Waistband self-adaptive control system and method

Abstract

The invention provides a self-adaptive waistband control system and method, and relates to the technical field of intelligent wearable equipment control, and the system comprises an obtaining module which is used for obtaining an acceleration signal and a gyroscope signal of the waist of a human body; the extraction module is used for performing feature extraction on the acceleration signal and the gyroscope signal to obtain a plurality of signal features; the first construction module is used for constructing a human body posture recognition model based on a random forest algorithm; the recognition module is used for inputting each signal feature into a human body posture recognition model based on a random forest algorithm for posture recognition, and determining a current human body posture; the determining module is used for determining a control mode corresponding to the current human body posture according to the current human body posture; and the control module is used for carrying out self-adaptive control on the light intensity of the infrared lamp beads in the waistband according to the determined control mode through the PI lambda D mu controller. The control mode and the light intensity can be adjusted in real time according to the actual situation, and the comfort is high.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent wearable device control, and in particular to a waistband adaptive control system and method. Background Art

[0002] With the rapid development of smart wearable devices and health monitoring technology, more and more smart devices are being used in daily life to help users manage their health. As one of the common wearable devices, belts are widely used in the fields of medical treatment, sports and life. By integrating sensors and control systems, they can monitor the posture, movement status and other physiological data of the human body in real time. These smart belts can effectively provide health feedback, help users adjust their posture, and avoid health problems caused by long-term bad posture, such as low back pain, spinal problems, etc.

[0003] In the prior art, many smart belt products use fixed control modes to assist users. These control modes fail to take into account the differences in individual needs in different postures and usually require users to manually adjust the control mode, which brings inconvenience and additional operational burden.

[0004] In addition, existing control systems lack flexibility in adapting to changes in human posture and cannot adjust control modes and lighting intensity in real time according to actual conditions, resulting in poor comfort and unable to effectively improve the energy efficiency of the equipment. Summary of the invention

[0005] In order to solve the technical problems that traditional smart belt products use fixed control modes to assist users, these control modes fail to take into account the differences in individual needs in different postures, and usually require users to manually adjust the control mode, which brings inconvenience and additional operating burden, and also lacks flexibility in adapting to changes in human posture, and cannot adjust the control mode and light intensity in real time according to actual conditions, resulting in poor comfort, and cannot effectively improve the energy efficiency of the device, the present invention provides a belt adaptive control system and method.

[0006] The technical solution provided by the embodiment of the present invention is as follows:

[0007] First aspect:

[0008] An embodiment of the present invention provides a waist belt adaptive control system, comprising:

[0009] An acquisition module is used to acquire acceleration signals and gyroscope signals of the human waist;

[0010] An extraction module, used for performing feature extraction on the acceleration signal and the gyroscope signal to obtain a plurality of signal features;

[0011] The first building module is used to build a human posture recognition model based on the random forest algorithm;

[0012] A recognition module, used for inputting each of the signal features into the human posture recognition model based on the random forest algorithm to perform posture recognition and determine the current human posture;

[0013] A determination module, used to determine a control mode corresponding to the current human body posture according to the current human body posture;

[0014] Control module for PI λ D μ The controller adaptively controls the light intensity of the infrared lamp beads in the belt according to the determined control mode.

[0015] Second aspect:

[0016] An embodiment of the present invention provides a waist belt adaptive control method, comprising:

[0017] S1: Obtain the acceleration signal and gyroscope signal of the human waist;

[0018] S2: extracting features from the acceleration signal and the gyroscope signal to obtain a plurality of signal features;

[0019] S3: Build a human posture recognition model based on random forest algorithm;

[0020] S4: inputting each of the signal features into the human posture recognition model based on the random forest algorithm to perform posture recognition and determine the current human posture;

[0021] S5: determining a control mode corresponding to the current human body posture according to the current human body posture;

[0022] S6: Via PI λ D μ The controller adaptively controls the light intensity of the infrared lamp beads in the belt according to the determined control mode.

[0023] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0024] In the present invention, by extracting features from the acceleration signal and gyroscope signal of the human waist, a plurality of signal features are obtained, each of which is input into a human posture recognition model based on a random forest algorithm for posture recognition, and the current human posture is determined. According to the current human posture, a control mode corresponding to the current human posture is determined, and a fixed control mode is no longer used to assist the user. The differences in individual needs under different postures can be taken into account, and the control mode corresponding to the current human posture is determined through PI. λ D μThe controller adaptively controls the light intensity of the infrared lamp beads in the belt according to the determined control mode. The user does not need to manually adjust the control mode, which provides convenience and reduces the operating burden. It has strong flexibility in adapting to changes in human posture and can adjust the control mode and light intensity in real time according to actual conditions. It is highly comfortable and can effectively improve the energy efficiency of the equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0026] Figure 1 A schematic structural diagram of a waist belt adaptive control system provided by an embodiment of the present invention;

[0027] Figure 2 A schematic structural diagram of a waist belt provided by an embodiment of the present invention;

[0028] Figure 3 A schematic diagram of the structure of a waistband hand controller provided by an embodiment of the present invention;

[0029] Figure 4 A schematic flow chart of a waist belt adaptive control method provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0030] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0031] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "example" in the present invention should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of the word "example" is intended to present the concept in a specific way. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or it can be either of the two.

[0032] In order to make the technical problems, technical solutions and advantages to be solved by the present invention more clear, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0033] Reference Manual Attached Figure 1 , shows a schematic structural diagram of a waist belt adaptive control system provided by an embodiment of the present invention.

[0034] Reference Manual Attached Figure 2, showing a schematic structural diagram of a waist belt provided by an embodiment of the present invention.

[0035] The embodiment of the present invention provides a waist belt adaptive control system 20, comprising:

[0036] The acquisition module 201 is used to acquire the acceleration signal and the gyroscope signal of the human waist.

[0037] In a possible implementation, the acquisition module 201 is specifically used for:

[0038] The three-axis acceleration signals of the X-axis, Y-axis and Z-axis of the human waist are obtained respectively through the three-axis accelerometer.

[0039] It should be noted that a three-axis accelerometer is a sensor used to measure the acceleration of an object along three orthogonal directions in space (X-axis, Y-axis, and Z-axis). It is able to detect changes in linear motion, such as movement, stillness, or vibration of an object. By detecting acceleration signals, a three-axis accelerometer can also sense the tilt angle of an object, because the projection of gravity acceleration on different axes will also change with the tilt of the object. Therefore, it is widely used in smartphones, wearable devices, and robots for positioning, posture detection, and motion tracking.

[0040] The three-axis gyroscope is used to obtain the three-axis gyroscope signals of the X-axis, Y-axis and Z-axis of the human waist respectively.

[0041] It should be noted that a three-axis gyroscope is a sensor used to measure the angular velocity of an object around three orthogonal axes (X-axis, Y-axis, and Z-axis). By detecting angular velocity signals, it can sense changes in the rotation or tilt of an object. Three-axis gyroscopes are commonly used for attitude stabilization control and rotational motion detection, such as attitude control of drones, automatic screen rotation of mobile phones, and head tracking in virtual reality devices. Unlike accelerometers, gyroscopes mainly measure rotational motion rather than linear displacement.

[0042] In the present invention, the accelerometer provides information on linear motion and gravity direction, which can help determine the changes in the linear acceleration of the human body. The gyroscope provides angular velocity information, which can accurately identify the changes in the rotation or tilt angle of the human body. Combining the two, the data on waist posture and movement status can be obtained more comprehensively and accurately, thereby improving the accuracy of posture recognition.

[0043] The extraction module 202 is used to extract features from the acceleration signal and the gyroscope signal to obtain multiple signal features.

[0044] In the present invention, if the original signals of the accelerometer and gyroscope are used directly for posture recognition, it may be affected by noise, sampling frequency and environmental interference, resulting in inaccurate recognition results. Through feature extraction, the original signals can be converted into more representative and informative features. These features can better reflect the core information of human posture, thereby improving the accuracy and robustness of the recognition model.

[0045] In a possible implementation, the extraction module 202 is specifically configured to:

[0046] Preprocess the acceleration signal and gyroscope signal.

[0047] Specifically, the spike noise and high-frequency noise in the three-axis acceleration signal (Acc XYZ) and the three-axis gyroscope signal (Gyro XYZ) are eliminated in turn by a median filter and a 20Hz third-order Butterworth low-pass filter. Subsequently, the three-axis acceleration signal is separated into a low-frequency gravity component (three-axis gravity acceleration signal, GravityAcc XYZ) and a high-frequency motion component (three-axis motion acceleration signal, BodyAcc XYZ) by a 0.3Hz Butterworth low-pass filter. The three-axis acceleration rate of change (BodyAcc Jerk XYZ) is obtained by calculating the time derivative of the motion component. By calculating the Euclidean distance of the above-mentioned multiple three-axis signals, the gravity acceleration amplitude (GravityAcc Mag), the motion acceleration amplitude (BodyAcc Mag) and the acceleration rate of change amplitude (BodyAcc Jerk Mag) are extracted. The three-axis angular velocity signal (BodyAngular Speed ​​XYZ) is directly obtained by using the de-noised three-axis gyroscope signal, and the three-axis angular acceleration signal (BodyAngularAcc XYZ) is extracted by calculating the time derivative of the three-axis angular velocity signal. The amplitude signals (BodyAngular Speed ​​Mag and BodyAngularAccMag) are extracted by calculating the Euclidean distance of the three-axis angular velocity and the three-axis angular acceleration. Finally, the three-axis motion acceleration, three-axis acceleration change rate, motion amplitude and other signals are converted from the time domain to the frequency domain through the fast Fourier transform (FFT), and finally 17 time domain and frequency domain signals are formed.

[0048] It should be noted that the median filter is a denoising method used to reduce noise in the signal, especially spike noise (such as isolated outliers). It sorts the data in a part of the signal (i.e., the filter window) by size and replaces the original signal value with the median value to remove the interference of extreme values ​​on the signal. The median filter is particularly suitable for removing random noise while retaining the edge characteristics of the signal (such as mutation points).

[0049] It should be noted that the 20Hz third-order Butterworth low-pass filter is a signal processing filter that allows signals below 20Hz to pass through while attenuating frequency components above 20Hz. It is a type of Butterworth filter, characterized by a flat frequency response curve within the passband, no ripples, and the ability to smoothly attenuate high-frequency parts. "Third-order" refers to the order of the filter, which has a stronger filtering effect and a faster attenuation rate, making it suitable for processing signals containing high-frequency noise.

[0050] It should be noted that the 0.3Hz Butterworth low-pass filter is a low-pass filter that is specifically used to retain signals with frequencies below 0.3Hz while attenuating components above 0.3Hz. This filter is often used to separate very low-frequency signal components from mixed signals, such as separating the gravity component (static signal) from the acceleration signal to analyze the motion trend.

[0051] Among them, the time domain signals (10 types) include: three-axis motion acceleration signal, three-axis gravity acceleration signal, motion acceleration amplitude, gravity acceleration amplitude, three-axis acceleration change rate signal, acceleration change rate amplitude, three-axis angular velocity signal, angular velocity amplitude, three-axis angular acceleration signal and angular acceleration amplitude.

[0052] Among them, the frequency domain signals (7 types) include: three-axis motion acceleration signal, three-axis acceleration change rate signal, motion acceleration amplitude, acceleration change rate amplitude, three-axis angular velocity signal, angular velocity amplitude and frequency domain representation of motion acceleration amplitude.

[0053] Feature extraction is performed on the preprocessed acceleration signal and gyroscope signal to obtain multiple signal features.

[0054] Optionally, the signal characteristics specifically include: arithmetic mean, standard deviation, median absolute deviation, maximum value, minimum value, average of the sum of squares, signal amplitude area, signal entropy, interquartile range, 4th-order Burg autoregression coefficient, Pearson correlation coefficient, the angle between the signal mean and the vector, the range between the minimum and maximum values, root mean square value, frequency signal skewness, frequency signal kurtosis, maximum frequency component, frequency signal weighted average and spectral energy of the frequency interval.

[0055] In the present invention, the peak noise and high-frequency noise are removed by a median filter and a 20Hz third-order Butterworth low-pass filter, and the gravity and motion components are separated by a 0.3Hz low-pass filter to ensure the purity and stability of the input signal, providing high-quality data for subsequent feature extraction and posture recognition. The generated time domain and frequency domain features cover the static and dynamic characteristics of human motion, providing a multi-dimensional, high-information input for the posture recognition model. Feature extraction compresses the data dimension, filters irrelevant or redundant information, reduces the computational overhead of the subsequent machine learning model, and improves real-time performance and processing efficiency.

[0056] The first building module 203 is used to build a human posture recognition model based on the random forest algorithm.

[0057] It should be noted that the random forest algorithm is an integrated machine learning algorithm that improves the accuracy and robustness of classification or regression tasks by combining the prediction results of multiple decision trees (usually dozens or hundreds of trees). It introduces randomness when constructing each decision tree, such as randomly sampling different subsets from the training data (i.e., out-of-bag sampling) and randomly selecting some features each time the tree splits the node, thereby reducing the risk of overfitting. The final prediction result of the random forest combines the outputs of each decision tree by voting (classification problems) or averaging (regression problems). Its advantages include high accuracy, robustness to noise and outliers, and the ability to handle high-dimensional data and complex relationships between features.

[0058] In the present invention, the random forest algorithm combines the prediction results of multiple decision trees and significantly improves the accuracy of posture recognition through a majority voting mechanism, and can more reliably distinguish different postures or dynamic behaviors.

[0059] The recognition module 204 is used to input each signal feature into a human posture recognition model based on a random forest algorithm to perform posture recognition and determine the current human posture.

[0060] In the present invention, the input signal features cover multiple dimensions in the time domain and frequency domain, including statistical features (such as mean, standard deviation), frequency features (such as frequency kurtosis, weighted average) and signal amplitude, etc. These features fully characterize the details of human posture and movement, and enhance the model's ability to distinguish different postures. Random forest can effectively handle nonlinear relationships and high-dimensional feature spaces, and its integrated learning mechanism can reduce the prediction error of a single decision tree, thereby improving the overall accuracy of the model.

[0061] In a possible implementation, the identification module 204 is specifically configured to:

[0062] Construct an initial forest consisting of multiple decision trees, each of which includes multiple nodes.

[0063] Calculate the information gain and Gini index of the current node of each decision tree:

[0064]

[0065] Where Gain() represents information gain, M represents the number of signal features in the current node, a represents the attribute of the signal feature, Ent() represents the entropy of the signal feature set, V represents the number of child nodes generated after node division, || represents taking the absolute value, and M v represents the number of signal features in the vth child node after node division, y represents the number of categories, and p k It represents the probability that the signal feature belongs to the kth category, and Gini() represents the Gini index.

[0066] Combined with the information gain and Gini index of the current node, the node division rule is dynamically determined:

[0067]

[0068] Among them, O represents the node division rule, min represents the minimum value, W{} represents the comprehensive function, α represents the weight coefficient of the Gini index, and θ represents the weight coefficient of the information gain.

[0069] According to the node division rules, each signal feature is used to divide the current node into nodes.

[0070] When the number of signal features in the current node is less than a preset number, node division is stopped to obtain multiple posture categories output by multiple decision trees.

[0071] According to each posture category, a voting mechanism is used to determine the posture category with the most votes as the current human posture.

[0072] In the present invention, by calculating the information gain and Gini index at the same time and dynamically adjusting the node partitioning rules, the model can more reasonably select split features and split points, and improve the ability to distinguish posture categories. Random forest determines the final category through the voting mechanism of multiple decision trees. This integrated learning method effectively reduces the error of a single decision tree, thereby significantly improving classification accuracy. The node partitioning rule comprehensively considers information gain and Gini index, and flexibly adapts to data distribution and feature importance by dynamically adjusting weight coefficients (α and θ). The partitioning rule that minimizes the error is determined by the comprehensive function W, ensuring that each step of splitting can maximize the classification performance.

[0073] The determination module 205 is used to determine the control mode corresponding to the current human body posture according to the current human body posture.

[0074] It should be noted that the control mode is divided into 5 levels, which are used to adjust the light intensity of the infrared lamp beads in the belt.

[0075] Optionally, Mode 1 is suitable for static state (such as standing or sitting for a long time), providing the lowest intensity of light to maintain basic comfort, and the default working time is 20 minutes; Mode 2 is suitable for light activity (such as slow walking or small adjustments in chair posture), providing low-intensity light to relieve waist pressure, and the default working time is 25 minutes; Mode 3 is suitable for medium activity (such as fast walking or climbing stairs), providing medium-intensity light to support waist comfort, and the default working time is 30 minutes; Mode 4 is suitable for high-intensity activity (such as jogging, carrying objects or squatting and other strenuous exercises), providing high-intensity light to help relieve fatigue, and the default working time is 15 minutes; Mode 5 is suitable for fatigue recovery (such as static or lying after exercise), providing the highest intensity of light to quickly relax waist muscles and promote recovery, and the default working time is 10 minutes. To ensure device safety and user experience, the maximum working time of all modes is limited to 60 minutes.

[0076] Reference Manual Attached Figure 3 , showing a schematic structural diagram of a waist belt hand controller provided by an embodiment of the present invention.

[0077] It should be noted that users can either rely on the system to automatically adjust the control mode according to the posture, or manually adjust the belt hand controller to set the light intensity and working time. The belt hand controller retains the adjustment function of the original button. Users can select the light intensity (5 levels are optional) and working time (optional range is 10 minutes to 60 minutes) according to personal needs through manual buttons. In addition, the hand controller provides the function of long pressing the power button for 3 seconds to turn the machine on and off. After the device is turned on, the default setting is 20 minutes of working time and 1 level of light intensity. The manual adjustment function is independent of the system's intelligent posture recognition mode. Users can use the automatic mode to achieve intelligent control, or choose the manual mode for more personalized and flexible operation.

[0078] In the present invention, by detecting the user's current posture and matching the corresponding control mode, the infrared light intensity can be adjusted in a personalized manner to meet the needs of different users in different states. Users do not need to manually switch gears frequently. The system can adjust the light intensity in real time according to the human body posture, reducing the user's operating burden and improving the intelligence of the device. In a posture that does not require high-intensity infrared light (such as a relatively static state), the system can automatically switch to a low-gear mode to reduce power consumption and increase the battery life of the device.

[0079] The control module 206 is used to control the λ D μThe controller adaptively controls the light intensity of the infrared lamp beads in the belt according to the determined control mode.

[0080] It should be noted that PI λ D μ The controller is a control algorithm that expands the traditional PID controller. It combines proportional (P), integral (I) and differential (D) control, and introduces the concepts of fractional integral (λ) and fractional differential (μ). Compared with the classic PID control, PI λ D μ The controller uses fractional-order operations in the integral and differential parts, which enables the controller to adjust the system response and accuracy more flexibly, especially when dealing with complex dynamic systems. By adjusting the order of fractional integrals and differentials, PI λ D μ The controller can better cope with the nonlinear characteristics of the system and improve the control accuracy, stability and adaptability of the system.

[0081] In the present invention, by PI λ D μ The controller performs adaptive control of infrared light intensity, which can achieve high-precision and flexible control, improve the energy efficiency and comfort of the equipment, and ensure effective health intervention. Adaptive adjustment can not only enhance the stability and accuracy of the equipment, but also provide users with a personalized and intelligent experience, ultimately improving user satisfaction and long-term dependence.

[0082] In a possible implementation, the control module 206 is specifically configured to:

[0083] Get the target light intensity in the current control mode.

[0084] Get the actual light intensity in the current control mode.

[0085] By PI λ D μ The controller outputs a control signal to adaptively control the light intensity of the infrared lamp beads in the belt.

[0086] Optionally, according to the following formula, through PI λ D μ The controller outputs control signals to adaptively control the light intensity of the infrared lamp beads in the belt:

[0087] u(h)=K p e(h)+K i D -λ e(h)+K d D μ e(h)

[0088] e(h)=E(h)-E 0

[0089] Among them, u(h) represents the control signal at time h, K p represents the proportional gain coefficient, e(h) represents the deviation between the target light intensity and the actual light intensity at time h, and K i Indicates the integral gain coefficient, K d Denotes the differential gain coefficient, D -λ represents fractional-order integration operation, λ represents the integration order, D μ represents fractional differential operation, μ represents the differential order, E(h) represents the actual light intensity of the infrared lamp beads in the belt at time h, and E 0 Indicates the target light intensity of the infrared lamp beads in the belt.

[0090] In the present invention, by PI λ D μ The controller realizes adaptive control of infrared light intensity, which not only provides high-precision and flexible control, but also effectively improves energy efficiency, extends device life, enhances comfort, and improves system stability and robustness. The controller's dynamic adjustment and fractional-order processing capabilities enable the present invention to achieve accurate, comfortable and efficient health intervention in different usage scenarios, optimize user experience and improve overall device performance.

[0091] In a possible implementation, the waist belt adaptive control system further includes:

[0092] The second constructing module 207 is used to construct an objective function with the goal of improving the comfort index during the control mode switching process.

[0093] Among them, the objective function is specifically:

[0094] maxf(δ)=w P ·C P +w H ·C H

[0095] Among them, max means taking the maximum value, f() means the objective function, and δ means PI λ D μ The control parameter set of the controller, PI λ D μ The control parameter set of the controller includes proportional gain coefficient, integral gain coefficient, differential gain coefficient, integral order and differential order, w P Represents the weight coefficient of the pressure comfort score, C P represents the pressure comfort score, w H Represents the weight coefficient of the thermal comfort score, CH Represents the thermal comfort score.

[0096] Optionally, the calculation method of the comfort index is as follows:

[0097] By evenly setting multiple pressure sensors in the waist belt, waist pressure data can be obtained:

[0098]

[0099] Among them, P i represents the waist pressure data obtained by the i-th pressure sensor, F i It represents the pressure exerted by the human body on the i-th pressure sensor, i=1,2,…n, n represents the total number of pressure sensors, and ΔS represents the sensing area of ​​the pressure sensor.

[0100] According to the waist pressure data obtained by each pressure sensor, the maximum waist pressure data is determined:

[0101] P max =max(P 1 ,P 2 ,...,P n )

[0102] Among them, P max Indicates the maximum waist pressure data, max means taking the maximum value, P 1 Indicates the waist pressure data obtained by the first pressure sensor, P 2 represents the waist pressure data obtained by the second pressure sensor, P n Represents the waist pressure data obtained by the nth pressure sensor.

[0103] According to the waist pressure data obtained by each pressure sensor, the average waist pressure data is determined:

[0104]

[0105] Among them, P avg Indicates average waist pressure data.

[0106] The pressure comfort score is determined based on the waist pressure data, maximum waist pressure data, and average waist pressure data obtained by each pressure sensor:

[0107]

[0108] Among them, C p Indicates the pressure comfort score.

[0109] Waist temperature data is obtained by evenly setting multiple temperature sensors in the waist belt.

[0110] According to the waist temperature data obtained by each temperature sensor, the maximum waist temperature data is determined:

[0111] H max =max(H 1 ,H 2 ,...,H m )

[0112] Among them, H max Indicates the maximum waist temperature data, H 1 Indicates the waist temperature data obtained by the first temperature sensor, H 2 Indicates the waist temperature data obtained by the second temperature sensor, H m Represents the waist temperature data obtained by the mth temperature sensor.

[0113] According to the waist temperature data obtained by each temperature sensor, the average waist temperature data is determined:

[0114]

[0115] Among them, H avg Indicates the average waist temperature data, H j represents the waist temperature data obtained by the jth temperature sensor, j=1, 2, ...m, and m represents the total number of temperature sensors.

[0116] The temperature comfort score is determined based on the waist temperature data, maximum waist temperature data, and average waist temperature data obtained by each temperature sensor:

[0117]

[0118] Among them, C H Represents the thermal comfort score.

[0119] The comfort index is calculated based on the pressure comfort score and the temperature comfort score:

[0120] ρ=w P ·C P +w H ·C H

[0121] Among them, ρ represents the comfort index, w P Represents the weight coefficient of pressure comfort score, w H Represents the weight coefficient of the thermal comfort score.

[0122] In the present invention, the pressure and temperature comfort scores of the waist are calculated through the data collected by the pressure sensor and the temperature sensor, which helps to dynamically adjust the control strategy to reduce the discomfort caused during wearing. Through real-time monitoring of pressure and temperature and calculation of the comfort index, the control system can reduce the light intensity and working time when it is not necessary, save battery energy, and extend the service life of the device. Through precise comfort adjustment, user discomfort caused by excessive pressure or high temperature is avoided, the overall user experience is improved, and the device is more in line with the requirements of long-term wear.

[0123] The optimization module 208 is used to optimize the PI according to the objective function through the krill swarm optimization algorithm. λ D μ Optimize the controller.

[0124] It should be noted that the Particle Swarm Optimization (PSO) algorithm is an optimization algorithm based on swarm intelligence, inspired by the foraging behavior of krill swarms. The algorithm simulates a group of particles flying randomly in the search space. The particles adjust their flight direction based on their own experience and the experience of the group to find the optimal solution to the problem. Each particle represents a potential solution, and the position and speed of the particle in the search space are updated iteratively until the optimal solution is found. PSO is widely used in fields such as function optimization and machine learning.

[0125] In a possible implementation, the optimization module 208 is specifically configured to:

[0126] Initialization parameters, set the maximum number of iterations of the krill swarm optimization algorithm.

[0127] The initial population is generated through Logistic chaos mapping. The initial population includes multiple krill individuals, each of which represents a feasible PI λ D μ The controller's control parameters are:

[0128] L z (t+1) = γL z (t)·(1-L z (t))

[0129] Among them, L z (t+1) represents the position of the zth krill individual at the t+1 iteration, γ represents the branching coefficient with a value of 3.95, and L z (t) represents the position of the zth krill individual at the tth iteration.

[0130] It should be noted that Logistic Chaos Mapping is a process that generates complex behaviors based on simple mathematical models and is often used to study chaotic phenomena. It describes an iterative process in which the value of a variable is updated based on its own current value and a control parameter. As the number of iterations increases, the process exhibits complex and unpredictable behaviors, called "chaos". This chaotic behavior is highly sensitive, that is, a small change in the initial conditions may lead to completely different results. Logistic Chaos Mapping has applications in many fields, such as random number generation, complex system modeling, and studying nonlinear dynamic systems in nature.

[0131] The objective function is used as the fitness function to calculate the fitness value of each krill individual, and the krill individual with the maximum fitness value is used as the current krill individual.

[0132] The inertia weight of foraging motion and induced motion is adaptively adjusted through the cosine control factor:

[0133]

[0134] Among them, ω 1 represents the inertia weight of the foraging motion after adaptive adjustment, ω 2 represents the inertia weight of the induced motion after adaptive adjustment, ω max represents the maximum value of the inertia weight, ω min represents the minimum value of the inertia weight, T represents the maximum number of iterations, and P represents a random number in the range of (0,1).

[0135] It should be noted that the cosine control factor is a factor used to adjust the search strategy or control process, usually adjusted by the cosine function. In the optimization algorithm, the cosine control factor can balance the relationship between exploration (global search) and exploitation (local search) by changing the cosine function. In some evolutionary computing and optimization algorithms, the cosine control factor helps gradually converge to the optimal solution while avoiding falling into the local optimum, thereby improving the ability of global search.

[0136] According to the inertia weights of the adaptively adjusted foraging motion and induced motion, the foraging motion and induced motion are adaptively adjusted:

[0137]

[0138] Among them, F' z represents the foraging movement of the zth krill individual after adaptive adjustment, F z represents the foraging movement of the zth krill individual, S f represents the foraging speed, β z represents the foraging direction of the z-th krill individual, represents the last foraging movement of the z-th krill individual, represents the current best foraging direction to attract the z-th krill individual, represents the historical best foraging direction to attract the z-th krill individual, G' z represents the induced motion of the zth krill individual after adaptive adjustment, G z represents the induced motion of the zth krill individual, G max represents the maximum induced velocity, b z represents the induction factor of the zth krill individual, represents the last induced movement of the z-th krill individual.

[0139] Update the current position of the krill individual according to the adaptively adjusted foraging movement and induced movement:

[0140] L z (t+Δt)=L z (t)+Δt(F' z +G' z +U z )

[0141] U z =U max ξ

[0142] Among them, L z (t+Δt) represents the position of the z-th krill individual at the t+Δt-th iteration, Δt represents the step length when the krill individual is updated, and U z represents the random diffusion rate of the zth krill individual, U max represents the maximum diffusion rate, ξ represents the random disturbance factor with a value range of [-1,1].

[0143] Through the Cauchy mutation operation, the position of the current krill individual is updated:

[0144]

[0145] in, represents the position of krill individuals after Cauchy mutation, represents the position of the z-th krill individual after Cauchy mutation at the t-th iteration, and C(0,1) represents a Cauchy distribution random variable.

[0146] It should be noted that the Cauchy mutation operation is a mutation operation based on the Cauchy distribution, which is often used in evolutionary algorithms and genetic algorithms to generate new candidate solutions. The Cauchy distribution has a "heavy tail" characteristic, that is, it has a thicker tail than the normal distribution and can jump over a larger range, which makes the Cauchy mutation operation have a stronger exploration ability in the search space. By applying the Cauchy mutation, the algorithm can more effectively jump out of the local optimum and promote the global search process, thereby enhancing the diversity and convergence of the algorithm.

[0147] When the maximum number of iterations is reached, the algorithm stops and outputs PI λ D μ The optimal control parameter set of the controller.

[0148] In the present invention, PI is optimized by using the krill swarm optimization algorithm (PSO) combined with Logistic chaos mapping, cosine control factor and Cauchy mutation operation. λ D μ The control parameters of the controller. The PSO algorithm simulates the foraging behavior of krill groups and can efficiently perform global searches to avoid local optimal solutions. The Logistic chaotic map generates the initial population, which improves the diversity of the search and the global exploration ability, while the cosine control factor dynamically adjusts the balance between global and local searches during the search process to ensure accurate convergence. The Cauchy mutation operation enhances the jumping ability and helps to break through the local optimum. By adaptively adjusting the controller parameters, the regulation of infrared light intensity is finally optimized, providing personalized and accurate health intervention solutions, and improving the performance, stability and user experience of the system.

[0149] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0150] In the present invention, by extracting features from the acceleration signal and gyroscope signal of the human waist, a plurality of signal features are obtained, each of which is input into a human posture recognition model based on a random forest algorithm for posture recognition, and the current human posture is determined. According to the current human posture, a control mode corresponding to the current human posture is determined, and a fixed control mode is no longer used to assist the user. The differences in individual needs under different postures can be taken into account, and the control mode corresponding to the current human posture is determined through PI. λ D μ The controller adaptively controls the light intensity of the infrared lamp beads in the belt according to the determined control mode. The user does not need to manually adjust the control mode, which provides convenience and reduces the operating burden. It has strong flexibility in adapting to changes in human posture and can adjust the control mode and light intensity in real time according to actual conditions. It is highly comfortable and can effectively improve the energy efficiency of the equipment.

[0151] Reference Manual Attached Figure 2, showing a schematic flow chart of a waist belt adaptive control method provided by the present invention.

[0152] The embodiment of the present invention also provides a waistband adaptive control method, which can be implemented by a waistband adaptive control device, and the waistband adaptive control device can be a terminal or a server. The processing flow of the waistband adaptive control method may include the following steps:

[0153] S1: Obtain the acceleration signal and gyroscope signal of the human waist.

[0154] In a possible implementation, S1 specifically includes sub-steps S101 and S102:

[0155] S101: Obtaining three-axis acceleration signals of the X-axis, Y-axis and Z-axis of the human waist respectively through a three-axis accelerometer.

[0156] S102: Obtain three-axis gyroscope signals of the X-axis, Y-axis and Z-axis of the human waist through the three-axis gyroscope.

[0157] S2: Extract features of the acceleration signal and the gyroscope signal to obtain multiple signal features.

[0158] In a possible implementation, S2 specifically includes sub-steps S201 and S202:

[0159] S201: Preprocess the acceleration signal and the gyroscope signal.

[0160] S202: Extract features from the preprocessed acceleration signal and gyroscope signal to obtain multiple signal features.

[0161] Optionally, the signal characteristics specifically include: arithmetic mean, standard deviation, median absolute deviation, maximum value, minimum value, average of the sum of squares, signal amplitude area, signal entropy, interquartile range, 4th-order Burg autoregression coefficient, Pearson correlation coefficient, the angle between the signal mean and the vector, the range between the minimum and maximum values, root mean square value, frequency signal skewness, frequency signal kurtosis, maximum frequency component, frequency signal weighted average and spectral energy of the frequency interval.

[0162] S3: Build a human posture recognition model based on random forest algorithm.

[0163] S4: Input each signal feature into a human posture recognition model based on a random forest algorithm to perform posture recognition and determine the current human posture.

[0164] In a possible implementation, S4 specifically includes sub-steps S401 to S406:

[0165] S401: construct an initial forest consisting of multiple decision trees, where the decision tree includes multiple nodes.

[0166] S402: Calculate the information gain and Gini index of the current node of each decision tree:

[0167]

[0168] Where Gain() represents information gain, M represents the number of signal features in the current node, a represents the attribute of the signal feature, Ent() represents the entropy of the signal feature set, V represents the number of child nodes generated after node division, || represents taking the absolute value, and M v represents the number of signal features in the vth child node after node division, y represents the number of categories, and p k It represents the probability that the signal feature belongs to the kth category, and Gini() represents the Gini index.

[0169] S403: Dynamically determine the node division rule based on the information gain and Gini index of the current node:

[0170]

[0171] Among them, O represents the node division rule, min represents the minimum value, W{} represents the comprehensive function, α represents the weight coefficient of the Gini index, and θ represents the weight coefficient of the information gain.

[0172] S404: According to the node division rule, use each signal feature to perform node division on the current node.

[0173] S405: Repeat sub-steps S402 to S404, and stop node division when the number of signal features in the current node is less than a preset number, to obtain multiple posture categories output by multiple decision trees.

[0174] S406: According to each posture category, a posture category with the most votes is determined as the current human body posture through a voting mechanism.

[0175] S5: According to the current human body posture, determine a control mode corresponding to the current human body posture.

[0176] S6: Via PI λ D μ The controller adaptively controls the light intensity of the infrared lamp beads in the belt according to the determined control mode.

[0177] In a possible implementation, S6 specifically includes sub-steps S601 to S603:

[0178] S601: Obtain the target light intensity in the current control mode.

[0179] S602: Obtain the actual light intensity in the current control mode.

[0180] S603: Through PI λ D μ The controller outputs a control signal to adaptively control the light intensity of the infrared lamp beads in the belt.

[0181] In a possible implementation manner, the waist belt adaptive control method further includes:

[0182] S7: Construct an objective function with the goal of improving the comfort index during the control mode switching process.

[0183] Among them, the objective function is specifically:

[0184] maxf(δ)=w P ·C P +w H ·C H

[0185] Among them, max means taking the maximum value, f() means the objective function, and δ means PI λ D μ The control parameter set of the controller, PI λ D μ The control parameter set of the controller includes proportional gain coefficient, integral gain coefficient, differential gain coefficient, integral order and differential order, w P Represents the weight coefficient of the pressure comfort score, C P represents the pressure comfort score, w H Represents the weight coefficient of the thermal comfort score, C H Represents the thermal comfort score.

[0186] Optionally, the calculation method of the comfort index is as follows:

[0187] By evenly setting multiple pressure sensors in the waist belt, waist pressure data can be obtained:

[0188]

[0189] Among them, P i represents the waist pressure data obtained by the i-th pressure sensor, F i It represents the pressure exerted by the human body on the i-th pressure sensor, i=1,2,…n, n represents the total number of pressure sensors, and ΔS represents the sensing area of ​​the pressure sensor.

[0190] According to the waist pressure data obtained by each pressure sensor, the maximum waist pressure data is determined:

[0191] P max =max(P 1 ,P 2 ,...,P n )

[0192] Among them, P max Indicates the maximum waist pressure data, max means taking the maximum value, P 1 Indicates the waist pressure data obtained by the first pressure sensor, P 2 represents the waist pressure data obtained by the second pressure sensor, P n Represents the waist pressure data obtained by the nth pressure sensor.

[0193] According to the waist pressure data obtained by each pressure sensor, the average waist pressure data is determined:

[0194]

[0195] Among them, P avg Indicates average waist pressure data.

[0196] The pressure comfort score is determined based on the waist pressure data, maximum waist pressure data, and average waist pressure data obtained by each pressure sensor:

[0197]

[0198] Among them, C p Indicates the pressure comfort score.

[0199] Waist temperature data is obtained by evenly setting multiple temperature sensors in the waist belt.

[0200] According to the waist temperature data obtained by each temperature sensor, the maximum waist temperature data is determined:

[0201] H max =max(H 1 ,H 2 ,...,H m )

[0202] Among them, H max Indicates the maximum waist temperature data, H 1 Indicates the waist temperature data obtained by the first temperature sensor, H 2 Indicates the waist temperature data obtained by the second temperature sensor, H m Represents the waist temperature data obtained by the mth temperature sensor.

[0203] According to the waist temperature data obtained by each temperature sensor, the average waist temperature data is determined:

[0204]

[0205] Among them, H avg Indicates the average waist temperature data, H j represents the waist temperature data obtained by the jth temperature sensor, j=1, 2, ...m, and m represents the total number of temperature sensors.

[0206] The temperature comfort score is determined based on the waist temperature data, maximum waist temperature data, and average waist temperature data obtained by each temperature sensor:

[0207]

[0208] Among them, C H Represents the thermal comfort score.

[0209] The comfort index is calculated based on the pressure comfort score and the temperature comfort score:

[0210] ρ=w P ·C P +w H ·C H

[0211] Among them, ρ represents the comfort index, w P Represents the weight coefficient of pressure comfort score, w H Represents the weight coefficient of the thermal comfort score.

[0212] S8: According to the objective function, PI is optimized by krill swarm optimization algorithm. λ D μ Optimize the controller.

[0213] It should be noted that the waistband adaptive control method can be implemented by the above-mentioned waistband adaptive control system and can achieve the same or similar technical effects. To avoid repetition, the present invention will not go into details.

[0214] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0215] In the present invention, by extracting features from the acceleration signal and gyroscope signal of the human waist, a plurality of signal features are obtained, each of which is input into a human posture recognition model based on a random forest algorithm for posture recognition, and the current human posture is determined. According to the current human posture, a control mode corresponding to the current human posture is determined, and a fixed control mode is no longer used to assist the user. The differences in individual needs under different postures can be taken into account, and the control mode corresponding to the current human posture is determined through PI. λ D μThe controller adaptively controls the light intensity of the infrared lamp beads in the belt according to the determined control mode. The user does not need to manually adjust the control mode, which provides convenience and reduces the operating burden. It has strong flexibility in adapting to changes in human posture and can adjust the control mode and light intensity in real time according to actual conditions. It is highly comfortable and can effectively improve the energy efficiency of the equipment.

[0216] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art who is familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

[0217] There are a few points to note:

[0218] (1) The drawings of the embodiments of the present invention only relate to the structures related to the embodiments of the present invention, and other structures may refer to the general design.

[0219] (2) For the sake of clarity, in the drawings used to describe the embodiments of the present invention, the thickness of the layers or regions is exaggerated or reduced, that is, these drawings are not drawn according to the actual scale. It is understood that when an element such as a layer, film, region or substrate is referred to as being "on" or "under" another element, the element may be "directly" "on" or "under" the other element or there may be intermediate elements.

[0220] (3) In the absence of conflict, the embodiments of the present invention and the features therein may be combined with each other to obtain new embodiments.

[0221] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. The protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A waist belt adaptive control system, characterized in that: include: An acquisition module is used to acquire acceleration signals and gyroscope signals of the human waist; An extraction module, used for performing feature extraction on the acceleration signal and the gyroscope signal to obtain a plurality of signal features; The first building module is used to build a human posture recognition model based on the random forest algorithm; A recognition module, used for inputting each of the signal features into the human posture recognition model based on the random forest algorithm to perform posture recognition and determine the current human posture; A determination module, used to determine a control mode corresponding to the current human body posture according to the current human body posture; Control module for PI λ D μ The controller adaptively controls the light intensity of the infrared lamp beads in the belt according to the determined control mode.

2. The waist belt adaptive control system according to claim 1, characterized in that: The acquisition module is specifically used for: The three-axis acceleration signals of the X-axis, Y-axis and Z-axis of the human waist are obtained respectively through the three-axis accelerometer; The three-axis gyroscope is used to obtain the three-axis gyroscope signals of the X-axis, Y-axis and Z-axis of the human waist respectively.

3. The waist belt adaptive control system according to claim 1, characterized in that: The extraction module is specifically used for: Preprocessing the acceleration signal and the gyroscope signal; Feature extraction is performed on the preprocessed acceleration signal and gyroscope signal to obtain multiple signal features.

4. The waist belt adaptive control system according to claim 3, characterized in that: The signal characteristics specifically include: arithmetic mean, standard deviation, median absolute deviation, maximum value, minimum value, average value of the sum of squares, signal amplitude area, signal entropy, interquartile range, 4th-order Burg autoregression coefficient, Pearson correlation coefficient, angle between signal mean and vector, range between minimum and maximum value, root mean square value, frequency signal skewness, frequency signal kurtosis, maximum frequency component, frequency signal weighted average value and spectral energy of frequency interval.

5. The waist belt adaptive control system according to claim 1, characterized in that: The identification module is specifically used for: Constructing an initial forest consisting of a plurality of decision trees, wherein the decision trees include a plurality of nodes; Calculate the information gain and Gini index of the current node of each decision tree; Combine the information gain and Gini index of the current node to dynamically determine the node division rules; According to the node division rule, using each of the signal features to perform node division on the current node; When the number of signal features in the current node is less than a preset number, node division is stopped to obtain multiple posture categories output by multiple decision trees; According to each of the posture categories, a voting mechanism is used to determine the posture category with the most votes as the current human body posture.

6. The waist belt adaptive control system according to claim 1, characterized in that: The control module is specifically used for: Get the target light intensity in the current control mode; Get the actual light intensity in the current control mode; By the PI λ D μ The controller outputs a control signal to adaptively control the light intensity of the infrared lamp beads in the belt.

7. The waist belt adaptive control system according to claim 1, characterized in that: Also includes: The second building module is used to build an objective function with the goal of improving the comfort index during the control mode switching process; An optimization module is used to optimize the PI according to the objective function through a krill swarm optimization algorithm. λ D μ Optimize the controller.

8. The waist belt adaptive control system according to claim 7, characterized in that: The calculation method of the comfort index is specifically as follows: Obtain waist pressure data by evenly arranging a plurality of pressure sensors in the waist belt; Determining maximum waist pressure data according to the waist pressure data acquired by each of the pressure sensors; Determining average waist pressure data according to the waist pressure data acquired by each of the pressure sensors; Determine a pressure comfort score according to the waist pressure data acquired by each of the pressure sensors, the maximum waist pressure data, and the average waist pressure data; By evenly arranging a plurality of temperature sensors in the waist belt, waist temperature data is obtained; Determining maximum waist temperature data according to the waist temperature data acquired by each of the temperature sensors; Determine average waist temperature data according to the waist temperature data acquired by each of the temperature sensors; determining a temperature comfort score according to the waist temperature data acquired by each of the temperature sensors, the maximum waist temperature data, and the average waist temperature data; The comfort index is calculated according to the pressure comfort score and the temperature comfort score.

9. A waist belt adaptive control method, characterized in that: include: S1: Obtain acceleration signal and gyroscope signal of human waist; S2: extracting features from the acceleration signal and the gyroscope signal to obtain a plurality of signal features; S3: Build a human posture recognition model based on random forest algorithm; S4: inputting each of the signal features into the human posture recognition model based on the random forest algorithm to perform posture recognition and determine the current human posture; S5: determining a control mode corresponding to the current human body posture according to the current human body posture; S6: Via PI λ D μ The controller adaptively controls the light intensity of the infrared lamp beads in the belt according to the determined control mode.

Images