function to be optimized are determined. Then, the training set and the validation set are input into the phase real-time monitoring module. After repeated iterations, the optimal hyperparameters of the phase real-time monitoring module are obtained.

[0083] Furthermore, the flexible capacitive sensor has excellent waterproofness and environmental robustness and can still work reliably at a water depth of 100 meters. In a strongly time-varying water flow environment, the sensor can still obtain muscle deformation signals with high signal-to-noise ratio.

[0084] Furthermore, the real-time feedback device includes a signal receiver, a processing unit, and a feedback device;

[0085] The signal receiver receives the output from the phase real-time monitoring module and transmits the data to the subsequent processing unit.

[0086] The processing unit processes and analyzes the data and generates corresponding real-time feedback signals according to preset algorithms and logics; the processing unit can be an embedded processor, a microcontroller, or a computer.

[0087] The feedback device gives the user corresponding real-time feedback information according to the preset feedback mode in the interaction module and the analysis result of the processing unit, and improves the swimming efficiency by optimizing the swimming posture of the swimmer. The feedback device can be in the form of auditory (earphones, buzzers, etc.), visual (display screens, etc.), or tactile (electrical stimulation, vibration, etc.).

[0088] Furthermore, the technical analysis system is used to analyze the swimming data in the data storage module, understand the technical characteristics of the swimmer, quantify the training effect, and put forward improvement suggestions and swimming training plans. The scoring method can be intelligently generated by a big data model, designed by an expert coach, or user-defined.

[0089] The present invention also provides a real-time feedback training method based on swimming phase monitoring. Applying the real-time feedback training system, it includes the following steps:

[0090] S1: The swimmer puts on the intelligent wearable device, selects the required feedback mode in the interaction module, inputs personal information, and then conducts swimming training;

[0091] S2: The intelligent wearable device collects the swimming data during the swimming process and uploads it to the phase real-time monitoring module and the data storage module;

[0092] S3: The phase real-time monitoring module classifies the swimming posture and continuously monitors the phase according to the muscle deformation signal;

[0093] S4: The real-time feedback device gives the user corresponding real-time feedback information according to the preset mode and the output of the phase real-time monitoring module, and improves the swimming efficiency by optimizing the swimming posture of the swimmer;

[0094] S5: After the training ends, according to the swimming data in the data storage module, the technical analysis system evaluates the swimming technique of the wearer and puts forward improvement suggestions and swimming training plans.

[0095] Figure 2 Shows the manufacturing process, exploded view and physical diagram of the flexible capacitive sensor. The flexible capacitive sensor 10 includes a waterproof elastic layer 2, a fixing frame 1, a housing, an upper electrode 3, a dielectric layer 4, a lower electrode 5 and a acquisition circuit 7. The housing includes an upper end cover 6, a lower end cover 8 and a sealing cover 9. The housing is 3D printed from a waterproof material, photosensitive resin. Due to assembly and wiring requirements, there are assembly holes and wiring holes on the housing. The inside of the housing is used to place the acquisition circuit, and waterproofing is achieved by pouring Ecoflex. The outside is used to fix the sensing unit and provides a hard boundary for its deformation. The structure of the flexible capacitive sensor consists of an upper electrode 3, a dielectric layer 4 and a lower electrode 5;

[0096] The acquisition circuit 7 is controlled by STM32F103RCT6, and signal acquisition is performed by AD7746, and the data is sent to the computer host through the serial port; the acquisition circuit 7 has two CAN ports for transmitting data, a UART port for sending data to the host, and a SWD port for downloading programs.

[0097] Further, the dielectric layer is made of a composite structure of polyurethane (PU) and calcium copper titanate (CCTO). The upper electrode and the lower electrode are conductive fabrics, and they are assembled through a silicone adhesive.

[0098] Further, the assembled capacitive sensing unit is fixed at the boss of the upper end cover 6 through a silicone adhesive, the electrodes are led out through wires, and the wires are connected to the acquisition circuit 7 through a wire groove.

[0099] Further, the waterproof elastic layer 2 is wrapped around the capacitive sensing unit, and its periphery is fixed at the groove of the upper end cover 6 through the fixing frame 1, and then Ecoflex is poured into the groove for sealing.

[0100] Further, the acquisition circuit 7 is placed inside the upper end cover 6 and the lower end cover 8. Ecoflex is poured into the housing through the opening of the lower end cover and completely submerges the acquisition circuit 7, and finally sealed through the sealing cover 9.

[0101] Figure 3 Shows the final structure of the flexible capacitive sensor.

[0102] Further, during actual use, in order to reduce the crosstalk noise of the capacitance, the ground wire of the acquisition circuit 7 is brought into contact with water to achieve common grounding, thereby improving the signal-to-noise ratio.

[0103] In order to verify the performance of the flexible capacitive sensor, a series of tests were carried out. Figure 4 a shows the performance of the flexible capacitive sensor under loading and unloading, Figure 4b shows the noise reduction effect generated by grounding the acquisition circuit together with water. Figure 4 c shows the influence of water temperature and water pressure on the sensing performance. Figure 4 d shows the influence of water flow on the sensing performance. Figure 4 e shows the fatigue resistance characteristics of the sensor. After 10,000 cycles of compression, the signal still has good repeatability.

[0104] Furthermore, the flexible capacitive sensor 10 is fixed to the human body by a nylon adhesive tape.

[0105] Figure 5 Shows the measurement principle of muscle deformation.

[0106] In actual use, the user sets the current required feedback mode through the interaction module and inputs personal information, which will be uploaded to the data storage module to establish a database. In this embodiment, the feedback mode input through the interaction module is: in breaststroke, butterfly stroke, backstroke, and freestyle, when the phase reaches T1%, T2%, T3%, and T4% respectively, the swimmer needs to be prompted to perform hand propulsion. Subsequently, the swimmer wears the smart wearable device, and the flexible capacitive sensors 10 are respectively placed at the muscle bellies of the tibialis anterior (TA), rectus femoris (RF), and gluteus maximus (GMAX) of the left and right legs of the swimmer. The specific situation is as Figure 6 shown.

[0107] During the swimming process, the swimmer can freely adjust the swimming posture and swimming frequency. The flexible capacitive sensor collects the muscle deformation signal and uploads it to the phase real-time monitoring module.

[0108] First, the long short-term memory network (LSTM) classifier in the first stage identifies the current swimming posture. Subsequently, according to the swimming posture category, the data is input into the phase regressor of the corresponding swimming posture, and the current phase percentage is obtained. The real-time feedback device receives the swimming posture category and the continuous phase change information from the phase real-time monitoring module. When the received information is one of the breaststroke phase T1%, butterfly stroke phase T2%, backstroke phase T3%, and freestyle phase T4%, the feedback device will emit a vibration for 1 s, the waterproof LED light will flash continuously, and at the same time, a voice of "hand propulsion" will occur to prompt the swimmer to change the swimming posture, thereby improving the swimming efficiency.

[0109] Figure 7 The muscle deformation signals of different swimming postures are given in, where lTA, rTA, lRF, rRF, lGMAX, and rGMAX respectively represent the left tibialis anterior, right tibialis anterior, left rectus femoris, right rectus femoris, left gluteus maximus, and right gluteus maximus.

[0110] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these modifications and variations.

Claims

1. A real-time feedback training system based on swimming phase monitoring, characterized in that, Comprising: An intelligent wearable device for swimmers to wear during training, collect the swimming data of the wearer, and upload it to the phase real-time monitoring module and the data storage module; The swimming data includes muscle deformation, current time, muscle balance, muscle strength, and muscle fatigue; The phase real-time monitoring module is used to monitor the swimming phases of different swimming strokes at different swimming frequencies to obtain the stroke category and continuous phase change; The interaction module is used for the user to select the current required feedback mode and input personal information, so as to achieve personalized customization of the user and expand the functions of the system; The communication module is used for communication within or between the data acquisition system, the phase real-time monitoring module, the interaction module, and the real-time feedback device; The real-time feedback device is used to give the user corresponding real-time feedback information according to the preset feedback mode and the output of the phase real-time monitoring module, and improve the swimming efficiency by optimizing the swimmer's stroke; The data storage module is used to store the personal information and swimming data of the wearer; The storage method of the data storage module is one of local storage, cloud storage, and memory storage; the personal information includes height, weight, age, and gender; The swimming data includes muscle deformation, current time, muscle balance, muscle strength, and muscle fatigue; The technical analysis system is used to analyze the swimming data in the data storage module, understand the technical characteristics of the swimmer, quantify the training effect, and put forward improvement suggestions and swimming training plans; The phase real-time monitoring module includes a swimming stroke classifier in the first stage and a swimming phase regressor in the second stage. The swimming stroke classifier is one of a neural network, a support vector machine, a decision tree, and a linear discriminant analysis. The swimming stroke classifier classifies the data of four swimming strokes, namely freestyle, breaststroke, backstroke, and butterfly stroke, based on the muscle deformation signal to obtain the swimming stroke category of the current swimmer. The swimming phase regressor includes a freestyle phase regressor, a breaststroke phase regressor, a backstroke phase regressor, and a butterfly stroke phase regressor. The swimming phase regressor is one of a linear regressor, a support vector regressor, a decision tree regressor, a random forest regressor, a gradient boosting regressor, and a neural network regressor. The phase real-time monitoring module inputs the current muscle deformation signal into the phase regressor corresponding to the swimming stroke according to the result of the first-stage swimming stroke classification, and monitors the swimming phases of different swimming strokes at different swimming frequencies to obtain the continuous change of the phase.

2. The real-time feedback training system based on swimming phase monitoring according to claim 1, characterized in that, The intelligent wearable device includes: Flexible capacitive sensing system, including a plurality of flexible capacitive sensors, which are used to measure muscle deformation signals during swimming. It includes a waterproof elastic layer, a fixing frame, a housing, an upper electrode, a dielectric layer, a lower electrode and a acquisition circuit. The housing includes an upper end cover, a lower end cover and a sealing cover. The housing is made of waterproof material. Assembly holes and wire holes for assembly and wiring are installed on the housing. It is used to fix the flexible capacitive sensing unit externally and provide a hard boundary for its deformation. The flexible capacitive sensing unit includes an upper electrode, a dielectric layer and a lower electrode. The dielectric layer is arranged between the upper electrode and the lower electrode. The flexible capacitive sensor is fixed at the belly of the muscle to be measured. When the muscle deforms, the dielectric layer is stressed and deformed, thereby changing the capacitance value. The waterproof elastic layer completely wraps the flexible capacitive sensor and fixes its periphery to the groove of the upper end cover through a fixing frame. Then, Ecoflex is poured into the groove for sealing. The acquisition circuit is arranged inside the housing assembled by the upper end cover and the lower end cover. Wires are connected to the electrodes through the wire grooves of the housing, which is used to acquire multi-channel capacitance signals and send all the data to the host computer after synchronization. Waterproof silicone is poured into the housing through the opening of the lower end cover to completely immerse the acquisition circuit, and then sealed with a sealing cover. Swimming tight suit, including swimming tight pants, swimming tight clothes and swimming fins. Nylon fasteners are sewn on the swimming tight suit, and the nylon fasteners are used to fix the flexible capacitive sensors.

3. The real-time feedback training system based on swimming phase monitoring according to claim 1, characterized in that, The interaction module is one of voice, touch, gesture and facial forms.

4. The real-time feedback training system based on swimming phase monitoring according to claim 1, wherein The communication method of the communication module is one of acoustic wave communication, electromagnetic wave communication, underwater acoustic optical fiber communication and wired cable communication.

5. The real-time feedback training system based on swimming phase monitoring according to claim 1, characterized in that The phase real-time monitoring module has undergone offline training before actual use, and the optimal hyperparameters are determined through an intelligent optimization algorithm. Among them, the intelligent optimization algorithm is one of genetic algorithm, ant colony algorithm and particle swarm algorithm. The specific steps are as follows: First, a large amount of swimming data is obtained through swimming experiments, and preprocessing and feature extraction are carried out. Then, real phase labels are obtained according to the synchronized experimental videos. Within a single swimming cycle, real phase labels from 0 - 100% are generated by linear interpolation. To avoid discontinuity when switching between adjacent cycles, that is, 99% directly changes to 0%, the error between the predicted phase and the real phase of the phase real-time monitoring module is represented by the angle between two vectors. The specific steps are as follows: First, through formula (1), the swimming phase percentage is converted into an angle between 0 and 2π θ , and through formulas (2) and (3), it is further mapped to two coordinate variables on the unit circle x and y， As shown in formula (4), the output of the phase real-time monitoring module Y p is also two coordinate variables x p and y p , in order to ensure that it can be mapped to the unit circle, it is normalized through formula (5), and finally the phase percentage error is obtained through formula (6) Y error； （1）； （2）； （3）； （4）； （5）； （6）； Among them, Y p is the output of the phase real-time monitoring module, Y pnor is the output after normalization processing, Y error is the phase percentage error of the prediction result of the phase real-time monitoring module, and The two vectors are respectively composed of the true phase points P on the unit circle, the predicted phase points P p and the origin of the coordinate system O ; All data is divided into a training set and a validation set according to a custom ratio, and the type of the intelligent optimization algorithm, the initial values of the hyperparameters of the phase real-time monitoring module and the objective function to be optimized are determined. Then, the training set and the validation set are input into the phase real-time monitoring module. After repeated iterations, the optimal hyperparameters of the phase real-time monitoring module are obtained.

6. The real-time feedback training system based on swimming phase monitoring according to claim 2, wherein, The flexible capacitive sensor has excellent waterproofness and environmental robustness and can still work reliably at a water depth of 100 meters. In a strongly time-varying water flow environment, the sensor can still obtain muscle deformation signals with high signal-to-noise ratio.

7. The real-time feedback training system based on swimming phase monitoring according to claim 1, characterized in that The real-time feedback device includes a signal receiver, a processing unit and a feedback device; The signal receiver receives the output from the phase real-time monitoring module and transmits the data to the subsequent processing unit; The processing unit processes and analyzes the data, and generates corresponding real-time feedback signals according to the preset algorithms and logics; The processing unit is one of an embedded processor, a microcontroller, or a computer; The feedback device gives the user corresponding real-time feedback information according to the preset feedback mode in the interaction module and the analysis result of the processing unit, improves the swimming efficiency by optimizing the swimmer's swimming posture, and the feedback device is in the form of audition, vision, or touch.

8. The real-time feedback training system based on swimming phase monitoring according to claim 1, wherein, The technical analysis system is used to analyze the swimming data in the data storage module, understand the technical characteristics of the swimmer, quantify the training effect, and put forward improvement suggestions and swimming training plans. The technical analysis system is intelligently generated by a big data model, designed by an expert coach, or customized by the user.

9. A real-time feedback training method based on swimming phase monitoring, characterized in that, Applying the real-time feedback training system according to any one of claims 1-8, comprising the following steps: S1: The swimmer puts on the intelligent wearable device, selects the required feedback mode in the interaction module, and inputs personal information, and then conducts swimming training; S2: The intelligent wearable device collects the swimming data during the swimming process and uploads it to the phase real-time monitoring module and the data storage module; S3: The phase real-time monitoring module conducts swimming posture classification and phase continuous monitoring according to the muscle deformation signal; S4: The real-time feedback device gives the user corresponding real-time feedback information according to the preset mode and the output of the phase real-time monitoring module, and improves the swimming efficiency by optimizing the swimmer's swimming posture; S5: After the training ends, according to the swimming data in the data storage module, the swimming technique of the wearer is evaluated by the technical analysis system, and improvement suggestions and swimming training plans are put forward.

Images