Personalized conditioning device for detecting and memorizing human physiological data

Abstract

The invention discloses a personalized conditioning device for detecting and memorizing human physiological data. The personalized conditioning device comprises a sensor system, a data processing unit, a cloud server, a user interface and a massage module, the sensor system obtains body information and posture information of a user and transmits the body information and the posture information to the data processing unit through communication, and the data processing unit sends the received information to the cloud server for processing and receives a user body 3D model and massage parameter data returned by the cloud server. The user interface is used for displaying health monitoring results and guidance suggestions to a user and controlling the massage module to execute massage according to the massage parameter data. Compared with the prior art, the 3D model can be constructed according to the sensor data, and the user can obtain customized massage suggestions and health guidance according to the physical conditions and requirements of the user.

Description

Technical Field

[0001] The present invention relates to the field of intelligent control technology, and more specifically, to a personalized conditioning device for detecting and memorizing human physiological data. Background Art

[0002] As a health care device, massage chairs have been widely used to relieve muscle tension, eliminate fatigue and promote blood circulation. However, traditional massage chairs usually only provide standardized massage programs and cannot be personalized according to the user's height, weight and muscle tension, which limits the maximization of massage effects. For example, for tall users, the preset program may not cover parts such as the neck; while for users with low muscle tension, the standard strength may be too strong and cause discomfort. In addition, massage chairs that lack memory functions cannot record users' usage habits and historical data, which is not conducive to long-term health management.

[0003] The disadvantages of the existing technology are that it cannot accurately identify the user's body shape and muscle condition, and lacks the ability to adaptively adjust the personalized massage intensity. These limitations result in the massage chair being unable to provide the best user experience.

[0004] The prior art discloses an intelligent massage method and massage device based on facial recognition, the method comprising the following steps: obtaining detected user body data through a facial recognition interface; uploading the obtained user body data to a cloud platform, obtaining a massage recommendation combination of the cloud platform, and performing the first massage action; after the first massage, obtaining the detected body data after the first massage through the facial recognition interface again; calculating a correction amount for the recommended massage combination, adjusting adjustable parameters according to the correction amount to obtain a recommended combination for the next massage, and performing the next massage action; collecting the user's evaluation data after each massage and storing it in historical evaluation data. The defect of this scheme is that it does not adaptively adjust the massage intensity and position in combination with electromyographic signals.

[0005] To this end, in combination with the above requirements and the defects of the prior art, the present application proposes a personalized conditioning device for detecting and memorizing human physiological data. Summary of the invention

[0006] The present invention provides a personalized conditioning device for detecting and memorizing human physiological data, which can build a 3D model based on sensor data. Users can obtain customized massage suggestions and health guidance based on their own physical conditions and needs.

[0007] The primary purpose of the present invention is to solve the above technical problems. The technical solutions of the present invention are as follows:

[0008] A first aspect of the present invention provides a personalized conditioning device for detecting and memorizing human physiological data, comprising: a sensor system, a data processing unit, a cloud server, a user interface and a massage module; the sensor system obtains user body information and posture information and transmits it to the data processing unit through communication, the data processing unit sends the received information to the cloud server for processing, and receives the user's body 3D model and massage parameter data returned by the cloud server, the user interface is used to display health monitoring results and guidance to the user, and control the massage module to perform massage according to the massage parameter data.

[0009] Furthermore, the sensor system includes: a depth camera, an infrared sensor, a pressure sensor, an inertial sensor and a surface electromyography sensor, wherein the depth camera is used to capture the user's 3D point cloud data and obtain the user's body shape and contour, the infrared sensor is used to scan the user's body and obtain contour and size information, the pressure sensor is arranged on the surface of the conditioning device to detect the pressure distribution of various parts of the user's body, the inertial sensor is used to capture the user's body posture and movement information, and the surface electromyography sensor is used to obtain the user's electromyography signal in real time.

[0010] Furthermore, the specific step of sending the received information to the cloud server for processing includes: synchronizing the data of different sensors in time and transmitting them to the cloud server through the data processing unit, using Kalman filtering in the cloud server to remove noise in the data, then normalizing the data, identifying and processing the missing data, using the interpolation algorithm to fill the missing values, calculating the 3D coordinates corresponding to each pixel according to the depth value and the camera intrinsic parameter, converting the collected data into point cloud data, using the filtering algorithm to remove isolated points and noise points, downsampling, and then using the feature matching algorithm for preliminary alignment, extracting feature points and descriptors from the point cloud, and then minimizing the distance between corresponding points through an iterative optimization algorithm, using a surface reconstruction algorithm to generate a smooth 3D surface, using computer vision technology to segment the 3D model, mapping the data of the pressure sensor to the 3D model, analyzing the pressure distribution of various parts of the user's body, identifying the pressure concentration area, using the depth camera and inertial sensor data to perform posture estimation, obtaining the posture information of various parts of the user's body, using a deep learning posture estimation network to detect key joints of the user's body, and automatically marking various parts of the user's body according to the joint point information and the 3D model.

[0011] Furthermore, the generation of the massage parameters includes the following steps: starting a surface electromyography sensor to obtain electromyographic signals from the user's body, decomposing the electromyographic signals using continuous variational mode decomposition, decomposing the signals into a set of intrinsic mode functions, calculating the correlation coefficient of each intrinsic mode function, and calculating a threshold value according to a dual correlation coefficient standard, dividing all intrinsic mode functions into three categories according to the threshold value: pure intrinsic mode functions, mixed intrinsic mode functions, and noisy intrinsic mode functions, denoising the mixed intrinsic mode functions, and after removing the noisy intrinsic mode functions, extracting multidimensional features, inputting them into a preset random forest regression model, outputting the muscle tension level, and calculating the massage parameters corresponding to the muscle tension level according to the preset standards.

[0012] Furthermore, the continuous variation mode decomposition decomposes the electromyographic signal and comprises the following steps:

[0013] The input signal f(t) is divided into two parts:

[0014] f(t)=u L (t)+f r (t)

[0015] In the formula, u L (t) is the Lth order mode, f r (t) is the residual signal, which includes the sum of the first-order mode and the (L-1)-order mode. The unprocessed signal is denoted by f u (t); When each mode is closely around its center frequency, the minimization constraint is as follows:

[0016]

[0017] where w L is the center frequency of the Lth mode, * is the convolution operation, is the partial derivative with respect to time t, and δ(t) is the Dirac function.

[0018] Furthermore, in f r (t) and u L Select the appropriate filter when the spectrum overlap of (t) is minimal. L (t) is the impulse response of the filter to the Lth mode, and its frequency response expression is:

[0019]

[0020] In the formula, α is the maximum equilibrium parameter, and J 2 The constraints are:

[0021]

[0022] The first-order mode is processed similarly to the (L-1)-order mode, assuming that β i (t) is the impulse response of the filter to the i-th order mode, and its frequency response is obtained

[0023]

[0024] Build J 3 The constraints are:

[0025]

[0026] To ensure that the assumptions hold, establish constraints:

[0027]

[0028] Build a constrained minimization model:

[0029]

[0030] According to the solution idea of ​​the model in the VMD algorithm, the augmented Lagrangian function is constructed to obtain the optimal solution:

[0031]

[0032] where λ is the Lagrange multiplier.

[0033] Furthermore, according to Parseval's theorem, the above formula can be transformed into:

[0034]

[0035] The alternating direction multiplier method can be used to iteratively solve the problem:

[0036]

[0037] In the formula, represents the Fourier transform of the original signal f(t), n is the number of iterations, and τ is the iteration step. express At the center frequency The Fourier transform result of the L mode at the nth iteration is shown in Figure 2, and the other modes are similar.

[0038] Furthermore, the calculation process of the dual correlation coefficient is as follows: First, calculate each eigenmode function u k The correlation coefficient between (t) and the original input signal f(t):

[0039]

[0040] Where CC(k) represents the correlation coefficient of the kth intrinsic mode function, and N represents the sample length of the signal;

[0041] The dual threshold screening of the dual correlation coefficient standard, the calculation threshold is as follows:

[0042]

[0043] Where P 1 and P 2 Respectively represent the high threshold (Hth) and the low threshold (Lth), CC max and CC aνg Represent the maximum and average values ​​of the correlation coefficients of each mode, respectively.

[0044] Further, the training process of the preset random forest regression model includes the following steps: the extracted feature data is divided into a training set, a validation set and a test set. The data is divided into 70% for training, 15% for validation and 15% for testing. Each tree is generated based on different random feature subsets and sample subsets, and the error within the node is minimized by selecting the best split point. After all trees grow independently, the random forest regression model averages the prediction results of each tree to obtain the final output value. In order to improve the performance of the model, it is usually necessary to tune the hyperparameters. Through the random search method, some combinations are randomly selected in the hyperparameter space for validation to select the optimal parameters. The validation set is used to evaluate the performance of the model. The evaluation indicators are: mean square error, determination coefficient, and finally the test set is used to evaluate the performance of the model on unseen data; as the conditioning device is used, the sensor will continuously collect new electromyographic signals and extract features. The muscle tension scores corresponding to these features can be corrected by user feedback or model prediction results, and the old data are weighted to ensure the sensitivity of the model to new trends. Combine new and old data and repeat the training process.

[0045] Furthermore, the random forest regression model predicts the current muscle tension based on the trained parameters. The muscle tension is output in the form of a quantitative score, with a value in the range of 0-100 representing the current muscle tension level; the data processing unit monitors the electromyographic signal in real time and uploads it to the cloud server, adjusts the massage parameters in real time according to the calculated muscle tension, and adjusts the massage intensity and frequency in real time according to the real-time changing electromyographic signal; the massage module adjusts the massage process in real time according to the dynamic massage parameters, and the user can set the evaluation time interval and massage plan according to the needs through the user interface.

[0046] Compared with the prior art, the technical solution of the present invention has the following beneficial effects:

[0047] The present invention provides a personalized conditioning device for detecting and memorizing human physiological data. According to the data collected by the sensor, a 3D model of the user's body is created in the cloud, which can realize the recognition and analysis of various body parts of the user, thereby providing a more accurate and comfortable massage experience; combined with the analysis of electromyographic signals, it can monitor the user's muscle tension in real time, and adjust the massage intensity according to the data, providing personalized and comfortable massage, which can enhance the user's comfort and massage effect, and improve the user experience of the massage chair. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 This is a schematic diagram of a personalized conditioning device for detecting and memorizing human physiological data according to the present invention.

[0049] Figure 2 Schematic diagram of system modules of a massage chair in one embodiment of the present invention.

[0050] Figure 3 The present invention is a flowchart of calculating massage parameters in one embodiment of the present invention. DETAILED DESCRIPTION

[0051] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0052] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0053] Example 1

[0054] like Figure 1 As shown, the present invention provides a personalized conditioning device for detecting and memorizing human physiological data, including: a sensor system, a data processing unit, a cloud server, a user interface and a massage module; the sensor system obtains user body information and posture information and transmits it to the data processing unit through communication, the data processing unit sends the received information to the cloud server for processing, and receives the user's body 3D model and massage parameter data returned by the cloud server, the user interface is used to display health monitoring results and guidance to the user, and control the massage module to perform massage according to the massage parameter data.

[0055] In this embodiment, a storage module is also included for storing massage parameters returned each time. The massage module includes a massage movement walking module, a massage movement kneading and tapping module, and the data processing unit sends signals to the massage movement walking module, the massage movement kneading and tapping module according to the received massage parameters, wherein the massage movement walking module is driven by a servo motor, and the movement of the massage movement up and down is driven by the walking drive motor to drive the gear on the rack to send the massage movement to a specified position, and the speed and direction of the kneading motor are adjusted by controlling the kneading motor drive circuit, so as to achieve the purpose of adjusting the speed of the kneading rate and the kneading intensity control, and the tapping massage achieves the effect of adjusting the magnitude of the tapping intensity by controlling the output power of the tapping motor.

[0056] The sensor system includes: a depth camera, an infrared sensor, a pressure sensor, an inertial sensor and a surface electromyography sensor. The depth camera is used to capture the user's 3D point cloud data and obtain the user's body shape and contour. The infrared sensor is used to scan the user's body and obtain contour and size information. The pressure sensor is arranged on the surface of the conditioning device to detect the pressure distribution of various parts of the user's body. The inertial sensor is used to capture the user's body posture and movement information. The surface electromyography sensor is used to obtain the user's electromyography signal in real time.

[0057] The specific steps of sending the received information to the cloud server for processing include: synchronizing the data of different sensors in time and transmitting them to the cloud server through the data processing unit, using Kalman filtering in the cloud server to remove noise in the data, then normalizing the data, identifying and processing the lost data, using the interpolation algorithm to fill the missing values, calculating the 3D coordinates corresponding to each pixel according to the depth value and the camera internal parameters, converting the collected data into point cloud data, using the filtering algorithm to remove isolated points and noise points, downsampling, and then using the feature matching algorithm for preliminary alignment, extracting feature points and descriptors from the point cloud, and then minimizing the distance between corresponding points through an iterative optimization algorithm, using a surface reconstruction algorithm to generate a smooth 3D surface, using computer vision technology to segment the 3D model, mapping the data of the pressure sensor to the 3D model, analyzing the pressure distribution of various parts of the user's body, identifying the pressure concentration area, using the depth camera and inertial sensor data to perform posture estimation, obtaining the posture information of various parts of the user's body, using a deep learning posture estimation network to detect key joints of the user's body, and automatically marking various parts of the user's body according to the joint information and the 3D model.

[0058] The generation of massage parameters includes the following steps: starting a surface electromyography sensor to obtain electromyographic signals from a user's body, decomposing the electromyographic signals using continuous variational mode decomposition, decomposing the signals into a set of intrinsic mode functions, calculating the correlation coefficient of each intrinsic mode function, and calculating a threshold value according to a dual correlation coefficient standard, dividing all intrinsic mode functions into three categories according to the threshold value: pure intrinsic mode functions, mixed intrinsic mode functions, and noisy intrinsic mode functions, denoising the mixed intrinsic mode functions, and after removing the noisy intrinsic mode functions, extracting multidimensional features, inputting them into a preset random forest regression model, outputting a muscle tension level, and calculating massage parameters corresponding to the muscle tension level according to preset standards.

[0059] The muscle tension is output in the form of a quantitative score, with a value in the range of 0-100 representing the current tension level of the muscle; the data processing unit monitors the electromyographic signal in real time and uploads it to the cloud server, adjusts the massage parameters in real time according to the calculated muscle tension, and adjusts the massage intensity and frequency in real time according to the real-time changing electromyographic signal; the massage module adjusts the massage process in real time according to the dynamic massage parameters, and the user can set the evaluation time interval and massage plan according to the needs through the user interface.

[0060] According to the above technical features, the present invention uses high-precision sensors and depth-sensing cameras, combined with data processing and machine learning technology, to create a 3D model of the user's body, and realize the recognition and analysis of various body parts of the user. The massage chair can accurately identify the body shapes of different users, thereby providing a more accurate and comfortable massage experience.

[0061] Combining surface electromyography sensors and intelligent algorithms, the massage chair can monitor the user's muscle tension in real time and adjust the massage intensity based on the data to provide personalized and comfortable massage. This adaptive adjustment of massage intensity can enhance the user's comfort and massage effect, and improve the user experience of the massage chair.

[0062] The storage module records the user's data, which can save time the next time you use it. At the same time, it can continuously learn and optimize the massage algorithm by recording the user's muscle status, feedback information and usage habits, helping users manage and track their health.

[0063] Example 2

[0064] Based on the above embodiment 1, combined with Figure 3 This embodiment describes in detail the specific process of calculating massage parameters of the present invention.

[0065] In a specific embodiment, Figure 2As shown in the figure, first start the surface electromyography sensor, use the data acquisition system to obtain the electromyographic signal in real time, send the collected signal to the main control module, the main control module connects to the network through the communication module, and sends the signal to the cloud for processing. Successive Variational Mode Decomposition (SVMD) is used to decompose the electromyographic signal and decompose the signal into a set of intrinsic mode functions (IMFs).

[0066] The calculation process of continuous variational mode decomposition (SVMD) is as follows:

[0067] Assume that the input signal f(t) is divided into two parts: f(t) = u L (t)+f r (t) where u L (t) is the Lth order mode, f r (t) is the residual signal, which includes the sum of the first-order mode and the (L-1)-order mode. The unprocessed signal is denoted by f u (t). Then we can get In order to realize the above assumptions, the following constraints need to be set and met:

[0068] Step 1: With each mode tightly centered around its center frequency, the minimization constraints are as follows:

[0069]

[0070] where w L is the center frequency of the Lth mode, * is the convolution operation, is the partial derivative with respect to time t, and δ(t) is the Dirac function.

[0071] Step 2: On f r (t) and u L Select the appropriate filter when the spectrum overlap of (t) is minimal. L (t) is the impulse response of the filter to the Lth mode, and its frequency response expression is:

[0072]

[0073] In the formula, α is the maximum equilibrium parameter, and J 2 The constraints are:

[0074]

[0075] Step 3: Perform similar processing on the first-order mode as that on the (L-1)-order mode, assuming β i (t) is the impulse response of the filter to the i-th order mode, and its frequency response is obtained

[0076]

[0077] Build J 3 The constraints are:

[0078]

[0079] Step 4: To ensure that the assumptions hold, establish constraints:

[0080]

[0081] Build a constrained minimization model:

[0082]

[0083] According to the solution idea of ​​the model in the VMD algorithm, the augmented Lagrangian function is constructed to obtain the optimal solution:

[0084]

[0085] where λ is the Lagrange multiplier.

[0086] According to Parseval's theorem, the above formula can be transformed into:

[0087]

[0088] The alternating direction multiplier method can be used to iteratively solve the problem:

[0089]

[0090] In the formula, represents the Fourier transform of the original signal f(t), n is the number of iterations, and τ is the iteration step. express At the center frequency The Fourier transform result of the L mode at the nth iteration is shown in Figure 2, and the other modes are similar.

[0091] After the EMG signal is decomposed by continuous variation mode, the correlation coefficient of each IMF is calculated, and the threshold is calculated according to the dual correlation coefficient criterion. In addition, all IMFs are divided into three categories according to the threshold: pure IMF, mixed IMF, and noisy IMF.

[0092] The calculation process of the dual correlation coefficient is as follows:

[0093] First, calculate each eigenmode function u k The correlation coefficient between (t) and the original input signal f(t):

[0094]

[0095] Where CC(k) represents the correlation coefficient of the kth intrinsic mode function, and N represents the sample length of the signal.

[0096] The dual threshold screening of the dual correlation coefficient standard, the calculation threshold is as follows:

[0097]

[0098] Where P 1 and P 2 Respectively represent the high threshold (Hth) and the low threshold (Lth), CC max and CC aνg Represent the maximum and average values ​​of the correlation coefficients of each mode, respectively.

[0099] These two thresholds are adaptive and can change according to the change of the signal. When the correlation coefficient of the intrinsic mode function is greater than the high threshold, it is considered that the intrinsic mode function has a good correlation with the original signal and is considered a pure intrinsic mode function; when the correlation coefficient value of the intrinsic mode function is between the two thresholds, it is identified as a mixed intrinsic mode function, and the corresponding intrinsic mode function is denoised by Savitzky-Gorey filtering (SG filter); when the correlation coefficient of the intrinsic mode function is less than the low threshold, it is regarded as a noise intrinsic mode function and needs to be removed.

[0100] Extract multi-dimensional features from the preprocessed EMG signals, including mean, interquartile range, median absolute deviation, range, variance, skewness, kurtosis, signal energy, absolute amplitude sum, and maximum kurtosis. Use these extracted features as input and input them into the random forest regression model. Training the random forest regression model requires the following steps: Divide the extracted feature data into training set, validation set, and test set. The data is divided into 70% for training, 15% for validation, and 15% for testing. Each tree is generated based on different random feature subsets and sample subsets. The error within the node is minimized by selecting the best split point. After all trees grow independently, the random forest regression model averages the prediction results of each tree to obtain the final output value. In order to improve the performance of the model, it is usually necessary to tune the hyperparameters. Through the random search method, some combinations are randomly selected in the hyperparameter space for validation to select the optimal parameters. Use the validation set to evaluate the performance of the model. The evaluation indicators are: mean square error, determination coefficient, and finally the test set is used to evaluate the performance of the model on unseen data. The random forest regression model needs to be continuously updated based on new data to maintain its prediction accuracy. As the massage chair continues to be used, the sensor will continuously collect new electromyographic signals and extract features. The muscle tension scores corresponding to these features can be corrected through user feedback or model prediction results. The old data is weighted to ensure the model's sensitivity to new trends. The training process is repeated by combining new and old data.

[0101] The model predicts the current muscle tension based on the trained parameters. The muscle tension is output in the form of a quantitative score, with a value in the range of 0-100 to represent the current muscle tension level. Each score has a corresponding massage parameter (the massage intensity of the kneading and tapping modules of the massage movement).

[0102] Example 3

[0103] Based on Example 1 and Example 2, Figure 2 As shown, this embodiment takes a massage chair as an application object to explain in detail the application results of the present invention.

[0104] In this embodiment, Figure 2 As shown, it includes a main control module, which is connected to the cloud server through the communication module ESP8266. The relevant technical process of the massage chair to realize the recognition of various body parts of the user is as follows:

[0105] (1) The sensor selection is as follows:

[0106] A depth camera is used to capture the user's 3D point cloud data and obtain the user's body shape and contour.

[0107] Infrared sensor that scans the user's body to obtain its contour and size information.

[0108] Pressure sensors are installed on the surface of the massage chair to detect the pressure distribution of various parts of the user's body.

[0109] Inertial sensors are used to capture user's body posture and movement information.

[0110] (2) Start the sensor, synchronize the data from different sensors in time and send them to the main control module. The main control module communicates with the server through the communication module and sends the data to the cloud for processing and analysis.

[0111] (3) Kalman filtering is used on the cloud to remove noise from the data, and then the data is normalized, missing data is identified and processed, and missing values ​​are filled using an interpolation algorithm. Data from different sensors are aligned to the same coordinate system to ensure data consistency. The 3D coordinates corresponding to each pixel are calculated based on the depth value and the camera intrinsic parameters, and the collected data is converted into point cloud data. Then point cloud processing is performed, and isolated points and noise points are removed using a filtering algorithm to achieve the purpose of denoising. Downsampling is performed to reduce the density of the point cloud and improve processing efficiency. Then, a feature matching algorithm is used for preliminary alignment, and feature points and descriptors are extracted from the point cloud for initial alignment. Then, an iterative optimization algorithm is used to minimize the distance between corresponding points for fine alignment. A surface reconstruction algorithm is used to generate a smooth 3D surface, and computer vision technology is used to segment the 3D model. The data from the pressure sensor is mapped to the 3D model, and the pressure distribution of various parts of the user's body is analyzed to identify the pressure concentration areas, which usually correspond to the key body parts of the user. Using the depth camera and inertial sensor data, posture estimation is performed to obtain posture information of various parts of the user's body. Use a deep learning posture estimation network to detect the key joints of the user's body, and automatically mark the user's body parts (such as head, shoulders, back, waist, legs, hips, etc.) based on the joint information and 3D model.

[0112] (4) Based on the user's height, weight, pressure distribution and body part information, the personalized massage parameters (the position of the massage movement) are calculated.

[0113] The main control module will send a signal to the massage movement walking module according to the received massage parameters. The massage movement walking module is driven by a servo motor. The massage movement moves up and down by the walking drive motor driving the gear on the rack to send the massage movement to the specified position. The main control module will also send the massage parameters to the kneading and tapping modules of the massage movement, and adjust the speed and direction of the kneading motor by controlling the kneading motor drive circuit, so as to achieve the purpose of adjusting the speed of kneading and kneading intensity. The tapping massage achieves the magnitude of the tapping intensity by controlling the output power of the tapping motor.

[0114] Repeat (2) to (4), and the main control module sends the massage parameters received each time to the storage module for storage.

[0115] Finally, the stored data is used to identify users and help them manage and track their health.

[0116] In the embodiments provided in the present application, it should be understood that the disclosed systems and methods can be implemented in other ways. A person of ordinary skill in the art can understand that all or part of the steps of implementing the above method embodiments can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium, which, when executed, executes the steps of the above method embodiments; and the aforementioned storage medium includes: a mobile storage device, a read-only memory (ROM), a random access memory (RAM), a disk or an optical disk, and other media capable of storing program codes.

[0117] Alternatively, if the above-mentioned embodiments of the present invention are implemented in the form of software function modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiments of the present invention can be essentially or partly reflected in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for a computer device to execute all or part of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: various media capable of storing program codes, such as mobile storage devices, ROM, RAM, magnetic disks or optical disks.

[0118] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the embodiments of the present invention. The icons describing the structural positional relationships in the accompanying drawings are only used for exemplary purposes and are not to be construed as limiting the present invention. For those of ordinary skill in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and is not possible to list all the embodiments here. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention shall be included in the scope of protection of the claims of the present invention.

Claims

1. A personalized conditioning device for detecting and memorizing human physiological data, characterized in that: It includes: a sensor system, a data processing unit, a cloud server, a user interface and a massage module; the sensor system obtains the user's body information and posture information and transmits it to the data processing unit through communication; the data processing unit sends the received information to the cloud server for processing, and receives the user's body 3D model and massage parameter data returned by the cloud server; the user interface is used to display health monitoring results and guidance to the user, and control the massage module to perform massage according to the massage parameter data.

2. A personalized conditioning device for detecting and memorizing human physiological data according to claim 1, characterized in that: The sensor system includes: a depth camera, an infrared sensor, a pressure sensor, an inertial sensor and a surface electromyography sensor. The depth camera is used to capture the user's 3D point cloud data and obtain the user's body shape and contour. The infrared sensor is used to scan the user's body and obtain contour and size information. The pressure sensor is arranged on the surface of the conditioning device to detect the pressure distribution of various parts of the user's body. The inertial sensor is used to capture the user's body posture and movement information. The surface electromyography sensor is used to obtain the user's electromyography signal in real time.

3. A personalized conditioning device for detecting and memorizing human physiological data according to claim 2, characterized in that: The specific steps of sending the received information to the cloud server for processing include: synchronizing the data of different sensors in time and transmitting them to the cloud server through the data processing unit, using Kalman filtering in the cloud server to remove noise in the data, then normalizing the data, identifying and processing the lost data, using the interpolation algorithm to fill the missing values, calculating the 3D coordinates corresponding to each pixel according to the depth value and the camera internal parameters, converting the collected data into point cloud data, using the filtering algorithm to remove isolated points and noise points, downsampling, and then using the feature matching algorithm for preliminary alignment, extracting feature points and descriptors from the point cloud, and then minimizing the distance between corresponding points through an iterative optimization algorithm, using a surface reconstruction algorithm to generate a smooth 3D surface, using computer vision technology to segment the 3D model, mapping the data of the pressure sensor to the 3D model, analyzing the pressure distribution of various parts of the user's body, identifying the pressure concentration area, using the depth camera and inertial sensor data to perform posture estimation, obtaining the posture information of various parts of the user's body, using a deep learning posture estimation network to detect key joints of the user's body, and automatically marking various parts of the user's body according to the joint information and the 3D model.

4. The personalized conditioning device for detecting and memorizing human physiological data according to claim 1, characterized in that: The generation of massage parameters includes the following steps: starting a surface electromyography sensor to obtain electromyographic signals from a user's body, decomposing the electromyographic signals using continuous variational mode decomposition, decomposing the signals into a set of intrinsic mode functions, calculating the correlation coefficient of each intrinsic mode function, and calculating a threshold value according to a dual correlation coefficient standard, dividing all intrinsic mode functions into three categories according to the threshold value: pure intrinsic mode functions, mixed intrinsic mode functions, and noisy intrinsic mode functions, denoising the mixed intrinsic mode functions, and after removing the noisy intrinsic mode functions, extracting multidimensional features, inputting them into a preset random forest regression model, outputting a muscle tension level, and calculating massage parameters corresponding to the muscle tension level according to preset standards.

5. A personalized conditioning device for detecting and memorizing human physiological data according to claim 4, characterized in that: The continuous variation mode decomposition decomposes the electromyographic signal and comprises the following steps: The input signal f(t) is divided into two parts: f(t)=u L (t)+f r (t) In the formula, u L (t) is the Lth order mode, f r (t) is the residual signal, which includes the sum of the first-order mode and the (L-1)-order mode. The unprocessed signal is denoted by f u (t); When each mode is closely around its center frequency, the minimization constraint is as follows: where w L is the center frequency of the Lth mode, * is the convolution operation, is the partial derivative with respect to time t, and δ(t) is the Dirac function.

6. A personalized conditioning device for detecting and memorizing human physiological data according to claim 5, characterized in that: In f r (t) and u L Select the appropriate filter when the spectrum overlap of (t) is minimal. L (t) is the impulse response of the filter to the Lth mode, and its frequency response expression is: In the formula, α is the maximum equilibrium parameter, and the J2 constraint is established as: The first-order mode is processed similarly to the (L-1)-order mode, assuming that β i (t) is the impulse response of the filter to the i-th order mode, and its frequency response is obtained Establish the J3 constraint as: To ensure that the assumptions hold, establish constraints: Build a constrained minimization model: According to the solution idea of ​​the model in the VMD algorithm, the augmented Lagrangian function is constructed to obtain the optimal solution: where λ is the Lagrange multiplier.

7. A personalized conditioning device for detecting and memorizing human physiological data according to claim 6, characterized in that: According to Parseval's theorem, the above formula can be transformed into: The alternating direction multiplier method can be used to iteratively solve the problem: In the formula, represents the Fourier transform of the original signal f(t), n is the number of iterations, and τ is the iteration step. express At the center frequency The Fourier transform result of the L mode at the nth iteration is shown in Figure 2, and the other modes are similar.

8. The personalized conditioning device for detecting and memorizing human physiological data according to claim 7, characterized in that: The calculation process of the dual correlation coefficient is as follows: First, calculate each eigenmode function u k The correlation coefficient between (t) and the original input signal f(t): Where CC(k) represents the correlation coefficient of the kth intrinsic mode function, and N represents the sample length of the signal; The dual threshold screening of the dual correlation coefficient standard, the calculation threshold is as follows: Where P1 and P2 represent the high threshold (Hth) and low threshold (Lth), respectively, CC max and CC aνg Represent the maximum and average values ​​of the correlation coefficients of each mode, respectively.

9. The personalized conditioning device for detecting and memorizing human physiological data according to claim 7, characterized in that: The training process of the preset random forest regression model includes the following steps: the extracted feature data is divided into a training set, a validation set and a test set. The data is divided into 70% for training, 15% for validation and 15% for testing. Each tree is generated based on different random feature subsets and sample subsets, and the error within the node is minimized by selecting the best split point. After all trees grow independently, the random forest regression model averages the prediction results of each tree to obtain the final output value. In order to improve the performance of the model, it is usually necessary to tune the hyperparameters. Through the random search method, some combinations are randomly selected in the hyperparameter space for validation to select the optimal parameters. The validation set is used to evaluate the performance of the model. The evaluation indicators are: mean square error, determination coefficient, and finally the test set is used to evaluate the performance of the model on unseen data; with the use of the conditioning device, the sensor will continuously collect new electromyographic signals and extract features. The muscle tension scores corresponding to these features can be corrected by user feedback or model prediction results, and the old data are weighted to ensure the sensitivity of the model to new trends. Combine new and old data and repeat the training process.

10. The personalized conditioning device for detecting and memorizing human physiological data according to claim 9, characterized in that: The random forest regression model predicts the current muscle tension based on the trained parameters. The muscle tension is output in the form of a quantitative score, with a value in the range of 0-100 representing the current muscle tension level; the data processing unit monitors the electromyographic signal in real time and uploads it to the cloud server, adjusts the massage parameters in real time according to the calculated muscle tension, and adjusts the massage intensity and frequency in real time according to the real-time changing electromyographic signal; the massage module adjusts the massage process in real time according to the dynamic massage parameters, and the user can set the evaluation time interval and massage plan according to the needs through the user interface.

Images