Intelligent control method for TWS earphone heart rate detection and earphone function adaptive adjustment

Abstract

The invention relates to the technical field of wearable equipment and intelligent audio processing, in particular to an intelligent control method for TWS earphone heart rate detection and earphone function adaptive adjustment, which comprises the following steps: S1, synchronously acquiring an original heart rate signal, a contact pressure signal and a motion acceleration signal; s2, performing real-time baseline correction on the original heart rate signal; s3, the motion acceleration signals are fused to calculate motion interference weight factors, and a multi-modal physiological feature set is generated; s4, outputting a multi-dimensional physiological state vector; s5, matching a predefined adjustment strategy library according to the multi-dimensional physiological state vector; and S6, analyzing the combined control instruction set. According to the TWS earphone, physiological signals are accurately collected through the ear canal composite sensor array, high-precision correction of heart rate detection is achieved in combination with dynamic compensation, feature extraction and physiological state classification, noise reduction, audio equalization and ear canal pressure adjustment parameters are intelligently matched based on an adjustment strategy library, and the TWS earphone can adapt to the physiological state of a user.

Description

Technical Field

[0001] The present invention relates to the technical fields of wearable devices and intelligent audio processing, and particularly to an intelligent control method for heart rate detection and adaptive adjustment of headphone functions of TWS earphones. Background Art

[0002] With the development of wireless audio technology, true wireless stereo (TWS) earphones have gradually become mainstream due to their convenience, intelligent features, and high-quality audio experience. However, TWS earphones are not only an audio playback device but also gradually developing towards health monitoring and intelligent adjustment. The heart rate detection technology based on ear canal optoelectronic sensing has been widely used in intelligent earphones due to its high measurement stability and comfort. In addition, to improve the user's auditory experience, TWS earphones are usually equipped with active noise reduction functions, audio equalization adjustment functions, and wearing comfort optimization mechanisms, enabling them to achieve noise reduction optimization and personalized audio adjustment in different environments.

[0003] However, in practical applications, due to changes in factors such as the wearing method of the earphones, the user's exercise state, and environmental noise, the accuracy of heart rate detection is easily disturbed, and at the same time, traditional noise reduction and audio adjustment methods are difficult to dynamically adapt to the user's real-time physiological state, affecting the wearing experience and detection accuracy. Summary of the Invention

[0004] Based on the above purposes, the present invention provides an intelligent control method for heart rate detection and adaptive adjustment of headphone functions of TWS earphones.

[0005] The intelligent control method for heart rate detection and adaptive adjustment of headphone functions of TWS earphones includes the following steps:

[0006] S1: Synchronously collect the original heart rate signal, contact pressure signal, and motion acceleration signal through an ear canal composite sensor array, which includes a pressure-sensitive layer, an infrared optoelectronic sensor, and a three-dimensional accelerometer;

[0007] S2: Generate a dynamic pressure compensation coefficient based on the contact pressure signal collected in S1, and perform real-time baseline correction on the original heart rate signal to output a preprocessed heart rate signal;

[0008] S3: Extract time-frequency domain feature parameters from the preprocessed heart rate signal, and at the same time fuse the motion acceleration signal to calculate the motion interference weight factor to generate a multi-modal physiological feature set;

[0009] S4: Input the multi-modal physiological feature set into a physiological state classification model to output a multi-dimensional physiological state vector including the user's exercise intensity level, cardiovascular load index, and auditory sensitivity;

[0010] S5: Match the predefined adjustment strategy library according to the multi-dimensional physiological state vector to generate a joint control instruction set for the noise reduction parameter instruction, the audio equalization instruction, and the ear canal pressure adjustment instruction;

[0011] S6: Parse the joint control instruction set through the headphone main control chip to synchronously drive the active noise reduction module to adjust the frequency band attenuation amount, the audio processing unit to update the equalization curve, and the micro air pump to adjust the ear canal contact pressure.

[0012] Optionally, the S1 specifically includes:

[0013] S11: Collect real-time pressure data at the interface between the ear canal and the headphone by closely contacting the pressure-sensitive layer set on the surface of the headphone earplug with the ear canal skin. The pressure detection range is 0.5 - 5.0 N, and the output is the contact pressure signal;

[0014] S12: Emit infrared light with a wavelength of 850 - 950 nm into the subcutaneous tissue layer of the ear canal through the infrared optoelectronic sensor embedded in the earplug of the headphone, and detect the change in the intensity of the reflected infrared light to realize the real-time collection of the photoplethysmogram signal. The sampling frequency is 50 - 200 Hz, and the output is the original heart rate signal;

[0015] S13: Real-time measure the acceleration changes in the three-axis directions during the head movement through the three-axis accelerometer integrated in the headphone body. The range is ±2g - ±16g, and the sampling frequency is 20 - 100 Hz. The output is the motion acceleration signal;

[0016] S14: Synchronously input the output signals of the pressure-sensitive layer, the infrared optoelectronic sensor, and the three-axis accelerometer into the data acquisition circuit of the composite sensor array to achieve the timestamp synchronization of multi-sensor data with a unified clock reference. The synchronization accuracy error is less than 1 ms, and a synchronized data frame is encapsulated.

[0017] Optionally, the S2 specifically includes:

[0018] S21: Perform a sliding window process on the contact pressure signal collected in S1. The window length is 1 - 3 seconds to calculate the short-time mean value of the contact pressure signal, and calculate the dynamic pressure compensation coefficient according to the ratio of the short-time mean value to the preset pressure reference value;

[0019] S22: Perform a band-pass filtering process on the original heart rate signal collected in S1 to filter out the DC drift and high-frequency noise components to obtain the filtered heart rate signal;

[0020] S23: Use the dynamic pressure compensation coefficient generated in S21 to perform real-time correction on the amplitude of the heart rate signal filtered in S22 to offset the signal amplitude change caused by the ear canal contact pressure fluctuation;

[0021] S24: Perform adaptive baseline tracking on the heart rate signal after amplitude correction, update the baseline value through recursive moving average method and perform real-time baseline subtraction, and finally output the preprocessed heart rate signal.

[0022] Optionally, the S24 specifically includes:

[0023] S241: Set the initial baseline value as the initial mean of the preset heart rate signal amplitude;

[0024] S242: Perform real-time sampling on the amplitude-corrected heart rate signal output by S23, and use the recursive moving average method to dynamically update the baseline value. The specific formula is: B(t) = α × H corr (t) + (1 - α) × B(t - 1), where B(t) represents the updated baseline value at the current moment; B(t - 1) represents the baseline value at the previous moment; α represents the baseline update coefficient, and the value range is 0.01 to 0.1;

[0025] S243: Use the baseline value updated by S242 to perform baseline subtraction on the real-time heart rate signal to eliminate the influence of baseline drift. The subtraction expression is: H out (t) = H corr (t) - B(t), where H out (t) represents the heart rate signal after real-time baseline subtraction;

[0026] S244: Output the heart rate signal after real-time baseline subtraction obtained by S243 as the preprocessed heart rate signal.

[0027] Optionally, the S3 specifically includes:

[0028] S31: Segment the heart rate signal preprocessed by S2 with an analysis window of a fixed length of 8 seconds, and the overlap rate between each segment of the signal is 50%;

[0029] S32: Perform time-domain analysis on each segment of the signal, and extract time-domain characteristic parameters such as the mean of adjacent heart beat peak intervals, the standard deviation of heart beat intervals, and the coefficient of variation of signal amplitude;

[0030] S33: Perform frequency-domain analysis on each segment of the signal, use the fast Fourier transform method to convert the time-domain signal to the frequency domain, and calculate frequency-domain characteristic parameters such as the low-frequency power of 0.04 - 0.15 Hz, the high-frequency power of 0.15 - 0.4 Hz, and the ratio of the two;

[0031] S34: Combine the time-domain and frequency-domain characteristic parameters extracted by S32 and S33 to obtain comprehensive time-frequency domain characteristic parameters of the heart rate signal;

[0032] S35: Calculate the vector magnitude integral of the three-axis motion acceleration signals collected in S1 to characterize the overall user motion intensity within the current 8-second window, and normalize it to the range of 0 to 1 as the motion interference weight factor;

[0033] S36: Perform feature fusion on the heart rate signal time-frequency domain feature parameters obtained in S34 and the motion interference weight factor obtained in S35 to form a multi-modal physiological feature set that includes heart rate time-frequency domain features and the motion interference weight factor.

[0034] Optionally, the specific steps of S4 are as follows:

[0035] S41: Input the multi-modal physiological feature set obtained in S3 into a pre-trained physiological state classification model, and the classification model is constructed using the support vector machine algorithm;

[0036] S42: Analyze the motion interference weight factor in the multi-modal physiological feature set through the physiological state classification model, and output the user's motion intensity level, where the motion intensity level includes three levels: low intensity, medium intensity, and high intensity;

[0037] S43: Analyze the time-domain feature and frequency-domain feature parameters of the heart rate signal in the multi-modal physiological feature set through the physiological state classification model, and calculate the cardiovascular load index;

[0038] S44: Based on the calculated motion intensity level and cardiovascular load index, combined with the pre-stored user auditory sensitivity mapping relationship table, determine the user's current auditory sensitivity level, where the auditory sensitivity level includes three levels: sensitive, normal, and dull;

[0039] S45: Combine the motion intensity level output by S42, the cardiovascular load index calculated by S43, and the auditory sensitivity level determined by S44 to generate a multi-dimensional physiological state vector for headphone adaptive adjustment.

[0040] Optionally, the specific steps of S42 are as follows:

[0041] S421: Input the motion interference weight factor obtained in S35 into the motion intensity level sub-classifier trained in S413, and perform analysis based on the decision function of this sub-classifier;

[0042] S422: According to the decision function of the motion intensity level sub-classifier, calculate the decision function value f act (M);

[0043] S423: Based on the magnitude of the decision function value f act (M), output the user's motion intensity level, and the expression is:

[0044]

[0045] Optionally, S5 specifically includes:

[0046] S51: Establish and pre-store an adjustment strategy library. The strategy library uses a multi-dimensional physiological state vector as the input index, corresponding to noise reduction parameter instructions, audio equalization instructions, and ear canal pressure adjustment instructions under different combined conditions respectively;

[0047] S52: According to the exercise intensity level, cardiovascular load index, and auditory sensitivity level in the multi-dimensional physiological state vector output by S4, quickly locate the unique strategy combination in the adjustment strategy library;

[0048] S53: Extract the corresponding noise reduction parameter instructions from the located strategy combination. The noise reduction parameter instructions include the attenuation amount of the active noise reduction frequency band;

[0049] S54: Extract the corresponding audio equalization instructions from the located strategy combination. The audio equalization instructions include the adjustment gain of the audio equalization curve;

[0050] S55: Extract the corresponding ear canal pressure adjustment instructions from the located strategy combination. The ear canal pressure adjustment instructions specifically include the target contact pressure value inside the ear canal;

[0051] S56: Combine and encapsulate the noise reduction parameter instructions, audio equalization instructions, and ear canal pressure adjustment instructions obtained in S53, S54, and S55 respectively to form a joint control instruction set.

[0052] Optionally, S51 specifically includes:

[0053] S511: Pre-define three dimensions of the user's multi-dimensional physiological state vector, including the exercise intensity level, the range of cardiovascular load index, and the auditory sensitivity level, and enumerate all state combinations;

[0054] S512: For each enumerated state combination, conduct ergonomics and auditory comfort experiments in advance to determine the corresponding optimal noise reduction parameter instructions, audio equalization instructions, and ear canal pressure adjustment instructions;

[0055] S513: Based on the experimental results obtained in S512, record and construct an adjustment strategy mapping table item by item, using the multi-dimensional physiological state vector as the index;

[0056] S514: Store the adjustment strategy mapping table constructed in S513 in the form of a data table in the built-in memory of the earphone to form a complete adjustment strategy library.

[0057] Optionally, S6 specifically includes:

[0058] S61: The headphone main control chip receives the combined control instruction set output by S56, and extracts the noise reduction parameter instruction, the audio equalization instruction, and the ear canal pressure adjustment instruction according to the data frame structure of the instruction set;

[0059] S62: The main control chip parses the extracted noise reduction parameter instruction into specific noise reduction frequency band attenuation amount values, and generates corresponding frequency band attenuation control signals to be sent to the active noise reduction module to drive the active noise reduction module to update the frequency band attenuation amount to the target value in real time;

[0060] S63: The main control chip parses the extracted audio equalization instruction into audio frequency band gain adjustment values, and updates the audio equalization filter coefficients in the audio processing unit;

[0061] S64: The main control chip parses the extracted ear canal pressure adjustment instruction into the target ear canal contact pressure, and controls the built-in micro air pump to adjust the air intake or air release amount of the inflatable airbag, and adjusts the actual contact pressure between the earphone insertion end and the ear canal wall in real time until the target pressure value is reached;

[0062] S65: The main control chip synchronously executes the control actions of S62, S63, and S64 based on a unified internal clock reference to ensure that the adjustment actions of the active noise reduction module, the audio processing unit, and the micro air pump are completed synchronously.

[0063] Advantages of the present invention:

[0064] In the present invention, based on the ear canal composite sensor array, the original heart rate signal, the contact pressure signal, and the motion acceleration signal are synchronously collected, and through the dynamic pressure compensation and real-time baseline correction technologies, the stability and accuracy of heart rate detection are significantly improved; by adopting the multi-modal feature extraction and physiological state classification model, the exercise intensity, cardiovascular load, and auditory sensitivity of the user can be accurately evaluated to form a multi-dimensional physiological state vector; this solution effectively solves the problem that the heart rate detection of existing TWS earphones is easily affected by the change of wearing pressure and motion interference, makes the heart rate data more reliable, and provides more accurate physiological monitoring results for users.

[0065] In the present invention, by establishing an adjustment strategy library, the real-time physiological state of the user is combined with active noise reduction, audio equalization, and ear canal contact pressure adjustment to achieve intelligent personalized adjustment; the headphone main control chip parses the combined control instruction set to ensure the synchronous adjustment of the active noise reduction module, the audio processing unit, and the micro air pump, so that the noise reduction effect, audio equalization, and wearing comfort can all match the physiological state of the user in real time; compared with the traditional fixed parameter adjustment method, this method greatly improves the intelligent adaptation ability of TWS earphones, enables them to dynamically adapt to user needs, and provides a better auditory experience and wearing comfort in different scenarios. Description of the Drawings

[0066] To more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only those of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0067] Figure 1 Schematic diagram of the intelligent control method for adaptively adjusting the functions of the earphone in the embodiment of the present invention;

[0068] Figure 2 Schematic diagram of the process for performing real-time baseline correction in the embodiment of the present invention. Detailed implementation manners

[0069] The following will describe the present invention in detail with reference to the drawings and specific embodiments. At the same time, it should be noted here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments. For some well-known technologies, those skilled in the art can also adopt other alternative methods for implementation; moreover, the drawings are only for more specifically describing the embodiments and are not intended to specifically limit the present invention.

[0070] It should be noted that in the specification, when referring to "an embodiment", "embodiment", "exemplary embodiment", "some embodiments", etc., it indicates that the described embodiment may include specific features, structures or characteristics, but not necessarily every embodiment includes such specific features, structures or characteristics. Additionally, when combining embodiments to describe specific features, structures or characteristics, implementing such features, structures or characteristics in combination with other embodiments (whether explicitly described or not) should be within the knowledge of those skilled in the relevant art.

[0071] Generally, terms can be understood at least in part from their use in the context. For example, at least in part depending on the context, the term "one or more" used herein can be used to describe any feature, structure or characteristic in a singular sense, or can be used to describe a combination of features, structures or characteristics in a plural sense. Additionally, the term "based on" can be understood as not necessarily aiming to convey a set of exclusive factors, but rather, at least in part depending on the context, allowing for the existence of other factors that may not be explicitly described.

[0072] As Figure 1 - Figure 2 shown, the intelligent control method for heart rate detection and adaptive adjustment of the functions of TWS earphones includes the following steps:

[0073] S1: Synchronously collect the original heart rate signal, contact pressure signal and motion acceleration signal through the ear canal composite sensor array, and this composite sensor array includes a pressure-sensitive layer, an infrared photoelectric sensor and a three-dimensional accelerometer;

[0074] S2: Generate a dynamic pressure compensation coefficient based on the contact pressure signal collected in S1, perform real-time baseline correction on the original heart rate signal, and output the preprocessed heart rate signal;

[0075] S3: Extract time-frequency domain characteristic parameters from the preprocessed heart rate signal, and at the same time fuse the motion acceleration signal to calculate the motion interference weight factor to generate a multi-modal physiological feature set;

[0076] S4: Input the multi-modal physiological feature set into the physiological state classification model, and output a multi-dimensional physiological state vector including the user's exercise intensity level, cardiovascular load index, and auditory sensitivity;

[0077] S5: Match the pre-defined adjustment strategy library according to the multi-dimensional physiological state vector to generate a joint control instruction set of noise reduction parameter instructions, audio equalization instructions, and ear canal pressure adjustment instructions;

[0078] S6: Parse the joint control instruction set through the headphone main control chip to synchronously drive the active noise reduction module to adjust the frequency band attenuation amount, the audio processing unit to update the equalization curve, and the micro air pump to adjust the ear canal contact pressure.

[0079] S1 specifically includes:

[0080] S11: Collect real-time pressure data at the contact interface between the ear canal and the headphone by closely contacting the pressure-sensitive layer set on the surface of the headphone earplug with the ear canal skin. The pressure detection range is 0.5 - 5.0 N, and the output is the contact pressure signal;

[0081] S12: Emitting infrared light with a wavelength of 850 - 950 nm into the subcutaneous tissue layer of the ear canal through the infrared photoelectric sensor embedded in the earplug of the headphone, and detecting the change in the intensity of the reflected infrared light to realize the real-time collection of the photoplethysmogram signal. The sampling frequency is 50 - 200 Hz, and the output is the original heart rate signal;

[0082] S13: Real-time measure the acceleration changes in the three-axis directions during the head movement through the three-axis accelerometer integrated in the headphone body. The measurement range is ±2g - ±16g, and the sampling frequency is 20 - 100 Hz, and the output is the motion acceleration signal;

[0083] S14: Synchronously input the output signals of the pressure-sensitive layer, the infrared photoelectric sensor, and the three-axis accelerometer into the data acquisition circuit of the composite sensor array to realize the timestamp synchronization of multi-sensor data with a unified clock reference. The synchronization accuracy error is less than 1 ms, and package to form a synchronized data frame; Through the implementation of the above sub-steps, the real-time synchronous acquisition of various sensor data can be accurately realized, providing a reliable data basis for the subsequent accurate processing of heart rate and the accurate evaluation of multi-dimensional physiological states.

[0084] S2 specifically includes:

[0085] S21: Perform a sliding window process on the contact pressure signal collected in S1. The window length is 1 to 3 seconds to calculate the short-time mean value of the contact pressure signal, and calculate the dynamic pressure compensation coefficient based on the ratio of the short-time mean value to the preset pressure reference value. The specific calculation formula is as follows: where k p represents the dynamic pressure compensation coefficient; P avg represents the short-time mean value of the contact pressure signal within the current sliding window; P ref represents the preset pressure reference value, and the value range is 1.5 to 2.5 N;

[0086] S22: Perform a band-pass filtering process on the original heart rate signal collected in S1. The filtering frequency range is 0.5 to 5 Hz to filter out the DC drift and high-frequency noise components and obtain the filtered heart rate signal;

[0087] S23: Use the dynamic pressure compensation coefficient generated in S21 to perform real-time correction on the amplitude of the heart rate signal filtered in S22 to offset the signal amplitude change caused by the ear canal contact pressure fluctuation. The specific calculation formula is as follows: where H corr (t) represents the heart rate signal after amplitude real-time correction; H filt (t) represents the filtered heart rate signal obtained in S22;

[0088] S24: Perform adaptive baseline tracking on the heart rate signal after amplitude correction, update the baseline value through the recursive moving average method and perform real-time baseline subtraction, and finally output the preprocessed heart rate signal. Through the implementation of the above sub-steps, the interference caused by the contact pressure fluctuation to the heart rate signal acquisition can be effectively eliminated, and a more stable and reliable preprocessed heart rate signal can be obtained, providing high-quality data support for subsequent accurate feature extraction and physiological state classification.

[0089] S24 specifically includes:

[0090] S241: Set the initial baseline value as the initial mean value of the preset heart rate signal amplitude and initialize the relevant parameters of the recursive moving average algorithm;

[0091] S242: Perform real-time sampling on the heart rate signal after amplitude correction output in S23, and use the recursive moving average method to dynamically update the baseline value. The specific formula is: B(t) = α × H corr (t)+(1 - α) × B(t - 1), where B(t) represents the updated baseline value at the current moment; B(t - 1) represents the baseline value at the previous moment; H corr (t) represents the heart rate signal after amplitude real-time correction; α represents the baseline update coefficient, and the value range is 0.01 to 0.1;

[0092] S243: Baseline subtraction is performed on the real-time heart rate signal using the baseline value obtained by updating with S242 to eliminate the influence of baseline drift. The subtraction expression is: H out (t) = H corr (t) - B(t), where H out (t) represents the heart rate signal after real-time baseline subtraction;

[0093] S244: Output the heart rate signal after real-time baseline subtraction obtained in S243 as the preprocessed heart rate signal for subsequent feature extraction; Through the specific implementation of the above sub-steps, the baseline drift interference in the heart rate signal can be efficiently and real-time eliminated, significantly improving the stability and reliability of the heart rate signal, and providing a high-quality data basis for the accurate extraction of subsequent feature parameters and the accurate assessment of physiological states.

[0094] S3 specifically includes:

[0095] S31: Segment the heart rate signal preprocessed by S2 with an analysis window of a fixed length of 8 seconds, and the overlap rate between each segment of the signal is 50%;

[0096] S32: Perform time-domain analysis on each segment of the signal, and extract time-domain feature parameters such as the mean of adjacent R-R intervals, the standard deviation of R-R intervals, and the coefficient of variation of signal amplitude; The specific calculation formulas are as follows:

[0097] Mean of R-R intervals:

[0098] Standard deviation of R-R intervals:

[0099] Coefficient of variation of signal amplitude: where RR i represents the adjacent R-R intervals; N represents the number of R-R intervals within the window; RR mean is the mean of R-R intervals, RR std is the standard deviation of R-R intervals, and CV is the coefficient of variation of signal amplitude;

[0100] S33: Perform frequency-domain analysis on each segment of the signal, use the fast Fourier transform (FFT) method to convert the time-domain signal to the frequency domain, and calculate frequency-domain feature parameters such as the low-frequency (LF) power of 0.04 - 0.15 Hz, the high-frequency (HF) power of 0.15 - 0.4 Hz, and their ratio (LF / HF); The specific calculation formulas are as follows:

[0101] Low-frequency power:

[0102] High-frequency power:

[0103] Power ratio: where PSD(f) is the power spectral density of the heart rate signal spectrum;

[0104] S34: Combine the time-domain and frequency-domain characteristic parameters extracted in S32 and S33 to obtain comprehensive time-frequency domain characteristic parameters of the heart rate signal;

[0105] S35: Calculate the vector magnitude integral of the three-axis motion acceleration signal collected in S1 to characterize the overall motion intensity of the user within the current 8-second window, and normalize it to the range of 0 to 1 as the motion interference weight factor; the specific calculation formula is as follows: where a x (t), a y (t), a z (t) represent the three-axis motion acceleration signals respectively; M is the motion interference weight factor, with a range of 0 to 1; M max is the preset maximum value of the motion integral;

[0106] S36: Perform feature fusion on the time-frequency domain characteristic parameters of the heart rate signal obtained in S34 and the motion interference weight factor obtained in S35 to form a multi-modal physiological feature set containing heart rate time-frequency domain characteristics and motion interference weight factors, and output it to the subsequent physiological state classification model; through the specific implementation of the above sub-steps, the characteristics of the heart rate signal and the motion signal can be extracted and fused with high precision, constructing a more comprehensive and accurate multi-modal physiological feature set, providing a reliable data basis for the accurate classification of the subsequent user's physiological state.

[0107] S4 specifically includes:

[0108] S41: Input the multi-modal physiological feature set obtained in S3 into a pre-trained physiological state classification model. The classification model is constructed using the support vector machine (SVM) algorithm to establish a mapping relationship based on the sample data;

[0109] The specific construction process of the physiological state classification model includes the following sub-steps:

[0110] S411: Collect multiple groups of labeled sample data with different exercise intensity levels, cardiovascular load indices, and auditory sensitivity levels as the training set. The number of samples is not less than 500 groups, and each sample contains a multi-modal physiological feature set;

[0111] S412: Perform feature normalization processing on the multi-modal physiological feature set of each group of samples in the training set. The normalization formula is as follows: where X is the original feature value; X′ is the normalized feature value; X min and X max are the minimum and maximum values of this feature in all training samples respectively;

[0112] S413: Based on the normalized feature data, using the exercise intensity level, cardiovascular load index, and auditory sensitivity level as output categories respectively, train three independent sub-classifiers through the support vector machine algorithm. The kernel function of each sub-classifier selects the radial basis kernel function, and the decision function expression is as follows: where f(x) is the decision function of the sub-classifier; x is the input normalized multi-modal physiological feature vector; x i is the support vector; y i is the class label corresponding to the support vector; α i is the weight coefficient corresponding to the support vector; γ is the kernel parameter of the RBF kernel function; b is the bias constant of the classifier;

[0113] S414: After completing the training of the above three sub-classifiers, combine them to form a physiological state classification model. The overall classification output expression of the model is: Y = {L act , CI, L aud}, where Y is the multi-dimensional physiological state vector output by the model; L act is the exercise intensity level output, with values of low intensity, medium intensity, or high intensity; CI is the cardiovascular load index output, with a value range of 0 - 100; L aud is the auditory sensitivity level output, with values of sensitive, normal, or dull.

[0114] S42: Analyze the exercise interference weight factor in the multi-modal physiological feature set through the physiological state classification model, and output the user's exercise intensity level, which includes three levels: low intensity, medium intensity, and high intensity;

[0115] S43: Analyze the time-domain features and frequency-domain feature parameters of the heart rate signal in the multi-modal physiological feature set through the physiological state classification model, and calculate the cardiovascular load index CI. The formula is: CI = w1 × RR mean + w2 × LF / HF + w3 × CV, where CI is the cardiovascular load index, with a value range of 0 - 100; RR mean is the average value of the heart beat peak interval obtained in S32; LF / HF is the ratio of low-frequency to high-frequency power obtained in S33; CV is the coefficient of variation of the heart rate signal amplitude obtained in S32; w1, w2, w3 are weight coefficients, and w1 + w2 + w3 = 1;

[0116] S44: According to the calculated exercise intensity level and cardiovascular load index, combined with the pre-stored user auditory sensitivity mapping relationship table, determine the user's current auditory sensitivity level, which includes three levels: sensitive, normal, and dull;

[0117] Table 1 Pre-stored User Auditory Sensitivity Mapping Relationship

[0118]

[0119]

[0120] Through the above Table 1, based on the user's physiological state combination obtained in real time, a quick match is made in the pre-stored user auditory sensitivity mapping relation table in S442 to determine the user's current auditory sensitivity level.

[0121] S45: Combine the exercise intensity level output by S42, the cardiovascular load index calculated by S43, and the auditory sensitivity level determined by S44 to generate a multi-dimensional physiological state vector for the adaptive adjustment of the earphone; Through the specific implementation of the above sub-steps, the real-time multi-dimensional physiological state vector of the user can be accurately output, providing an accurate basis for the subsequent personalized adaptive adjustment of the earphone.

[0122] S42 specifically includes:

[0123] S421: Input the motion interference weight factor obtained in S35 into the motion intensity level sub-classifier trained in S413, and analyze it based on the decision function of this sub-classifier;

[0124] S422: According to the decision function of the motion intensity level sub-classifier, calculate the decision function value f act (M), and the formula is: where, f act (M) is the decision function value output by the motion intensity level classifier; M is the input motion interference weight factor; M i is the motion interference weight factor corresponding to the support vector of the motion intensity level sub-classifier; y i is the motion intensity level category label corresponding to the support vector; α i is the weight coefficient corresponding to the support vector; γ is the kernel parameter of the RBF kernel function used by the motion intensity level sub-classifier; b is the bias constant of the motion intensity level sub-classifier;

[0125] S423: Based on the magnitude of the decision function value f act (M), output the user's motion intensity level, and the expression is:

[0126]

[0127] S5 specifically includes:

[0128] S51: Establish and pre-store an adjustment strategy library. The strategy library uses the multi-dimensional physiological state vector as the input index, and respectively corresponds to the noise reduction parameter instructions, audio equalization instructions, and ear canal pressure adjustment instructions under different combination conditions;

[0129] S52: Based on the exercise intensity level, cardiovascular load index, and auditory sensitivity level in the multi-dimensional physiological state vector output by S4, quickly locate the unique policy combination in the adjustment strategy library;

[0130] S53: Extract the corresponding noise reduction parameter instructions from the located policy combination. The noise reduction parameter instructions include the attenuation amount of the active noise reduction frequency band, and the expression is: D anc =[D low , D mid , D high , where D anc is the noise reduction parameter instruction vector; D low , D mid , D high correspond to the attenuation amounts of the low frequency (20 - 200 Hz), middle frequency (200 - 2000 Hz), and high frequency (2000 - 5000 Hz) bands respectively;

[0131] S54: Extract the corresponding audio equalization instructions from the located policy combination. The audio equalization instructions include the audio equalization curve adjustment gain, and the expression is: G eq =[G low , G mid ,, G high , where G eq is the audio equalization instruction gain vector; G low , g mid ,, G high correspond to the audio gain adjustment amounts of the low frequency (20 - 200 Hz), middle frequency (200 - 2000 Hz), and high frequency (2000 - 5000 Hz) bands respectively;

[0132] S55: Extract the corresponding ear canal pressure adjustment instructions from the located policy combination. The ear canal pressure adjustment instructions specifically include the target contact pressure value inside the ear canal, and the expression is: P ear =P base ±ΔP, where P ear is the target contact pressure; P base is the standard ear canal contact pressure; ΔP is the pressure adjustment value determined according to the user's current multi-dimensional physiological state vector, and the value range is 0 - 0.5 N;

[0133] S56: Combine and encapsulate the noise reduction parameter instructions, audio equalization instructions, and ear canal pressure adjustment instructions obtained in S53, S54, and S55 respectively to form a joint control instruction set, and output it to the headphone main control chip for subsequent execution of corresponding adaptive adjustment actions; Through the specific implementation of the above sub-steps, it is possible to quickly and accurately map the user's real-time multi-dimensional physiological state to the optimal headphone adjustment strategy, output a specific and clear joint control instruction set, and provide direct and effective technical support for subsequent headphone adaptive adjustment.

[0134] The adjustment strategy library in S51 specifically includes:

[0135] S511: Pre-define three dimensions of the user's multi-dimensional physiological state vector, including the exercise intensity level, the range of cardiovascular load index, and the auditory sensitivity level, and enumerate all state combinations;

[0136] S512: For each enumerated state combination, conduct ergonomics and auditory comfort experiments in advance to determine the corresponding optimal noise reduction parameter instructions, audio equalization instructions, and ear canal pressure adjustment instructions;

[0137] S513: Based on the experimental results obtained in S512, record and construct an adjustment strategy mapping table item by item, using the multi-dimensional physiological state vector as an index;

[0138] Table 2 Strategy Mapping Table

[0139]

[0140] Perform matching mapping through the above Table 2.

[0141] S514: Store the adjustment strategy mapping table constructed in S513 in the form of a data table in the headphone built-in memory to form a complete adjustment strategy library for real-time matching of the user's multi-dimensional physiological state vector and generating a joint control instruction set; Through the specific implementation of the above sub-steps, it is possible to achieve accurate strategy mapping based on a large amount of experimental data, ensure that the adaptive adjustment actions of the headphones are always highly matched with the user's real-time physiological state, and significantly improve the optimization effect of headphone wearing comfort and auditory experience.

[0142] S6 specifically includes:

[0143] S61: The headphone main control chip receives the joint control instruction set output by S56, and extracts the noise reduction parameter instructions, audio equalization instructions, and ear canal pressure adjustment instructions according to the data frame structure of the instruction set;

[0144] S62: The main control chip parses the extracted noise reduction parameter instructions into specific noise reduction frequency band attenuation amount values, and generates corresponding frequency band attenuation control signals to send to the active noise reduction module to drive the active noise reduction module to update the frequency band attenuation amount to the target value in real time;

[0145] S63: The main control chip parses the extracted audio equalization instruction into an audio frequency band gain adjustment value, and updates the audio equalization filter coefficients in the audio processing unit. The expression is: where, H eq (f) is the updated audio equalization filter frequency response function; H eq0 (f) is the audio equalization filter frequency response function under standard conditions; G eq (f) is the audio equalization gain adjustment given in S54;

[0146] S64: The main control chip parses the extracted ear canal pressure adjustment instruction into a target ear canal contact pressure, and controls the built-in micro air pump to adjust the air intake or air release of the inflatable airbag, and adjusts the actual contact pressure between the earphone earbud and the ear canal wall in real time until the target pressure value is reached. The contact pressure control expression is: P err (t) = P ear -P act (t), where, P err (t) is the ear canal contact pressure adjustment error; P ear is the target ear canal contact pressure determined in S55; -P act (t) is the actually measured ear canal contact pressure at the current moment;

[0147] S65: The main control chip synchronously executes the control actions of S62, S63, and S64 based on a unified internal clock reference, ensuring that the adjustment actions of the active noise reduction module, the audio processing unit, and the micro air pump are synchronously completed, and realizing the overall adaptive adjustment of the headphone functions; Through the specific implementation of the above sub-steps, each function execution module can be accurately parsed and driven synchronously, enabling the active noise reduction effect, audio output equalization, and wearing pressure of the headphones to precisely and real-time adapt to the user's current physiological state, and significantly improving the user's comprehensive wearing comfort and the quality of the auditory experience.

[0148] The present invention covers any substitutions, modifications, equivalent methods, and solutions made within the spirit and scope of the present invention. In order to enable the public to have a thorough understanding of the present invention, specific details are described in detail in the following preferred embodiments of the present invention, and those skilled in the art can fully understand the present invention without these detailed descriptions. Additionally, in order to avoid unnecessary confusion to the essence of the present invention, well-known methods, processes, procedures, components, and circuits are not described in detail.

[0149] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. An intelligent control method for TWS earphone heart rate detection and adaptive adjustment of earphone functions, characterized in that: The following steps are involved: S1: synchronously collects the original heart rate signal, contact pressure signal and motion acceleration signal through the ear canal composite sensor array, which includes a pressure sensitive layer, an infrared photoelectric sensor and a three-dimensional accelerometer; S2: Generates a dynamic pressure compensation coefficient based on the contact pressure signal collected by S1, performs real-time baseline correction on the original heart rate signal, and outputs a preprocessed heart rate signal; S3: extracting time-frequency domain feature parameters from the preprocessed heart rate signal, and fusing the motion acceleration signal to calculate the motion interference weight factor to generate a multimodal physiological feature set; S4: inputting the multimodal physiological feature set into the physiological state classification model, and outputting a multidimensional physiological state vector including the user's exercise intensity level, cardiovascular load index and auditory sensitivity; S5: Generate a joint control instruction set of noise reduction parameter instructions, audio equalization instructions and ear canal pressure adjustment instructions according to the multi-dimensional physiological state vector matching predefined adjustment strategy library; S6: The headphone main control chip parses the joint control instruction set to synchronously drive the active noise reduction module to adjust the frequency band attenuation, the audio processing unit to update the equalization curve, and the micro air pump to adjust the ear canal contact pressure.

2. The intelligent control method for TWS earphone heart rate detection and adaptive adjustment of earphone function according to claim 1 is characterized in that: The S1 specifically includes: S11: The pressure sensitive layer disposed on the surface of the earphone inlet is in close contact with the ear canal skin to collect real-time pressure data at the contact interface between the ear canal and the earphone. The pressure detection range is 0.5 to 5.0N, and the output is a contact pressure signal. S12: The infrared photoelectric sensor embedded in the earphone inlet emits infrared light with a wavelength of 850 to 950 nm into the subcutaneous tissue layer of the ear canal, and detects the intensity change of the reflected infrared light to realize the real-time acquisition of the photoelectric volume pulse wave signal. The sampling frequency is 50 to 200 Hz, and the output is the original heart rate signal; S13: The three-dimensional accelerometer integrated in the headset body measures the acceleration changes in the three-axis directions during the head movement in real time, with a measurement range of ±2g to ±16g, a sampling frequency of 20 to 100Hz, and outputs a motion acceleration signal; S14: The output signals of the pressure sensitive layer, infrared photoelectric sensor and three-dimensional accelerometer are synchronously input to the data acquisition circuit of the composite sensor array, and the timestamp synchronization of the multi-sensor data is realized with a unified clock reference, the synchronization accuracy error is less than 1ms, and the synchronized data frames are encapsulated.

3. The intelligent control method for TWS earphone heart rate detection and adaptive adjustment of earphone function according to claim 1, characterized in that: The S2 specifically includes: S21: performing sliding window processing on the contact pressure signal collected by S1, where the window length is 1 to 3 seconds, to calculate the short-time mean value of the contact pressure signal, and calculating the dynamic pressure compensation coefficient according to the ratio of the short-time mean value to the preset pressure reference value; S22: performing bandpass filtering on the original heart rate signal collected by S1 to filter out DC drift and high-frequency noise components, thereby obtaining a filtered heart rate signal; S23: using the dynamic pressure compensation coefficient generated by S21 to perform real-time correction on the amplitude of the heart rate signal filtered by S22 to offset the signal amplitude change caused by the ear canal contact pressure fluctuation; S24: Adaptively track the heart rate signal after amplitude correction, update the baseline value through the recursive moving average method and perform real-time baseline subtraction, and finally output the preprocessed heart rate signal.

4. The intelligent control method for TWS earphone heart rate detection and adaptive adjustment of earphone function according to claim 3 is characterized in that: The S24 specifically includes: S241: Setting the baseline initial value to the initial mean value of the preset heart rate signal amplitude; S242: Real-time sampling of the amplitude-corrected heart rate signal output by S23, and dynamic updating of the baseline value using the recursive moving average method. The specific formula is: B(t) = α × H corr (t)+(1-α)×B(t-1), where B(t) represents the updated baseline value at the current moment; B(t-1) represents the baseline value at the previous moment; α represents the baseline update coefficient, which ranges from 0.01 to 0.1; S243: Using the baseline value updated in S242 to perform baseline subtraction on the real-time heart rate signal to eliminate the influence of baseline drift, the subtraction expression is: out (t) = H corr (t)-B(t), where H out (t) represents the heart rate signal after real-time baseline subtraction; S244: Output the real-time baseline-subtracted heart rate signal obtained in S243 as the preprocessed heart rate signal.

5. The intelligent control method for TWS earphone heart rate detection and adaptive adjustment of earphone function according to claim 1, characterized in that: The S3 specifically includes: S31: segment the heart rate signal preprocessed by S2 into analysis windows with a fixed length of 8 seconds, and the overlap rate between each signal segment is 50%; S32: Perform time domain analysis on each signal segment to extract time domain characteristic parameters of the mean of the intervals between adjacent heart beat peaks, the standard deviation of the heart beat intervals, and the coefficient of variation of the signal amplitude; S33: Perform frequency domain analysis on each signal segment, convert the time domain signal into the frequency domain using the fast Fourier transform method, and calculate and obtain frequency domain characteristic parameters such as the low frequency power of 0.04-0.15 Hz, the high frequency power of 0.15-0.4 Hz, and the ratio of the two; S34: combining the time domain and frequency domain feature parameters extracted by S32 and S33 to obtain comprehensive time-frequency domain feature parameters of the heart rate signal; S35: Calculate the vector amplitude integral of the three-axis motion acceleration signal collected by S1 to represent the overall motion intensity of the user in the current 8-second window, and normalize it to a range of 0 to 1 as a motion interference weight factor; S36: Fusing the time-frequency domain characteristic parameters of the heart rate signal obtained in S34 with the motion interference weight factor obtained in S35 to form a multimodal physiological feature set including the time-frequency domain characteristics of the heart rate and the motion interference weight factor.

6. The intelligent control method for TWS earphone heart rate detection and adaptive adjustment of earphone function according to claim 1, characterized in that: The S4 specifically includes: S41: inputting the multimodal physiological feature set obtained in S3 into a pre-trained physiological state classification model, wherein the classification model is constructed using a support vector machine algorithm; S42: Analyze the motion interference weight factor in the multimodal physiological feature set through the physiological state classification model, and output the user's exercise intensity level, where the exercise intensity level includes three levels: low intensity, medium intensity, and high intensity; S43: Analyze the time domain characteristics and frequency domain characteristic parameters of the heart rate signal in the multimodal physiological feature set through the physiological state classification model, and calculate the cardiovascular load index; S44: determining the user's current hearing sensitivity level according to the calculated exercise intensity level and cardiovascular load index and in combination with a pre-stored user hearing sensitivity mapping table, wherein the hearing sensitivity level includes three levels: sensitive, normal, and dull; S45: The exercise intensity level output by S42, the cardiovascular load index calculated by S43, and the auditory sensitivity level determined by S44 are combined to generate a multi-dimensional physiological state vector for adaptive adjustment of the headphone.

7. The intelligent control method for TWS earphone heart rate detection and adaptive adjustment of earphone function according to claim 6, characterized in that: The S42 specifically includes: S421: inputting the motion interference weight factor obtained in S35 into the motion intensity level sub-classifier trained in S413, and performing analysis based on the decision function of the sub-classifier; S422: Calculate the decision function value f of the input motion interference weight factor according to the decision function of the motion intensity level sub-classifier act (M); S423: Based on the decision function value f act The size of (M) outputs the user's exercise intensity level, and the expression is:

8. The intelligent control method for TWS earphone heart rate detection and adaptive adjustment of earphone function according to claim 1, characterized in that: The S5 specifically includes: S51: Establishing and pre-storing an adjustment strategy library, the strategy library uses a multi-dimensional physiological state vector as an input index, and corresponds to noise reduction parameter instructions, audio equalization instructions, and ear canal pressure adjustment instructions under different combination conditions; S52: quickly locate a unique strategy combination in the adjustment strategy library according to the exercise intensity level, cardiovascular load index and auditory sensitivity level in the multi-dimensional physiological state vector output by S4; S53: extracting a corresponding noise reduction parameter instruction from the positioning strategy combination, wherein the noise reduction parameter instruction includes an active noise reduction frequency band attenuation amount; S54: extracting a corresponding audio equalization instruction from the positioned strategy combination, where the audio equalization instruction includes an audio equalization curve adjustment gain; S55: extracting a corresponding ear canal pressure adjustment instruction from the positioning strategy combination, where the ear canal pressure adjustment instruction specifically includes a target contact pressure value inside the ear canal; S56: The noise reduction parameter instructions, audio equalization instructions, and ear canal pressure adjustment instructions respectively obtained in S53, S54, and S55 are combined and packaged to form a joint control instruction set.

9. The intelligent control method for TWS earphone heart rate detection and adaptive adjustment of earphone function according to claim 8, characterized in that: The S51 specifically includes: S511: pre-define three dimensions of the user's multi-dimensional physiological state vector, including exercise intensity level, cardiovascular load index range, and auditory sensitivity level, and enumerate all state combinations; S512: For each enumerated state combination, ergonomic and auditory comfort experiments are conducted in advance to determine the corresponding optimal noise reduction parameter instructions, audio equalization instructions, and ear canal pressure adjustment instructions; S513: Based on the experimental results obtained in S512, record and construct a regulation strategy mapping table item by item, using the multi-dimensional physiological state vector as an index; S514: storing the adjustment strategy mapping table constructed in S513 in the form of a data table in the built-in memory of the headset to form a complete adjustment strategy library.

10. The intelligent control method for TWS earphone heart rate detection and adaptive adjustment of earphone function according to claim 1, characterized in that: The S6 specifically includes: S61: the headphone main control chip receives the joint control instruction set output by S56, and extracts the noise reduction parameter instruction, the audio equalization instruction and the ear canal pressure adjustment instruction according to the data frame structure of the instruction set; S62: the main control chip parses the extracted noise reduction parameter instruction into a specific noise reduction frequency band attenuation value, generates a corresponding frequency band attenuation control signal and sends it to the active noise reduction module, so as to drive the active noise reduction module to update the frequency band attenuation to the target value in real time; S63: The main control chip interprets the extracted audio equalization instruction into an audio frequency band gain adjustment value, and updates the audio equalization filter coefficient in the audio processing unit; S64: The main control chip interprets the extracted ear canal pressure adjustment instruction into the target ear canal contact pressure, and controls the built-in micro air pump to adjust the air intake or deflation of the inflatable airbag, and adjusts the actual contact pressure between the earphone inlet and the ear canal wall in real time until the target pressure value is reached; S65: The main control chip synchronously executes the control actions of S62, S63 and S64 with a unified internal clock reference to ensure that the adjustment actions of the active noise reduction module, the audio processing unit and the micro air pump are completed synchronously.

Images