Foot massage control method based on massage shoe wearing device and massage shoe

Abstract

The present application relates to the field of information technology, and relates to a foot massage control method based on a massage shoe wearing device and a massage shoe, and the method comprises the following steps: acquiring initial three-dimensional model data of a foot, and extracting a foot acupoint reference position coordinate from the initial three-dimensional model data; if the real-time offset exceeds a preset threshold, a position correction algorithm is used to adjust the reference position coordinate to obtain an updated acupoint target point; starting from the updated acupoint target point, a foot physiological signal in a massage process is acquired, and the physiological signal is processed to judge an improvement degree of a foot blood circulation state; according to the improvement degree of the foot blood circulation state, a fuzzy logic control method is used to calculate a massage intensity correction value to obtain a personalized massage intensity parameter; the final massage rhythm sequence is used to drive an executing mechanism to apply massage to the foot acupoint, and real-time user physiological response data is acquired to iteratively optimize subsequent adjustment.

Description

Technical Field

[0001] This invention relates to the field of information technology, and more particularly to massage shoes, specifically to a foot massage control method based on a massage shoe wearable device and the massage shoes themselves. Background Technology

[0002] The integration of smart wearable devices with health monitoring technology is becoming an important direction for improving personal health management. This field aims to provide users with convenient and effective daily care experiences by deeply integrating technology with human needs. Especially in foot massage devices, precise acupoint recognition and personalized massage controls are considered key to enhancing product value. Foot massage not only relieves fatigue but also promotes blood circulation, which is of great significance for the recovery of modern people after prolonged sitting or high-intensity exercise.

[0003] However, current foot massage devices on the market exhibit significant shortcomings in practical applications. While many products possess certain intelligent functions, they often overlook the complexity and dynamic changes in the user's foot characteristics, resulting in a substantial reduction in massage effectiveness. In particular, the location of foot acupoints and the adjustment of massage intensity generally lack a deep understanding of the human body's true state, making it difficult to adapt to the unique needs of different users in different scenarios. This problem manifests in the user experience as inaccurate massage positioning or inappropriate pressure, affecting comfort and actual effectiveness.

[0004] A deeper technological challenge lies in the precise location of acupoints on the feet and the real-time capture of the body's responses during massage. Foot acupoints, being tiny and complexly distributed points on the body, can vary slightly in location due to individual foot shape differences, shoe wearing habits, and even foot fatigue levels. Traditional positioning methods often cannot handle this dynamic shift, causing the massage to deviate from the target area. More importantly, even when the positioning is accurate, the massage device struggles to adjust the intensity or rhythm based on the user's immediate foot reactions. For example, changes in blood circulation or muscle tension in the user's feet cannot be effectively perceived; the device simply operates mechanically according to a preset program, lacking interactivity with the body's condition.

[0005] Therefore, how to dynamically capture subtle changes in acupoints on the feet during user wear and adjust the massage strategy based on the real-time physiological responses of the feet during massage has become a key issue in improving the practicality and comfort of smart massage devices. In practical applications, this manifests as follows: when users use massage shoes after a long day at work, the device may fail to accurately locate fatigue-related acupoints or adjust the massage intensity based on improved blood circulation in the feet, leading to an ineffective or uncomfortable massage experience and severely impacting user experience and health benefits. Summary of the Invention

[0006] This invention provides a foot massage control method and a massage shoe based on a wearable massage shoe device, mainly comprising:

[0007] The process involves: acquiring initial 3D model data of the foot; extracting reference coordinates of acupoints from the initial 3D model data; acquiring dynamic deformation signals of the foot during user activity based on the reference coordinates of the acupoints; analyzing the deformation feature patterns in the dynamic deformation signals to determine the real-time offset of the acupoints; if the real-time offset exceeds a preset threshold, adjusting the reference coordinates using a position correction algorithm to obtain updated acupoint target points; acquiring foot physiological signals during the massage process from the updated acupoint target points; processing the physiological signals to determine the degree of improvement in foot blood circulation; calculating massage intensity correction values ​​using a fuzzy logic control method based on the degree of improvement in foot blood circulation to obtain personalized massage intensity parameters; acquiring feedback data from a foot muscle tension sensor for the personalized massage intensity parameters; fine-tuning the intensity parameters through a feedback loop mechanism if the feedback data shows muscle tension higher than a preset threshold to determine the final massage rhythm sequence; and using the final massage rhythm sequence to drive the actuator to apply massage to the acupoints of the foot, while acquiring real-time user physiological response data for iterative optimization of subsequent adjustments. Furthermore, the step of acquiring initial three-dimensional model data of the foot and extracting the reference position coordinates of acupoints on the foot from the initial three-dimensional model data includes: acquiring initial three-dimensional model data of the foot through sensors, obtaining the point cloud distribution of the foot surface from the initial three-dimensional model data; preprocessing the point cloud distribution of the foot surface using image processing technology, removing noise through edge detection filtering to obtain the edge features of the foot contour; extracting acupoint reference points based on the edge features of the foot contour, determining the preliminary position coordinates of the acupoints using geometric center calculation; calibrating the reference offset from the preliminary position coordinates of the acupoints, adjusting the deviation through coordinate transformation, and obtaining the reference position coordinates of the acupoints on the foot. Furthermore, the step of acquiring dynamic deformation signals of the foot during user activity based on the reference coordinates of the foot acupoints, analyzing the deformation feature patterns in the dynamic deformation signals, and determining the real-time offset of the foot acupoints includes: continuously acquiring dynamic deformation signals of the foot during user activity using a signal acquisition device, the signal acquisition device including a flexible strain sensor; performing signal preprocessing on the dynamic deformation signals, the preprocessing including filtering and normalization operations, to obtain normalized time-series signal data; inputting the normalized time-series signal data into a convolutional neural network, the convolutional neural network including a feature extraction layer, the feature extraction layer analyzing deformation feature patterns from the signals; and calculating the real-time offset of each acupoint based on the deformation feature patterns and a pre-established acupoint region division, combined with the reference coordinates of the acupoints, through feature mapping relationships to determine the real-time position of the foot acupoints.Furthermore, if the real-time offset exceeds a preset threshold, a position correction algorithm is used to adjust the reference position coordinates to obtain the updated acupoint target position. This includes: extracting deformation feature patterns from the dynamic deformation signal; determining if the threshold exceeds the extracted deformation feature patterns; if the offset corresponding to the feature patterns exceeds the preset threshold, using a position correction algorithm to process the acupoint reference position coordinates; the position correction algorithm calculates coordinate correction values ​​based on the offset and a preset correction coefficient to obtain the updated acupoint reference coordinates; calculating the real-time position offset vector of each acupoint through feature mapping based on the updated acupoint reference coordinates and grid area division; and determining the updated acupoint target position by combining the real-time position offset vector and the updated acupoint reference coordinates. Furthermore, the step of acquiring foot physiological signals during the massage process from the updated acupoint target points and processing the physiological signals to determine the degree of improvement in foot blood circulation includes: acquiring foot skin temperature and pulse signals during the massage process from the updated acupoint target points; processing the signals using signal fusion technology to obtain a fused data sequence; calculating the temperature change rate and pulse wave conduction time based on the fused data sequence to determine the correlation pattern between temperature and pulse; calculating the difference in blood filling rate in different areas of the foot using the correlation pattern to obtain a blood flow uniformity index; and determining the degree of improvement in foot blood circulation by combining the pulse wave characteristics in the correlation pattern if the uniformity index exceeds a preset threshold. Furthermore, the step of calculating a massage intensity correction value using a fuzzy logic control method based on the degree of improvement in foot blood circulation to obtain personalized massage intensity parameters includes: acquiring foot blood circulation data from a biosensor, judging changes in flow velocity in the data using a preset threshold to obtain an improvement degree value; defining membership functions of input variables using a fuzzy logic control method for the improvement degree value, performing fuzzification processing to determine a fuzzy set; calculating output variables using inference rules through the fuzzy set to obtain a defuzzified intensity correction value; and generating personalized massage intensity parameters based on the intensity correction value, using user vital sign data from the user device and combining the user vital sign data.Furthermore, regarding the personalized massage intensity parameters, the step involves acquiring feedback data from a foot muscle tension sensor. If the feedback data shows that the muscle tension is higher than a preset threshold, the intensity parameters are fine-tuned through a feedback loop mechanism to determine the final massage rhythm sequence. This includes: acquiring feedback data from a foot pressure sensor and processing the pressure values ​​of each area of ​​the foot using foot pressure distribution monitoring data to obtain a comprehensive tension index; if the comprehensive tension index is higher than a preset threshold, the adjustment range is calculated by iteratively comparing the difference between the current index and the threshold using a feedback loop mechanism to obtain the fine-tuned intensity parameters; integrating biosignal data acquired from biosensors to generate a rhythm adjustment sequence using the fine-tuned intensity parameters; and applying a sequence fusion method to merge the rhythm elements in the sequence to determine the final massage rhythm sequence based on the rhythm adjustment sequence. Furthermore, the step of using the final massage rhythm sequence to drive the actuator to apply massage to acupoints on the feet and acquiring real-time user physiological response data to iteratively optimize subsequent adjustments includes: using the final massage rhythm sequence to generate a control command sequence for the corresponding actuator, the control command sequence including force value and time interval, and simultaneously activating the biosensor group to collect heart rate variability signals and surface electromyography signals; extracting the time-domain standard deviation feature of the heart rate variability signal and calculating the root mean square value feature of the surface electromyography signal, normalizing the two feature values ​​and fusing them to obtain a two-dimensional comprehensive physiological state vector; acquiring a pre-established ideal relaxation state vector, calculating the Euclidean distance between the comprehensive physiological state vector and the ideal relaxation state vector, and defining the distance value as the real-time state deviation; if the real-time state deviation is greater than a preset threshold, then using the gradient descent method, iteratively updating the force value with the goal of reducing the real-time state deviation, calculating the adjustment gradient of the force value in the control command sequence to obtain the force adjustment coefficient; and performing a scalar multiplication operation on each force parameter in the final massage rhythm sequence according to the force adjustment coefficient to generate an updated rhythm sequence for the next massage cycle. Furthermore, the pre-established acupoint region division includes: the acupoint region is divided into a preset grid region on the foot surface, the grid region corresponds to the deformation feature pattern, and the feature mapping relationship is obtained by multiplying the deformation feature pattern by the corresponding displacement vector of the grid region to obtain an offset vector. Furthermore, the acquisition of foot blood circulation data from the biosensor includes: the biosensor collects foot skin temperature signals and pulse signals, and processes the signals using signal fusion technology to obtain the foot blood circulation data, the foot blood circulation data including temperature change rate and pulse wave conduction time.

[0008] A massage shoe includes a massage shoe that uses a foot massage control method based on a massage shoe wearable device for foot massage control.

[0009] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects:

[0010] This invention discloses a method for personalized foot massage based on wearable devices. It acquires initial 3D model data of the foot and uses image processing technology to obtain the coordinates of acupoint reference positions. Analyzing the dynamic deformation signals of the foot during user activity, it utilizes a convolutional neural network to analyze deformation characteristics, calculates acupoint offsets in real time, and adjusts coordinates using a position correction algorithm to ensure massage accuracy. Furthermore, this invention integrates foot skin temperature and pulse signals to determine the degree of improvement in blood circulation. A fuzzy logic control method is used to calculate the massage intensity correction value, and parameters are fine-tuned using muscle tension feedback data. Finally, a personalized massage rhythm sequence is generated to drive the actuator to apply appropriate intensity and rhythm. Simultaneously, user physiological response data is collected in real time for iterative optimization and adjustment. This invention, through multi-signal fusion and intelligent algorithms, significantly improves the accuracy and personalization of massage, effectively improving foot blood circulation and muscle condition, providing users with a comfortable and scientific health management experience. Attached Figure Description

[0011] Figure 1 This is a flowchart of the foot massage control method based on a wearable massage shoe device according to the present invention.

[0012] Figure 2 This is a schematic diagram of the framework of step S103 in this invention. Detailed Implementation

[0013] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this specification, and not all embodiments. Based on the embodiments in this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this specification.

[0014] like Figures 1-2 The foot massage control method based on the massage shoe wearable device in this embodiment may specifically include:

[0015] Step S101: Initial three-dimensional model data of the foot is collected by the built-in sensor of the wearable device, and the collected data is preprocessed by image processing technology to obtain the reference position coordinates of the acupoints on the foot.

[0016] Initial 3D model data of the foot is collected using built-in sensors in a wearable device, and the point cloud distribution of the foot surface is obtained from this initial 3D model data. Image processing techniques are used to preprocess the foot surface point cloud distribution, removing noise through edge detection filtering to obtain the foot contour edge features. Acupoint reference points are extracted from these foot contour edge features, and geometric center calculation is used to determine the preliminary coordinates of the acupoint positions. The reference offset is calibrated from these preliminary acupoint position coordinates, and the deviation is adjusted through coordinate transformation to obtain the reference position coordinates of the foot acupoints.

[0017] In one implementation, acquiring initial three-dimensional model data of the foot using sensors built into a wearable device can be achieved through the following process. The wearable device, such as a smart foot massage shoe or a foot scanner, incorporates depth sensors and an inertial measurement unit, which work together to capture geometric information about the foot.

[0018] Specifically, after the user puts on the device, the sensors initiate a scanning mode. The depth sensor generates point cloud data by emitting infrared light and measuring the reflection time, while the inertial measurement unit records changes in foot posture to ensure data consistency under dynamic conditions. This acquisition method is suitable for everyday health monitoring scenarios, such as users scanning their feet at home to obtain initial 3D model data, including the sole surface and contour point set. This method allows the data acquisition process to cover multiple angles of the foot, providing a foundation for complete 3D reconstruction. Furthermore, sensor calibration needs to be considered during the acquisition process.

[0019] In one possible implementation, the device first performs self-calibration, adjusting the focal length and resolution of the depth sensor to accommodate different foot sizes.

[0020] It's important to note that this calibration is based on a pre-defined reference model, such as a standard foot template, and corrects errors by comparing the collected data with the template. This improves the accuracy of the initial 3D model data, providing reliable input for subsequent preprocessing. In practical applications, such as when a user is standing or walking, sensors collect data in real time, forming a point cloud dataset containing tens of thousands of points, representing the surface morphology of the foot. Preprocessing the collected data using image processing techniques is the next crucial step.

[0021] Specifically, preprocessing includes noise removal, point cloud registration, and surface smoothing. First, a Gaussian filter is applied to the acquired point cloud data to remove random noise, ensuring a uniform distribution of data points. Then, point cloud registration is performed, aligning multiple frames of data using an iterative nearest-point algorithm to form a unified foot model.

[0022] In one embodiment, preprocessing also involves mesh reconstruction, using triangulation methods to convert the point cloud into a surface mesh model. This preprocessing technique is common in the field of foot health, such as in traditional Chinese medicine massage devices, used to prepare the data foundation for acupoint identification. Through these steps, the raw data is transformed into a structured 3D model, reducing the computational burden of subsequent processing.

[0023] Preferably, feature enhancement is introduced during the preprocessing stage.

[0024] For example, edge detection algorithms can be applied to highlight the anatomical features of the foot, such as the arch and phalangeal contours.

[0025] Specifically, edge detection, based on gradient calculation, identifies high-contrast regions in the model, thereby enhancing the visibility of potential acupoint locations. This enhancement method is applicable to various scenarios, such as static scanning or dynamic gait analysis, ensuring the robustness of the preprocessing results. The reference coordinates of foot acupoint locations are obtained based on the preprocessed 3D model.

[0026] Specifically, a foot acupoint database is first established, containing location descriptions of standard Traditional Chinese Medicine acupoints, such as the Yongquan acupoint located at the anterior third of the sole of the foot. A matching algorithm is then used to compare the preprocessed model with the database to calculate the three-dimensional coordinates of the acupoints.

[0027] In one embodiment, a keypoint detection method, such as a machine learning-based feature extractor, is used to identify keypoints (landmarks) in the model and then map them to the acupoint coordinate system.

[0028] It should be noted that the coordinate calculation process involves the application of a transformation matrix to convert local coordinates into a global reference, for example, defining the coordinate axes with the heel as the origin. This method ensures the accuracy of acupoint locations and can be used to guide massage functions in wearable devices. In another implementation, the acquisition of acupoint reference coordinates can be combined with personalized user data.

[0029] For example, when a user inputs their foot size or previous scan records, the system adjusts the database parameters to achieve adaptive positioning.

[0030] Specifically, the algorithm first extracts the curvature features of the preprocessed model, such as calculating the curvature values ​​of various points on the surface, and then matches them with an acupoint feature library to generate a coordinate list. This personalized adjustment is particularly useful in foot care scenarios, such as providing customized acupoint maps for users of different ages, enhancing the versatility of the technology. Through this implementation, coordinate errors are controlled at the millimeter level, achieving reliable benchmark positioning. Further, the integration of preprocessing and coordinate acquisition can form a closed-loop process.

[0031] In one possible implementation, the preprocessed output is directly input into a coordinate calculation module, which uses a convolutional neural network to analyze the model texture and predict the probability distribution of acupoints. The final coordinates are then determined based on a probability threshold. This integrated approach manifests as real-time processing in wearable devices; for example, an acupoint map can be displayed immediately after a user scans their feet for self-massage guidance.

[0032] Understandably, the technical effect of the above process lies in improving the accuracy and efficiency of foot acupoint location. In practical applications, such as in wearable TCM health devices, this method can support multiple functions, such as acupoint stimulation or virtual reality display, without the need for external equipment assistance. In a preferred embodiment, for users with specific foot morphologies, such as flat feet, the sensor angle is adjusted during the acquisition process to capture more details of the sole, and a deformation correction step is added during preprocessing to ensure model accuracy. During coordinate acquisition, the algorithm incorporates morphological compensation to calculate the adjusted acupoint positions. This variation demonstrates the flexibility of the technical solution within the same field, such as adapting to the health monitoring needs of users with different foot types. Finally, through these implementation methods, the technical features of the claims are fully supported, including the real-time nature of data acquisition and the accuracy of preprocessing, as well as the benchmark definition of acupoint coordinates, ensuring the practicality and scalability of the solution in the field of foot health.

[0033] Step S102: Based on the reference coordinates of the acupoints on the foot, obtain the dynamic deformation signal of the foot during the user's activity, and analyze the deformation features in the signal through a convolutional neural network to determine the real-time offset of the acupoints on the foot.

[0034] The system acquires the reference coordinates of acupoints on the foot, which are pre-determined through static measurements. A signal acquisition device, including a flexible strain sensor, continuously acquires dynamic deformation signals of the foot during user activity. These dynamic deformation signals reflect changes in the foot's surface morphology. The dynamic deformation signals undergo preprocessing, including filtering and normalization, to obtain regularized time-series signal data. This regularized time-series signal data is then input into a convolutional neural network (CNN), which includes a feature extraction layer. This feature extraction layer analyzes deformation feature patterns from the signal. Based on these deformation feature patterns and a pre-defined acupoint region division (divided into pre-defined grid regions on the foot surface), and combined with the reference coordinates of the acupoints, the system calculates the real-time offset of each acupoint using a feature mapping relationship. This feature mapping relationship is obtained by multiplying the deformation feature pattern by the corresponding displacement vector of the grid region to obtain the offset vector, thus determining the real-time position of the acupoint.

[0035] In one implementation, the coordinates of the reference positions of acupoints on the foot can be obtained using three-dimensional foot scanning technology.

[0036] Specifically, an optical 3D scanner is used to perform a static scan of the user's feet to obtain 3D point cloud data that includes the surface contour of the foot and key anatomical landmarks.

[0037] For example, the approximate locations of key acupoints such as Yongquan (KI1) and Taichong (LR3) can be pre-marked on the skin surface of the foot. After scanning, these marked points are identified using point cloud processing software, and their three-dimensional coordinates are used as the reference coordinates for the acupoint.

[0038] It should be noted that the reference coordinates are determined when the user is standing still and the foot is not subjected to significant external force, providing a reference origin for subsequent calculations of dynamic offset. Furthermore, the dynamic deformation signals of the foot during user activity can be acquired using a flexible sensor array integrated into the insole or sock.

[0039] In one possible implementation, the sensor array consists of multiple distributed flexible strain sensors and pressure sensors. These sensors are arranged in a grid to monitor changes in tensile and compressive strain and pressure distribution in different areas of the sole of the foot during activities such as walking and running.

[0040] Specifically, the sensor array continuously acquires signals at a specific sampling frequency (e.g., 100 Hz), converts the analog signals into digital signals, and outputs them to form a dynamic deformation signal sequence that changes over time and reflects the multidimensional changes in foot morphology. Based on the acquired dynamic deformation signals, a convolutional neural network is used to analyze them to extract deformation features.

[0041] In one embodiment, the convolutional neural network is specifically designed to process data that combines one-dimensional temporal signals with two-dimensional spatial distributions.

[0042] Specifically, the input dynamic deformation signal can be constructed as a multi-channel two-dimensional matrix, where the rows represent the time series and the columns represent the signal values ​​at different spatial locations (sensor nodes), with each sensor type (such as strain or pressure) serving as an independent channel. The network can contain multiple convolutional layers, pooling layers, and fully connected layers.

[0043] For example, the first set of convolutional layers uses one-dimensional convolutional kernels that slide along the time dimension to capture the dynamic patterns of signal evolution over time; subsequent convolutional layers may use two-dimensional convolutional kernels, considering both the relationship between time and the sensor's spatial location, thereby extracting deep deformation features that contain both temporal dynamics and spatial distribution patterns. During the training phase, the network uses a large amount of labeled data containing real acupoint offsets synchronously measured by a high-precision motion capture system under various activity modes. The network optimizes its internal parameters by minimizing the error (such as mean squared error) between the predicted and actual offsets, ultimately learning the ability to directly map acupoint offsets from complex raw deformation signals.

[0044] Preferably, the convolutional neural network can include residual connection structures to alleviate the vanishing gradient problem during deep network training and improve the stability of feature extraction. The final output layer of the network can be a fully connected layer, with the number of neurons matching the number of acupoints to be tracked multiplied by the coordinate dimensions (e.g., three-dimensional coordinates x, y, z). The activation function of this layer can be a linear function, directly outputting the predicted offset of each acupoint on each coordinate axis. In another implementation, to improve sensitivity to weak deformation signals, preprocessing can be performed before the signal is input into the network.

[0045] For example, the original sensor signals can be normalized to eliminate the influence of individual differences between different sensors; or bandpass filtering can be performed to retain the main components related to human gait frequency and filter out high-frequency noise. The process of determining the real-time offset of acupoints on the foot is to decode the feature vector output by the convolutional neural network into specific spatial displacement values.

[0046] Specifically, the value output by the network represents the offset relative to the reference coordinates.

[0047] For example, for the Yongquan acupoint, the network might output a three-dimensional vector (Δx, Δy, Δz), where Δx represents the offset in the anterior-posterior direction of the foot, Δy represents the offset in the lateral-lateral direction of the foot, and Δz represents the offset in the vertical direction. These offsets are calculated continuously in real time, reflecting the instantaneous changes in the acupoint's location during the movement.

[0048] Understandably, the above technical solutions can be applied to various scenarios that require precise understanding of the biomechanical state of the foot and changes in the location of acupoints.

[0049] For example, in the design and effectiveness evaluation of customized orthotic insoles, real-time monitoring of the displacement of key foot support points (corresponding to specific acupoints) during walking allows for more precise assessment of the insole's support effect and dynamic adjustments. Similarly, in the fields of traditional Chinese medicine rehabilitation or sports science, analyzing the dynamic trajectories of specific acupoints under different exercise modes such as Tai Chi and running can provide quantitative evidence for movement standardization assessment or rehabilitation training guidance. In one specific embodiment, a flexible sensor array is embedded in a pair of specially designed sports socks. After wearing them, the user engages in walking activities for a period of time. The system synchronously records sensor signals and processes them in real time through a pre-trained convolutional neural network. Finally, the system visualizes the trajectory of several major acupoints on the foot moving continuously with the gait cycle in the form of a dynamic 3D model on a display device. This implementation intuitively demonstrates the technical solution's ability to analyze and reconstruct the dynamics of the invisible internal soft tissues of the foot through measurable signals.

[0050] Step S103: If the real-time offset of the acupoint on the foot exceeds the preset threshold, the position correction algorithm is used to adjust the reference position coordinates to obtain the updated target acupoint position.

[0051] Dynamic deformation signals are acquired through foot deformation monitoring, and deformation feature patterns are extracted from these signals. A threshold is determined for each extracted deformation feature pattern. If the offset corresponding to the feature pattern exceeds a preset threshold, a position correction algorithm is used to process the acupoint reference coordinates. The position correction algorithm calculates a coordinate correction value based on the offset and a preset correction coefficient, resulting in updated acupoint reference coordinates. Based on the updated acupoint reference coordinates and the grid area division, a real-time position offset vector for each acupoint is calculated using a feature mapping relationship. Combining the real-time position offset vector with the updated acupoint reference coordinates, the updated target acupoint location is determined.

[0052] In one implementation, the preset threshold can be determined based on the physiological displacement range of foot acupoints in a specific activity mode.

[0053] Specifically, statistical analysis can be performed by collecting a large amount of data on foot acupoint deviations from standard user movements (such as normal walking).

[0054] For example, the vertical displacement range of the Yongquan acupoint during a complete gait cycle might be statistically estimated to be between 0 and 15 millimeters. Therefore, a preset threshold could be set as the upper limit of this statistical range, such as 15 millimeters.

[0055] It should be noted that the threshold can be set separately for different acupoints and different coordinate directions, forming a set of threshold vectors. When the component of the real-time offset output by the convolutional neural network in any direction exceeds its corresponding threshold, the position correction process is triggered. Furthermore, the purpose of the position correction algorithm is to progressively adjust the original reference coordinates when it is determined that they may no longer be accurate due to long-term adaptive changes in the foot or slight sensor drift.

[0056] In one embodiment, the algorithm may employ an exponentially weighted moving average strategy.

[0057] Specifically, the algorithm maintains a "corrected reference coordinate" for each acupoint. When the real-time offset exceeds a threshold, the algorithm does not immediately replace the original reference with the current predicted position. Instead, it performs a weighted fusion of the acupoint's absolute position calculated at the current moment (i.e., the original reference coordinate plus the real-time offset) with the corrected reference coordinate from the previous moment.

[0058] For example, the new corrected reference coordinates = α * (original reference coordinates + current real-time offset) + (1 - α) * previous corrected reference coordinates. Here, α is a smoothing factor between 0 and 1, for example, 0.1. This algorithm can smoothly track the slow changes in the acupoint reference position, avoiding drastic fluctuations in reference coordinates caused by single prediction errors or instantaneous abnormal movements. In another possible implementation, the position correction algorithm can incorporate the statistical characteristics of historical offsets.

[0059] For example, the system continuously records the predicted offset sequence of each acupoint within a time window (e.g., the last 100 steps). When the current offset exceeds a threshold, the algorithm checks the historical sequence. If it finds that recent offsets are consistently biased to the same side and the magnitude is large, it may indicate a systematic change in foot condition or sensor characteristics. In this case, the algorithm can calculate the median or robust mean of the offsets within the historical window and use it as a correction factor, superimposing it onto the original reference coordinates at a certain ratio to obtain the updated acupoint target location. This method can resist occasional outlier interference, making the correction of the reference coordinates more robust.

[0060] Understandably, the updated acupoint target locations will serve as the new reference origin for calculating real-time offsets in the next time period. By periodically or trigger-basedly executing the aforementioned threshold judgment and correction process, the system enables the acupoint tracking reference frame to adapt to long-term changes in the user's foot condition, maintaining the accuracy of the monitoring system.

[0061] For example, in scenarios used for long-term gait rehabilitation monitoring, as patients recover foot muscle strength and mobility, their arch shape and force patterns may change. This solution can automatically adjust the acupoint baseline to ensure that the offset in subsequent analysis always reflects the dynamic changes relative to the current normal foot condition, rather than a fixed initial state.

[0062] Preferably, the threshold itself can also be designed to be adaptively adjustable.

[0063] For example, the system can continuously monitor the distribution of offsets over a period of time. If it finds that the overall distribution center of the offsets has drifted significantly, it will adjust the baseline value of the threshold accordingly so that it always matches the current typical activity range.

[0064] Step S104: Starting from the updated acupoint target points, obtain the foot skin temperature and pulse signals during the massage process, process these signals through signal fusion technology, and determine the degree of improvement in foot blood circulation.

[0065] Starting from the updated acupoint target locations, the skin temperature and pulse signals of the foot during the massage process are acquired. These signals are processed using signal fusion technology to obtain a fused data sequence. Based on the fused data sequence, the temperature change rate and pulse wave propagation time are calculated to determine the correlation pattern between temperature and pulse. Using this correlation pattern, the difference in blood filling rate in different areas of the foot is calculated to obtain a blood flow uniformity index. If the uniformity index exceeds a preset threshold, the degree of improvement in foot blood circulation is determined by combining the pulse wave characteristics in the correlation pattern.

[0066] In one implementation, starting from the updated acupoint target points, the accuracy of acupoint location must first be ensured. These acupoint target points may be updated based on previous scans or user input, for example, by using infrared imaging equipment to identify and adjust the position of specific acupoints on the sole of the foot, such as Yongquan or Taichong. After updating, these points serve as a starting reference for massage, guiding the massage device or manual operation to focus on these areas, thereby monitoring physiological changes in real time during the massage.

[0067] Specifically, obtaining foot skin temperature and pulse signals during a massage can be achieved through integrated sensors.

[0068] For example, an infrared temperature sensor is attached to the skin surface of the foot to continuously record temperature data. For instance, the temperature might be 32 degrees Celsius at the start of a massage, potentially rising to 34 degrees Celsius as the massage progresses, indicating increased local blood flow. Simultaneously, pulse signals are acquired using a photoplethysmography (PPG) sensor placed at the dorsalis pedis artery to capture pulse waveform changes, such as differences in peak and trough values, reflecting heart rate and vasodilation. The acquisition frequency of these signals can be set to 10 times per second to ensure real-time data transmission. Further, processing these signals using signal fusion technology is a core step. Signal fusion involves synchronously integrating temperature and pulse data, for example, using a weighted averaging algorithm. First, the temperature signal is filtered to remove noise, and then aligned with the time series of the pulse signal.

[0069] It should be noted that the fusion process can be achieved by calculating composite indices.

[0070] For example, the rate of temperature change and pulse wave amplitude can be vector-superimposed to form a fusion vector. The specific process includes data normalization, mapping temperature values ​​to a range of 0 to 1, and normalizing the pulse amplitude before multiplying the results to obtain a fusion score. A higher score indicates stronger signal synergy, reflecting improved blood circulation.

[0071] In one embodiment, in a foot massager device, this fusion can be performed by an embedded processor with a processing time controlled within 1 second to support real-time feedback.

[0072] Preferably, the degree of improvement in foot blood circulation is determined based on the analysis of the fused signals.

[0073] For example, a threshold is set such that a fusion score above 0.7 is considered a significant improvement, while a score below 0.3 indicates no significant change. This judgment can be made by comparing data before and after massage. For instance, if the fusion score is 0.4 before massage and rises to 0.8 after massage, it indicates that the circulation status has improved by approximately 100%.

[0074] In one possible implementation, this judgment is applied to a home foot bath scenario, where users can view the results through an app and receive guidance on massage duration.

[0075] Understandably, this technology can also be applied in TCM foot massage centers. For example, massage therapists can use handheld devices to collect signals, fuse them, and then determine the improvement in the client's circulation to help adjust the massage intensity. This versatility ensures the solution is suitable for different user groups, such as foot care for the elderly or athletes. In another embodiment, signal fusion can incorporate timing analysis to further improve accuracy.

[0076] For example, temperature and pulse signals are input into a sliding window model with a window size of 30 seconds. The average fusion value within the window is calculated to dynamically determine the improvement trend. This method provides more stable evaluation results, especially avoiding signal fluctuation interference during prolonged massages. Furthermore, to enhance flexibility, the acquisition device can be modularly designed, such as having a temperature module independent of the pulse module for easy replacement. The fusion technology supports multiple algorithm variants, such as replacing it with Kalman filtering to handle noisy environments, thus maintaining effectiveness in different sub-scenarios of foot massage.

[0077] For example, in an office foot relaxation scenario, users wear portable sensors that integrate signals during the massage to determine circulation improvement and help relieve fatigue.

[0078] Specifically, if the improvement level reaches a moderate level after integration, the system can suggest extending the massage time to 15 minutes.

[0079] In one embodiment, the judgment result can be output as a quantitative indicator, such as the percentage of improvement, and presented on a display screen to ensure that users can intuitively understand the technical effect.

[0080] Step S105: Based on the degree of improvement in foot blood circulation, a fuzzy logic control method is used to calculate the massage intensity correction value to obtain personalized massage intensity parameters.

[0081] Foot blood circulation data is acquired from biosensors, and changes in flow velocity within the data are judged using a preset threshold to obtain an improvement level value. For this improvement level value, a membership function for the input variables is defined using a fuzzy logic control method, and fuzzification processing is performed to determine a fuzzy set. Using this fuzzy set, inference rules are applied to calculate the output variable, obtaining a defuzzified intensity correction value. Based on this intensity correction value, user vital sign data is acquired from the user device and combined with this data to generate personalized massage intensity parameters.

[0082] In one implementation, the degree of improvement is assessed by monitoring the blood circulation in the feet, which forms the basis for personalized massage.

[0083] Specifically, the system first collects physiological data from the user's feet, such as using infrared sensors to detect changes in blood flow velocity and temperature. This data is used to calculate the degree of improvement in blood circulation, for example, by comparing current blood flow indicators with the initial state to obtain a percentage improvement value. This process ensures that massage intensity adjustments are based on actual physiological feedback, avoiding discomfort caused by fixed parameters. In this way, the system can dynamically respond to changes in the user's body, providing a more personalized massage experience. Furthermore, a fuzzy logic control method is used to calculate the massage intensity correction value. Fuzzy logic is a control method that handles uncertainty and fuzzy information; it simulates human decision-making processes by defining fuzzy sets and membership functions.

[0084] For example, the degree of improvement in blood circulation can be used as an input variable, divided into three fuzzy sets: low, medium, and high. Each set corresponds to a membership function used to quantify the membership degree of the input value. Then, inference is performed based on a pre-defined fuzzy rule base, such as "if the degree of improvement is low, the correction value is increased." This method allows for the handling of continuous physiological changes, rather than binary judgments, thus obtaining a more accurate correction output.

[0085] Preferably, a defuzzification step is introduced during the calculation process to convert the fuzzy output into a crisp value. Specifically, the centroid method is used to calculate the precise value of the correction, for example, by integrating the output distribution of fuzzy rules and calculating a weighted average. The key to this step lies in the design of the rule base, which needs to be optimized based on clinical data, such as adjusting the rule weights for the blood circulation characteristics of users of different ages. Through this implementation, fuzzy logic control can effectively cope with the variability in foot massage, ensuring that the correction value reflects the true degree of improvement.

[0086] One possible implementation considers the application of various foot massage scenarios.

[0087] For example, in a home foot bath massager, the system monitors changes in blood circulation after the user soaks in real time. If the improvement is moderate, the massage intensity is adjusted to 1.2 times the initial value. This scenario emphasizes the integration of portable devices, with sensors embedded in the massage pad and a computing module executing fuzzy logic algorithms via an embedded processor. This diversity covers everyday relaxation scenarios, demonstrating the versatility of the technology.

[0088] It should be noted that the inputs to fuzzy logic can also include auxiliary variables, such as user feedback on comfort ratings.

[0089] Specifically, by using the degree of improvement and comfort as multiple inputs, a two-dimensional fuzzy rule is constructed, such as "if the improvement is high and the comfort is low, then slightly reduce the correction value." This extension enhances the accuracy of personalized parameters and avoids bias caused by a single variable. In practice, this method adapts to different users' physical conditions by iteratively optimizing the rule base.

[0090] For example, in rehabilitation massage equipment, for users with chronic foot circulation problems, the system first assesses the degree of baseline improvement, then applies fuzzy control to calculate an initial correction value, which is dynamically updated in subsequent massage cycles. This embodiment emphasizes long-term use scenarios, and the calculation process involves real-time adjustment of rules, such as modifying the boundaries of member functions based on continuous monitoring data to maintain the stability of massage intensity. Furthermore, after obtaining personalized massage intensity parameters, the system outputs control signals to the massage actuator.

[0091] For example, correction values ​​can be applied to air pressure or vibration modules to adjust to a user-specific intensity level, such as adjusting from a standard intensity level 5 to level 7. This output ensures a closed-loop implementation of the technical solution, providing practical benefits in the field of health management, such as improving user foot comfort and circulatory health.

[0092] In one embodiment, fuzzy logic parameters are optimized by incorporating user historical data.

[0093] Specifically, the improvement levels and correction values ​​from previous massage sessions are stored to form a dataset, which is then used for machine learning adjustments to the rule base, such as refining the shape of membership functions through simple gradient methods. This approach extends the adaptability of the technology and makes it suitable for reusable scenarios.

[0094] Understandably, the technical effect of this method lies in improving the targeting of massage, using fuzzy logic to handle physiological uncertainties, and achieving more effective promotion of blood circulation, without relying on precise mathematical models.

[0095] For example, in a portable foot massager, when implementing the above process, low-power computing is prioritized to ensure efficient operation of fuzzy inference on the microcontroller. This scenario demonstrates the application potential of the technology in mobile health devices.

[0096] Step S106: For personalized massage intensity parameters, obtain feedback data from the foot muscle tension sensor. If the feedback data shows that the muscle tension is higher than the preset threshold, then fine-tune the intensity parameters through the feedback loop mechanism to determine the final massage rhythm sequence.

[0097] Foot pressure sensors acquire feedback data from foot muscle tension sensors, and foot pressure distribution monitoring data is combined to process pressure values ​​from different areas of the foot to obtain a comprehensive tension index. If the comprehensive tension index exceeds a preset threshold, a feedback loop mechanism is used to iteratively compare the current index with the threshold to calculate the adjustment range, resulting in a fine-tuned force parameter. This fine-tuned force parameter is then used to integrate biosignal data acquired from biosensors to generate a rhythm adjustment sequence. Based on this rhythm adjustment sequence, a sequence fusion method is applied to merge the rhythm elements in the sequence to determine the final massage rhythm sequence.

[0098] In one implementation, feedback data obtained through a foot muscle tension sensor is used to adjust personalized massage intensity parameters, which is key to ensuring that the massage process adapts to the user's physiological state.

[0099] Specifically, sensors are embedded in the massage device to monitor changes in foot muscle tension in real time, such as using piezoelectric sensors to detect pressure signals during muscle contraction. This data is acquired digitally and converted into quantifiable tension values, such as pressure units per square centimeter, providing a basis for subsequent judgments. This process emphasizes the real-time nature and accuracy of the data to avoid adjustment deviations caused by delays. Furthermore, determining whether the muscle tension shown in the feedback data exceeds a preset threshold is the starting point for the adjustment mechanism. The preset threshold is set based on the user's initial physiological assessment, for example, 1.5 times the standard tension for middle-aged users.

[0100] Specifically, the collected tension value is compared with a threshold; if it exceeds the threshold, a fine-tuning process is triggered. This judgment is executed by an embedded processor, dynamically updating the threshold based on the user's historical data to ensure adaptation to different physical conditions.

[0101] Preferably, a feedback loop mechanism is used to fine-tune the force parameters, and the mechanism initiates an iterative process when the muscle tension is higher than a threshold.

[0102] Specifically, the feedback loop involves multiple data acquisitions and parameter adjustments. For example, the initial force parameter is set to a moderate level. If the tension remains high, the force value is gradually reduced, with each adjustment being 10% of the original value. This loop is achieved through closed-loop control, where the sensor continuously provides feedback on the adjusted tension changes until the tension drops below a threshold. This mechanism handles physiological variability, provides stable adjustments, and avoids excessive one-time changes.

[0103] In one possible implementation, the loop integration algorithm module has a maximum number of iterations of 5 to prevent infinite iteration.

[0104] It's important to note that the core of the feedback loop mechanism lies in the detailed implementation of its iterative fine-tuning process. First, the mechanism starts with initial force parameters, collecting sensor data to calculate the current tension. Then, it generates an error signal by comparison; if the error is positive, the force is reduced; if negative, it is maintained or slightly increased. This process simulates human feedback response; for example, in a foot massager, it is executed every 30 seconds, gradually optimizing the parameters to a suitable range. This detailed process ensures the accuracy of fine-tuning and is suitable for long-term massage scenarios.

[0105] For example, in home foot massage devices, when the above mechanism is applied, if the user's foot muscle tension exceeds a threshold, the feedback loop fine-tunes the intensity from level 5 to level 3, ultimately determining the rhythm sequence as slow vibration combined with intermittent pressure. This scenario demonstrates the versatility of the mechanism in daily relaxation, with sensor data driving adjustments to enhance the user experience.

[0106] In one embodiment, considering a rehabilitation center scenario, the feedback loop is extended to multi-sensor fusion, such as incorporating a temperature sensor to assist in determining stress levels.

[0107] Specifically, if tension is high and temperature rises, the intensity of the massage is reduced preferentially to avoid overheating and discomfort. This implementation emphasizes the flexibility of the mechanism, enhancing the accuracy of fine-tuning through data fusion. Furthermore, the final massage rhythm sequence is generated based on the fine-tuned intensity parameters.

[0108] Specifically, adjustment parameters are mapped to rhythm modes; for example, a low-intensity sequence corresponds to a long, gentle massage, while a high-intensity sequence corresponds to a short, strong massage. This sequence is output to the actuator via control signals to achieve a closed-loop application.

[0109] Understandably, the implementation of this method in a portable massage pad emphasizes low-power design. The feedback loop runs on a microprocessor, and the sequence, once determined, is stored as a user profile for easy reuse. This approach covers mobile scenarios, demonstrating the adaptability of the technology.

[0110] For example, in professional foot massage equipment, if feedback indicates that tension remains above a threshold, a loop mechanism introduces manual user input as an aid, and the fine-tuned sequence includes a gradual rhythm for progressive relaxation. This embodiment extends to interactive applications, ensuring personalized parameters. In one implementation, the entire process is integrated into a software module; sensor data is processed and looped, and the output sequence is applied to different foot areas, such as a targeted rhythm for the ankle. This modular design facilitates expansion and is suitable for various device types.

[0111] Step S107: The final massage rhythm sequence is used to drive the wearable device actuator to apply corresponding force and rhythm to the acupoints on the feet, and real-time user physiological response data is obtained to iteratively optimize subsequent adjustments.

[0112] Using the final massage rhythm sequence, a corresponding control command sequence for the actuator is generated. This control command sequence includes force values ​​and time intervals. Simultaneously, a biosensor array worn on the user's ankle and instep is activated to collect heart rate variability and surface electromyography (EMG) signals. The time-domain standard deviation feature is extracted from the heart rate variability signal, and the root mean square (RMS) feature is calculated from the EMG signal. These two feature values ​​are normalized and then fused to obtain a two-dimensional comprehensive physiological state vector. A pre-established ideal relaxation state vector is obtained, and the Euclidean distance between the comprehensive physiological state vector and the ideal relaxation state vector is calculated. This distance is defined as the real-time state deviation. If the real-time state deviation exceeds a preset threshold, gradient descent is used to iteratively update the force values ​​to reduce the real-time state deviation. The adjustment gradient for the force values ​​in the control command sequence is calculated to obtain a force adjustment coefficient. Based on the force adjustment coefficient, a scalar multiplication operation is performed on each force parameter in the final massage rhythm sequence to generate an updated rhythm sequence for the next massage cycle.

[0113] In one embodiment, the wearable device includes an actuator, such as a vibrating motor or a pneumatic pump, for applying massage to acupoints on the feet. The device first generates a final massage rhythm sequence based on preset parameters; for example, the sequence is defined as a combination of a series of intensity values ​​and time intervals, with the intensity gradually increasing from a gentle 5 Newtons to a moderate 15 Newtons, and the rhythm presented as pulses at 60 times per minute. This sequence is designed based on Traditional Chinese Medicine acupoint theory, targeting key points such as the Yongquan and Taichong acupoints to ensure stimulation promotes blood circulation. The device converts the sequence into electrical signals via a built-in controller, driving the actuator to precisely execute the physical massage process on the feet. This approach is suitable for everyday foot care scenarios, such as home use or office relaxation, ensuring the continuity and effectiveness of the massage. Furthermore, the process of applying the corresponding intensity and rhythm involves an acupoint positioning module.

[0114] For example, the wearable device is equipped with a sensor array that can automatically identify the foot contour and locate acupoints. For instance, it scans the sole of the foot using an infrared sensor, calculates the acupoint coordinates, and then adjusts the output of the actuator. Force control uses a proportional-integral-derivative algorithm to adjust the motor speed, ensuring the force remains stable within a specified range, while the rhythm is achieved through intermittent vibrations via a timer. This approach is applicable to various foot massage scenarios, such as a slow rhythm for fatigue recovery or a rapid pulse for pain relief, demonstrating the flexibility of the technical solution.

[0115] It should be noted that the process of applying acupoints originates from traditional Chinese medicine practice, but the combination with modern sensor technology has improved accuracy.

[0116] Specifically, obtaining real-time user physiological response data is accomplished through biosensors integrated into wearable devices. These sensors monitor metrics such as heart rate, skin conductivity, and body temperature changes. For example, the heart rate sensor uses photoplethysmography (PPG) to acquire data once per second, while skin conductivity measures sweat gland activity via electrodes to assess relaxation levels. Data acquisition occurs in real-time during the massage, and the device converts the raw signals into digital form and stores them in local memory. This method ensures data immediacy and supports subsequent analysis.

[0117] In one possible implementation, if user feedback indicates a decrease in heart rate, suggesting the massage is effective, the system records this response for optimization purposes. This physiological response data focuses on health monitoring, helping to quantify the massage's effectiveness.

[0118] Preferably, the mechanism for iteratively optimizing subsequent adjustments is based on a feedback loop algorithm.

[0119] In one embodiment, the system first analyzes the collected physiological response data, such as calculating heart rate variability. If the variability increases beyond a threshold, such as 0.1, the current rhythm sequence needs adjustment. The optimization process involves machine learning models, such as using gradient descent to iteratively update sequence parameters. Specific steps include initial sequence input, data feedback input to the model, error calculation, and adjustment of intensity and rhythm intervals, ultimately outputting an optimized sequence to drive the device. This iteration can be repeated multiple times until the physiological response reaches a stable state, such as a heart rate maintained at 60 to 80 beats per minute. The innovation of this process lies in its real-time adaptation to the user's state, enhancing the personalization of massage, and its application in foot care devices can bring more precise health benefits. In another embodiment, massage programs are adjusted for different user groups.

[0120] For example, for middle-aged users, the initial rhythm sequence emphasizes gentle intensity to avoid overstimulation. The system uses physiological data, such as a 0.5-degree Celsius increase in body temperature, as an optimization signal, iteratively increasing the rhythm frequency to 80 times per minute. This scenario is suitable for home health management, and the device can connect to a mobile app to display the optimization progress. Furthermore, if data indicates a decrease in skin conductivity, the system automatically reduces the intensity to ensure safety. This diverse implementation demonstrates the versatility of the technological solution within the same foot massage field.

[0121] It is understandable that the principle behind the above optimization process lies in a closed-loop control system, where physiological response serves as an input variable driving parameter adjustment.

[0122] For example, response data is quantified into vector form. The system compares the current vector with the target relaxation state, minimizing the difference through multiple iterations to maximize the massage effect. This explanation clarifies the detailed process of iterative operations, the logical chain from data input to parameter output, ensuring the continuity of adjustments. In actual operations, this mechanism can effectively handle physiological variations in users, such as high heart rates under stress, bringing a relaxation effect through gradual optimization.

[0123] For example, in one specific embodiment, the wearable device is applied in a foot massage center setting. After the user puts it on, the system initiates a default sequence, applying a massage of 10 Newtons of force at 50 beats per minute to the Yongquan acupoint, while sensors collect heart rate data in real time. If the heart rate drops from 90 beats per minute to 70 beats per minute, the system iteratively adjusts the rhythm to 40 beats per minute and the force to 8 Newtons. This adjustment, based on data analysis, avoids the limitations of static massage and demonstrates the adaptability of the technology. Furthermore, the driving details of the device's actuators include power management and signal amplification.

[0124] For example, the controller uses pulse width modulation (PWM) technology to generate drive signals, ensuring that actuators such as pneumatic pumps operate with low power consumption. This implementation supports extended massage sessions and is applicable in various wellness scenarios, such as portable use while traveling.

[0125] In one embodiment, the processing of physiological response data involves a filtering step to remove noise.

[0126] For example, the system uses a low-pass filter to process the heart rate signal to ensure accuracy before inputting it into the optimization module. This preprocessing enhances the reliability of the iteration and provides stable feedback in foot massage applications. Finally, through the above methods, the technical solution achieves precise stimulation and dynamic optimization of foot acupoints, demonstrating practical value in the field of health devices.

[0127] It should be noted that the above examples are merely some specific embodiments of the present invention. Obviously, the present invention is not limited to the above embodiments and many variations are possible. All variations that can be directly derived or conceived by those skilled in the art from the content disclosed in this invention should be considered within the scope of protection of this invention.

Claims

1. A foot massage control method based on a wearable massage shoe device, characterized in that, include: Acquire initial three-dimensional model data of the foot, and extract the reference position coordinates of acupoints on the foot from the initial three-dimensional model data; Based on the reference coordinates of the foot acupoints, dynamic deformation signals of the feet during user activities are obtained, deformation feature patterns in the dynamic deformation signals are analyzed, and the real-time offset of the foot acupoints is determined. If the real-time offset exceeds a preset threshold, a position correction algorithm is used to adjust the reference coordinates to obtain updated acupoint target points. Starting from the updated acupoint target points, physiological signals of the feet during the massage process are obtained, and the physiological signals are processed to determine the degree of improvement in foot blood circulation. Based on the degree of improvement in foot blood circulation, a fuzzy logic control method is used to calculate the massage intensity correction value to obtain personalized massage intensity parameters. For these personalized massage intensity parameters, feedback data from a foot muscle tension sensor is acquired. If the feedback data shows that muscle tension is higher than a preset threshold, the intensity parameters are fine-tuned through a feedback loop mechanism to determine the final massage rhythm sequence. This final massage rhythm sequence is then used to drive the actuator to apply massage to acupoints on the foot, and real-time user physiological response data is acquired to iteratively optimize subsequent adjustments.

2. The foot massage control method based on a wearable massage shoe device as described in claim 1, characterized in that, The process of acquiring initial three-dimensional model data of the foot and extracting reference coordinates of acupoints from the initial three-dimensional model data includes: acquiring initial three-dimensional model data of the foot through sensors; obtaining the point cloud distribution of the foot surface from the initial three-dimensional model data; preprocessing the point cloud distribution of the foot surface using image processing technology, removing noise through edge detection filtering to obtain the edge features of the foot contour; extracting acupoint reference points based on the edge features of the foot contour, determining the preliminary position coordinates of the acupoints using geometric center calculation; calibrating the reference offset from the preliminary position coordinates of the acupoints, adjusting the deviation through coordinate transformation, and obtaining the reference position coordinates of the acupoints on the foot.

3. The foot massage control method based on a wearable massage shoe device as described in claim 1, characterized in that, The step of acquiring dynamic deformation signals of the foot during user activity based on the reference coordinates of the foot acupoints, analyzing the deformation feature patterns in the dynamic deformation signals, and determining the real-time offset of the foot acupoints includes: continuously acquiring dynamic deformation signals of the foot during user activity using a signal acquisition device, the signal acquisition device including a flexible strain sensor; performing signal preprocessing on the dynamic deformation signals, the preprocessing including filtering and normalization operations, to obtain normalized time-series signal data; inputting the normalized time-series signal data into a convolutional neural network, the convolutional neural network including a feature extraction layer, the feature extraction layer analyzing deformation feature patterns from the signals; and calculating the real-time offset of each acupoint based on the deformation feature patterns and a pre-established acupoint region division, combined with the reference coordinates of the acupoints, through feature mapping relationships to determine the real-time position of the foot acupoints.

4. The foot massage control method based on a wearable massage shoe device as described in claim 1, characterized in that, If the real-time offset exceeds a preset threshold, a position correction algorithm is used to adjust the reference position coordinates to obtain the updated acupoint target position. This includes: extracting deformation feature patterns from the dynamic deformation signal; determining if a threshold exceeds the extracted deformation feature patterns; if the offset corresponding to the feature pattern exceeds the preset threshold, using a position correction algorithm to process the acupoint reference position coordinates; the position correction algorithm calculates coordinate correction values ​​based on the offset and a preset correction coefficient to obtain the updated acupoint reference coordinates; calculating the real-time position offset vector of each acupoint through feature mapping based on the updated acupoint reference coordinates and grid area division; and determining the updated acupoint target position by combining the real-time position offset vector and the updated acupoint reference coordinates.

5. The foot massage control method based on a wearable massage shoe device as described in claim 1, characterized in that, The step of acquiring foot physiological signals during the massage process from the updated acupoint target points and processing the physiological signals to determine the degree of improvement in foot blood circulation includes: acquiring foot skin temperature and pulse signals during the massage process from the updated acupoint target points; processing the signals using signal fusion technology to obtain a fused data sequence; calculating the temperature change rate and pulse wave conduction time based on the fused data sequence to determine the correlation pattern between temperature and pulse; calculating the difference in blood filling rate in different areas of the foot using the correlation pattern to obtain a blood flow uniformity index; and determining the degree of improvement in foot blood circulation by combining the pulse wave characteristics in the correlation pattern if the uniformity index exceeds a preset threshold.

6. The foot massage control method based on a wearable massage shoe device as described in claim 1, characterized in that, The step of calculating a massage intensity correction value using a fuzzy logic control method based on the degree of improvement in foot blood circulation to obtain personalized massage intensity parameters includes: acquiring foot blood circulation data from a biosensor, wherein the biosensor collects foot skin temperature signals and pulse signals, and processes the signals using signal fusion technology to obtain the foot blood circulation data, which includes temperature change rate and pulse wave conduction time; judging the change in flow velocity in the data using a preset threshold to obtain an improvement degree value; defining a membership function of the input variable for the improvement degree value using a fuzzy logic control method, performing fuzzification processing to determine a fuzzy set; calculating the output variable using inference rules through the fuzzy set to obtain a defuzzified intensity correction value; and generating personalized massage intensity parameters based on the intensity correction value by acquiring user vital sign data from the user device and combining the user vital sign data.

7. The foot massage control method based on a wearable massage shoe device as described in claim 1, characterized in that, The process involves acquiring foot muscle tension sensor feedback data for the personalized massage intensity parameters. If the feedback data shows that the muscle tension is higher than a preset threshold, the intensity parameters are fine-tuned through a feedback loop mechanism to determine the final massage rhythm sequence. This includes: acquiring foot muscle tension sensor feedback data from a foot pressure sensor and processing the pressure values ​​of each area of ​​the foot using foot pressure distribution monitoring data to obtain a comprehensive tension index; if the comprehensive tension index is higher than a preset threshold, a feedback loop mechanism is used to iteratively compare the difference between the current index and the threshold to calculate the adjustment range and obtain the fine-tuned intensity parameters; integrating biosignal data acquired from biosensors using the fine-tuned intensity parameters to generate a rhythm adjustment sequence; and applying a sequence fusion method to merge the rhythm elements in the sequence to determine the final massage rhythm sequence.

8. The foot massage control method based on a wearable massage shoe device as described in claim 1, characterized in that, The process of using the final massage rhythm sequence to drive the actuator to apply massage to acupoints on the feet and acquiring real-time user physiological response data for iterative optimization of subsequent adjustments includes: using the final massage rhythm sequence to generate a control command sequence for the corresponding actuator, the control command sequence including force value and time interval, and simultaneously activating a biosensor group to collect heart rate variability signals and surface electromyography signals; extracting the time-domain standard deviation feature of the heart rate variability signal and calculating the root mean square value feature of the surface electromyography signal, normalizing the two feature values ​​and fusing them to obtain a two-dimensional comprehensive physiological state vector; acquiring a pre-established ideal relaxation state vector, calculating the Euclidean distance between the comprehensive physiological state vector and the ideal relaxation state vector, and defining this distance value as the real-time state deviation; if the real-time state deviation is greater than a preset threshold, using the gradient descent method, iteratively updating the force value with the goal of reducing the real-time state deviation, calculating the adjustment gradient of the force value in the control command sequence to obtain the force adjustment coefficient; and performing a scalar multiplication operation on each force parameter in the final massage rhythm sequence according to the force adjustment coefficient to generate an updated rhythm sequence for the next massage cycle.

9. The foot massage control method based on a wearable massage shoe device as described in claim 3, characterized in that, The pre-established acupoint region division includes: the acupoint region is divided into a preset grid region on the foot surface, the grid region has a corresponding relationship with the deformation feature pattern, and the feature mapping relationship is obtained by multiplying the deformation feature pattern with the corresponding displacement vector of the grid region to obtain the offset vector.

10. A massage shoe, characterized in that, The invention includes massage shoes, wherein the massage shoes employ the foot massage control method based on a massage shoe wearable device according to any one of claims 1-9 for foot massage control.

Images