A single accelerometer based multi-functional system

Abstract

The application discloses a multifunctional system based on a single accelerometer for a wearable device, comprising: an accelerometer module: collecting three-axis acceleration data of a user's arm; a data acquisition and preprocessing module: performing vector norm calculation and data standardization processing on the collected three-axis acceleration data, and extracting relevant features according to different functions; a threshold test module: performing threshold judgment according to the acceleration change and angle characteristics of arm movement, and filtering out invalid actions; an LDA classification module: classifying the data that has undergone preprocessing and threshold test through a linear discriminant analysis model, and identifying different arm movements; a function module: including wrist-lifting screen-on and screen-off, step counting, motion pattern recognition and sleep evaluation functions, and performing corresponding response control according to the classification results. Through the single accelerometer, multiple functions such as wrist-lifting screen-on and screen-off, step counting, motion pattern recognition and sleep evaluation are accurately realized.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of wearable devices, and relates to a multifunctional system based on a single accelerometer, further to a system for step counting, real-time wrist lifting and screen turning on and off, motion pattern recognition and sleep evaluation. BACKGROUND

[0002] With the rapid development of technology, wearable device technology has gradually entered a more efficient and intelligent stage. As typical representatives, smartwatches and smartbands have been widely used in daily life due to their portability and multifunctionality. These devices usually rely on multi-sensor systems, such as accelerometers, gyroscopes and magnetometers, to realize gesture recognition, motion pattern analysis and health monitoring and other functions.

[0003] However, although multi-sensor cooperation can significantly improve the accuracy and comprehensiveness of data, this technical solution also brings many challenges. First, the complexity of multi-sensor systems increases the difficulty of hardware design and integration, which increases the research and production cost of devices. At the same time, the simultaneous work of multiple sensors significantly increases the power consumption of the device, which poses a severe test to the endurance and user experience of the device.

[0004] In terms of function implementation, the traditional threshold method is widely used in the implementation of wrist lifting and screen turning on and off and other interactive functions. However, this method is prone to false triggering in complex actions or environments, resulting in poor user experience. In addition, existing step counting algorithms and motion pattern recognition technologies also face the problem of insufficient accuracy. Step counting algorithms often rely on simple acceleration changes or gait patterns, and use fixed thresholds or maximum value detection based on vibration amplitude to calculate the number of steps, which is difficult to adapt to changing motion states (such as slow walking, fast running), especially in low-frequency actions, which is prone to step counting errors or omissions. Motion pattern recognition technology usually relies on simple feature analysis and fixed rule classification, such as distinguishing walking, running, and stationary states by analyzing the mean and fluctuation amplitude of acceleration data. However, this kind of rule classification method has great limitations when facing diversified user behaviors, especially in scenarios where multiple motion patterns appear alternately, which is prone to confusion.

[0005] Therefore, the existing wearable device technology has obvious deficiencies in the complexity of multi-sensor systems, the limitations of threshold methods, the accuracy of step counting algorithms, and the accuracy of motion pattern recognition. These challenges limit the performance and user experience of the device. SUMMARY

[0006] The purpose of the present application is to provide a multifunctional system based on a single accelerometer to solve the problems raised in the background art.

[0007] The purpose of the application is realized by the following technical solutions:

[0008] A multifunctional system based on a single accelerometer for wearable devices, comprising:

[0009] An accelerometer module: for collecting three-axis acceleration data of the user's arm;

[0010] A data acquisition and preprocessing module: for vector norm calculation, data standardization processing of the collected three-axis acceleration data, and extraction of relevant features according to different functions;

[0011] A threshold test module: for threshold judgment according to the acceleration change and angle characteristics of arm movement, and filtering out invalid actions;

[0012] An LDA classification module: for classifying the data that has been preprocessed and threshold tested through a linear discriminant analysis model, and identifying different arm movements;

[0013] A function module: including wrist-lifting screen-on and screen-off, step counting, motion pattern recognition, and sleep evaluation functions, and corresponding response control according to the classification results.

[0014] As a further improvement of an embodiment of the application, the data acquisition and preprocessing module includes a data processing module that collects three-axis acceleration data and calculates the vector norm of three-axis acceleration in each time window, then normalizes the collected data to ensure that data of different dimensions can be compared within the same range, and then extracts features from the data that has been standardized in each time window.

[0015] As a further improvement of an embodiment of the application, the threshold test module includes a data window for time threshold testing, and the data after feature extraction is filtered through the data window to obtain valid data, and the acceleration change of the Z-axis and the inclination change of the Y-axis are detected for the valid data to preliminarily judge the lifting and lowering actions of the arm.

[0016] As a further improvement of an embodiment of the application, the LDA classification module is based on multi-dimensional features in the acceleration data, and generates a classification model through linear discriminant analysis model classification training according to the feature vectors of different actions, for real-time prediction of the user's arm movements.

[0017] As a further improvement of an embodiment of the application, the function module determines whether the current arm movement is wrist-lifting or wrist-lowering to determine the screen-on or screen-off control through the LDA model, specifically including the following steps:

[0018] Step 1.1, the smart watch uses a three-axis accelerometer with a frequency of 100Hz;

[0019] Step 1.2, the accelerometer collects 100 times of three-axis acceleration data of the arm per second, which is transmitted to the data processing module in real time to obtain the vector norm of acceleration through vector norm calculation, and the normalized feature value is obtained through the normalization processing formula as follows:

[0020]

[0021] Wherein, feature i is the i-th original feature, MEAN i is the mean value of feature i, SCALE i is the standard deviation of feature i, scaled_feature i is the normalized feature value.

[0022] Step 1.3, in each time window, the angle and acceleration features of the arm are extracted; when the Z-axis acceleration is negative and the Y-axis inclination angle is less than 0.6, it is judged as a wrist lifting action.

[0023] Step 1.4, the features extracted in step 1.3 are input into the LDA classification module, and it is judged whether it is a wrist lifting screen-on action or a wrist lowering screen-off action according to the LDA classification result.

[0024] As a further improvement of an embodiment of the application, the function module identifies whether the current is a gait action through an LDA model, specifically including the following steps:

[0025] Step 2.1, the smart watch adopts a three-axis accelerometer with a frequency of 100Hz;

[0026] Step 2.2, collect three-axis acceleration data and transmit to the data processing module in real time, analyze the acceleration signal through fast Fourier transform, and extract frequency features;

[0027] Step 2.3, input the frequency features extracted in step 2.2 into the LDA classification module to identify whether the current is a gait action;

[0028] Step 2.4, after identifying the gait action, the peak value of each step is identified through a local maximum value detection algorithm, and after removing outliers, frequency domain analysis is performed for anti-interference processing, and moving average filtering is used to distinguish running and walking.

[0029] As a further improvement of an embodiment of the application, the function module judges the motion mode classification step as follows:

[0030] Step 3.1, the smart watch adopts a three-axis accelerometer with a frequency of 100Hz;

[0031] Step 3.2, collect triaxial acceleration data every three seconds, and extract features from the data in each time window;

[0032] Step 3.3, normalize the features extracted in step 3.2, the formula is as follows:

[0033]

[0034] Where y represents the extracted features; through normalization processing to ensure that the feature values are uniformly processed in different dimensions; then the extracted features are input into the LDA model to optimize the classification accuracy, the formula is as follows:

[0035]

[0036] Where W is the coefficient matrix, X is the motion feature value, and L is the intercept vector;

[0037] Step 3.4, by judging each element of each output value, greater than 0 is considered to be positively correlated, less than 0 is negatively correlated, and combined with the determination rule of motion mode to screen; based on the continuous 20 times statistical output to judge the current motion state to realize mode switching and response.

[0038] As a further improvement of an embodiment of the application, wherein the determination rule of the motion mode is as follows: when more than 2 positive correlations or all negative correlation values are less than a set threshold, the output is excluded from identification, but its statistics are still accumulated.

[0039] As a further improvement of an embodiment of the application, wherein the function module performs the sleep evaluation function step as follows:

[0040] Step 4.1, the smart watch uses an accelerometer to monitor the user's night movement, and the sampling frequency of the sensor is set to 50Hz;

[0041] Step 4.2, collect triaxial acceleration data of the user's arm every 30 seconds, monitor the arm movement and movement frequency;

[0042] Step 4.3, average square wavelength and standard deviation calculation is performed on the low frequency motion features extracted in step 3.2, and then by analyzing the motion amplitude, frequency and duration, it is judged whether the user is in deep sleep, light sleep or wake state; wherein the formula of average square wavelength is as follows:

[0043]

[0044] And the standard deviation formula is as follows:

[0045]

[0046] Where yk representative acceleration data, n is the total number of data in the sliding window;

[0047] Step 4.4, record and analyze the data obtained in step 4.3 to generate a sleep report.

[0048] As a further improvement of an embodiment of the present application, in step 4.3, the criteria for threshold judgment are as follows: according to the extracted low-frequency motion features and the set threshold, different thresholds are set according to the square wavelength and s, and a score score of the current motion intensity is obtained, and the average value of score is used to judge whether it is light sleep, deep sleep or wake state.

[0049] The above technical solution has the following beneficial effects: the core innovation of the patent is to accurately realize multiple functions such as wrist lifting screen-on and screen-off, step counting, motion mode recognition, and sleep evaluation by using a single accelerometer, combining advanced signal processing and existing machine learning technologies such as linear discriminant analysis (LDA) model. This method not only simplifies the hardware structure, significantly reduces power consumption and system complexity, but also optimizes the algorithm to ensure efficient and accurate motion recognition and state judgment under limited sensor input.

[0050] Compared with the prior art which relies on multiple sensors, the patent has the outstanding advantage of efficient processing and classification ability of limited input data. Through feature extraction and machine learning model, it ensures more reliable motion recognition and state monitoring in complex environments and different user behavior patterns. This technology not only improves the user experience of the device, but also makes the device more suitable for portable and low-power application scenarios, greatly promoting the application of single-sensor devices in the field of smart wearables. BRIEF DESCRIPTION OF DRAWINGS

[0051] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only exemplary, and for those skilled in the art, other drawings can be derived from the provided drawings without creative labor.

[0052] The structures, proportions, sizes, etc. shown in the specification are only used to cooperate with the content disclosed in the specification, to be understood and read by those skilled in the art, and do not define the limiting conditions for the implementation of the present application, so they do not have technical significance. Any modification of structure, change of proportion relationship or adjustment of size, without affecting the effect and purpose that the present application can produce, should still fall within the scope of the technical content disclosed by the present application.

[0053] Figure 1 The system architecture structure schematic diagram provided by the present application is shown in the figure. DETAILED DESCRIPTION

[0054] It should be noted that the embodiments and features in the embodiments in the present application can be combined with each other without conflict. The present application will be described in detail below with reference to the drawings and in combination with embodiments.

[0055] It should be noted that, unless otherwise specified, all technical and scientific terms used in the present application have the same meaning as that generally understood by those skilled in the art to which the present application belongs.

[0056] In the present application, unless otherwise specified, the orientation words such as "upper", "lower", "top", "bottom" are generally directed to the directions shown in the drawings, or are directed to the vertical, perpendicular or gravity directions of the components themselves; similarly, for the convenience of understanding and description, "inner" and "outer" refer to the inner and outer relative to the contour of the components themselves, but the above orientation words are not used to limit the present application.

[0057] A multifunctional system based on a single accelerometer, through a single accelerometer, realizes multiple functions such as wrist lifting screen-on and screen-off, step counting, motion mode recognition and sleep evaluation. The scheme is designed for wearable devices, has the advantages of simple structure, low power consumption and high precision, and is suitable for long-term daily use. The technical scheme of the present application and its implementation steps will be described in detail below.

[0058] The system architecture of the present application is shown in the figure, mainly including the following modules: Figure 1

[0059] Accelerometer module 10: collect three-axis acceleration data of the user's arm as the main input source.

[0060] Data acquisition and preprocessing module 20: perform vector norm calculation and data standardization processing on the collected three-axis acceleration data, and extract relevant features according to different functions.

[0061] Threshold test module 30: perform threshold judgment according to the acceleration change and angle characteristics of arm movement, and filter out invalid actions.

[0062] LDA classification module 40: classify the data that has passed the preprocessing and threshold test through a linear discriminant analysis model, and recognize different arm movements such as wrist lifting, wrist lowering, gait and motion mode.

[0063] Function module 50: includes the functions of wrist lifting screen-on and screen-off, step counting, motion mode recognition and sleep evaluation, and performs corresponding response control according to the classification results.

[0064] ​The data acquisition and preprocessing module 20 in this embodiment is introduced as follows:

[0065] The accelerometer is installed on the wristwatch device and is responsible for collecting acceleration data in three-axis directions. It samples 100 times per second and transmits the data to the upper processing module (data processing module).

[0066] Vector norm calculation: In each time window, the system calculates the vector norm of three-axis acceleration to reflect the overall motion intensity of the arm.

[0067] Data standardization: The collected data is normalized to ensure that data of different dimensions can be compared within the same range.

[0068] Feature extraction: In each time window, 9 features are extracted, including mean, variance, root mean square, peak value, direction change, spectral energy, wavelength, waveform coefficient, and average absolute value of acceleration, for subsequent LDA classification.

[0069] The threshold test module 30 in this embodiment is introduced as follows:

[0070] Angle and acceleration threshold judgment: By detecting the acceleration change of the Z-axis and the inclination change of the Y-axis, the system preliminarily judges the lifting and lowering actions of the arm. For example, when the Z-axis acceleration is less than 0 and the Y-axis inclination absolute value is less than 0.6, it is determined as a wrist lifting action.

[0071] Time threshold test: To prevent frequent screen-on and screen-off, the system sets a time threshold, with at least 15 data windows (about 1.5 seconds) between each action. Only with continuous valid data, the corresponding operation will be triggered.

[0072] The LDA classification module 40 in this embodiment is introduced as follows:

[0073] LDA model training: The system classifies and trains the model through the linear discriminant analysis (LDA) model according to the multi-dimensional features in the acceleration data, generating a classification model for real-time prediction of user's arm movements. The training data of the model covers different arm movements, such as wrist lifting, wrist lowering, gait, etc.

[0074] Real-time classification: During system operation, the LDA model classifies the data in each time window, and according to the feature vectors of different actions, it is classified into corresponding action categories, such as wrist lifting, gait or static.

[0075] Linear discriminant analysis (English: Linear Discriminant Analysis, abbreviated: LDA) in this embodiment is a widely used data analysis method, commonly used for feature dimension reduction and classification problems.

[0076] The introduction of the functional module 50 in this embodiment is as follows:

[0077] 1. Wrist-lifting and screen-off function

[0078] Screen-on control: When the LDA model judges that the current arm movement is a wrist-lifting movement, the system confirms the effectiveness of the movement through threshold testing and triggers the screen to turn on.

[0079] Screen-off control: When a wrist-lowering movement is detected (LDA classification result is wrist-lowering category) and passes the threshold test, the system will trigger the screen to turn off. By setting a time threshold, the system prevents frequent screen-on and screen-off operations and saves power.

[0080] 2. Step counting function

[0081] Gait signal detection: The system identifies gait signals through frequency domain analysis of acceleration signals. FFT (Fast Fourier Transform) is used to extract gait features such as step frequency and amplitude.

[0082] LDA gait classification: The system inputs gait data into the LDA model and classifies the movement as gait or other movements such as running, walking, etc. through feature analysis.

[0083] Step count calculation: In combination with the local maximum value detection algorithm, the system calculates the number of steps in real time by analyzing the step length changes when identifying gait. For non-gait signals (such as stationary or interference movements), the system filters them out to avoid false counting.

[0084] 3. Motion pattern recognition function

[0085] Motion feature extraction: The system extracts features such as walking, running, and standing through acceleration data under different motion patterns. The extracted features include acceleration mean, standard deviation, direction change, etc.

[0086] LDA classifier: The LDA model is used to classify the extracted features, and the system can accurately identify different motion patterns and perform corresponding processing based on the classification results, such as recording motion data or prompting the user's current motion state.

[0087] 4. Sleep evaluation function

[0088] Low-frequency motion feature analysis: At night, the system analyzes the low-frequency motion features of the arm to determine the user's sleep state. Sleep states are divided into deep sleep, light sleep, and wakefulness. Based on the smoothness and frequency change of acceleration data, the system can distinguish between these states.

[0089] Time threshold: To avoid false positives in a short period of time, the system analyzes the state of multiple consecutive data windows to ensure the smoothness of state changes and generates a detailed sleep report.

[0090] In summary, the system architecture of the present application has the following functions:

[0091] Data acquisition: The accelerometer module collects real-time three-axis acceleration data.

[0092] Data preprocessing: The system performs vector norm calculation and standardization processing on the data, and extracts features.

[0093] Threshold judgment and LDA classification: Through LDA model and threshold test, the system identifies the user's arm movement.

[0094] Function implementation: According to the classification result, the system triggers the wrist lifting screen brightening, step counting, motion mode recognition or sleep evaluation function respectively.

[0095] The technical solution proposed by the present application provides an intelligent watch system integrating step counting, wrist lifting screen brightening and screen off, motion mode recognition and sleep evaluation functions through a single accelerometer combined with signal processing and linear discriminant analysis model (LDA model).

[0096] Compared with the prior art, the present application has significant technical progress in many aspects, and brings the following beneficial effects:

[0097] 1. Reduce system complexity and power consumption, improve device endurance

[0098] The prior art generally relies on a multi-sensor system, increasing hardware complexity and energy consumption, affecting the portability and user experience of the device. The present application realizes multi-functional application by using only a single accelerometer combined with optimized signal processing technology. Since multiple sensors no longer need to work simultaneously, the power consumption of the system is significantly reduced, and the device can maintain efficient operation for a longer period of time, improving the user's endurance experience. In addition, the hardware structure of the system is more simple, reducing the design and manufacturing cost, so that the device has significant advantages in portability and performance.

[0099] 2. Improve the accuracy of wrist lifting screen brightening and screen off function, reduce false triggering

[0100] The existing wrist lifting screen brightening and screen off function mostly relies on simple threshold method, which is easy to be triggered in complex motion scenarios, causing the screen to frequently brighten and dim. The present application can accurately identify effective wrist lifting motion by combining LDA model and multi-dimensional feature extraction, and significantly reduce false triggering by time and angle threshold test. Since the system can adapt to the motion changes of different users, users will not frequently consume the battery due to false triggering in daily use, further improving the efficiency and user experience of the device.

[0101] 3. Significantly improve the accuracy of step counting function

[0102] The step counting algorithm in the prior art is prone to step counting omission or error under different gaits (such as slow walking, fast running, etc.), especially with large step counting error in complex environments. The present application accurately identifies gait signals through LDA model classification and combines a local maximum value detection algorithm to maintain high-precision step counting in variable motion states. Whether it is slow walking, fast walking or running, the present application can ensure the accuracy of step counting through feature extraction and classification model, avoid step counting omission and error, and meet the high requirements of users for motion data.

[0103] 4. Improve the reliability and adaptability of motion pattern recognition

[0104] Most existing motion pattern recognition systems rely on rule-based classification algorithms and cannot effectively handle complex motion scenarios, easily confusing different motion patterns. The present application can accurately distinguish different motion patterns such as walking, running and standing through multi-dimensional feature analysis based on the LDA model. Since the LDA model can automatically learn the features of motion patterns from data, the system can adapt to diverse user behaviors, reducing errors and confusion in pattern recognition and improving the overall recognition accuracy and stability of the system.

[0105] 5. Optimize the accuracy and power consumption of sleep evaluation function

[0106] Existing sleep evaluation systems usually rely on multiple sensors or complex machine learning models, which have relatively high accuracy but also relatively high power consumption, especially in long-term use, where device battery consumption becomes a major problem. The present application analyzes the low-frequency motion characteristics of the accelerometer without relying on complex machine learning models to determine the user's sleep state, such as deep sleep, light sleep and wakefulness. This method not only reduces the power consumption of the system, but also ensures the accuracy of sleep evaluation, allowing the device to provide reliable sleep monitoring reports under low power consumption conditions.

[0107] 6. Improve the cost-effectiveness of the device

[0108] Since the present application achieves multiple functions through a single accelerometer, reducing the dependence on other sensors, the system hardware structure is simplified, and the design and manufacturing costs are significantly reduced. The overall cost-effectiveness of the device is greatly improved, with strong market competitiveness. In addition, the simplified design also shortens the production cycle of the device, further improving economic efficiency.

[0109] In summary, the present application optimizes hardware structure and algorithm design, significantly improving the accuracy of functions and user experience while reducing system complexity and power consumption. This technical solution not only has significant advantages in wrist lifting, screen brightness, step counting, motion pattern recognition and sleep evaluation, but also provides a new direction for the development of future smart wearable devices.

[0110] The present application provides a variety of single-accelerometer-based smartwatch implementation modes, through the organic cooperation of multiple algorithms, realizing multiple functions such as step counting, motion mode recognition, wrist-lifting screen-on and screen-off sleep monitoring. We not only designed adaptive algorithms in each functional module, but also combined anti-interference processing, feature extraction, LDA classification and statistical analysis techniques, so that the system has high reliability and accuracy in complex environments. The following examples detail the specific implementation steps and conditions of these functional modules.

[0111] Example 1: Wrist-lifting screen-on and screen-off function

[0112] Step 1.1: Hardware preparation

[0113] • Equipment: Use a smartwatch with a three-axis accelerometer.

[0114] • Installation: The accelerometer is embedded in the wristwatch, and the sampling frequency of the sensor is set to 100 Hz to ensure high-precision detection of arm movement.

[0115] Step 1.2: Data acquisition and processing

[0116] • Data acquisition: The accelerometer collects 100 three-axis acceleration data of the arm per second, which is transmitted to the data processing module in real time.

[0117] • Vector norm calculation: Calculate the vector norm of acceleration by the following formula:

[0118]

[0119] Where v is the vector, containing multiple components a x ,a y ,a z , square the sum of each component, and then take the square root of the result to get the Euclidean norm. This vector norm is used to measure the overall amplitude of arm movement.

[0120] • Standardization processing: The standardization processing formula is used to scale the data so that it is on the same scale, usually zero mean and unit variance. The mean MEAN and standard deviation SCALE arrays are trained on the training set to scale the features, the formula is as follows:

[0121]

[0122] Where feature i is the i-th original feature, MEAN i is the mean of feature i, and SCALE i is the standard deviation of feature i (or the range of features or other scaling values), scaled_featurei are the standardized feature values.

[0123] Step 1.3: Feature Extraction and Threshold Judgment

[0124] • Feature Extraction: In each time window, extract the angle and acceleration features of the arm, focusing on the change of Z-axis acceleration (consistent with the vertical direction).

[0125] • Threshold Judgment: When the Z-axis acceleration is negative and the Y-axis tilt angle is less than 0.6, it is preliminarily judged as a wrist lifting action. By setting a time threshold (such as at least 1.5 seconds between each action), avoid frequent triggering of screen brightness.

[0126] Step 1.4: LDA Classification and Control

[0127] • LDA Classification: Input the extracted features into the LDA classification module, and the classifier identifies whether it is a wrist lifting or wrist lowering action.

[0128] • Screen Brightness and Sleep Control: According to the LDA classification result, the system confirms that the wrist lifting is an effective action, triggering the screen brightness; if the classification result is wrist lowering, triggering the sleep function.

[0129] Example 2: Step Counting Function

[0130] Step 2.1: Hardware Preparation

[0131] • Equipment: Use a smart watch with an accelerometer, and the sampling frequency of the sensor remains at 100Hz.

[0132] Step 2.2: Data Collection and Signal Processing

[0133] • Data Collection: The watch collects three-axis acceleration data and transmits it to the data processing module. Through FFT (Fast Fourier Transform) analysis of the acceleration signal, extract the frequency characteristics for gait recognition.

[0134] Step 2.3: Feature Extraction and Gait Recognition

[0135] • Feature Extraction: Extract the frequency, amplitude and stride of gait signals as the basis for gait recognition.

[0136] • LDA Classification: Input the extracted features into the LDA classification module, and identify whether it is a gait action such as walking or running.

[0137] Step 2.4: Step Counting

[0138] • Local Maximum Detection: After recognizing the gait action, the system uses a local maximum detection algorithm to identify the peak value of each step. The system updates the step count in real time according to the detected step increment.

[0139] Anti-interference processing: outlier rejection, frequency domain analysis, threshold screening for irregular interference actions. Specific interference actions are filtered through three-axis acceleration feature extraction, and running and walking are distinguished through moving average filtering to ensure the accuracy of gait recognition.

[0140] Example 3: Motion pattern recognition function

[0141] Step 3.1: Hardware preparation

[0142] Device: use a smartwatch with an accelerometer, set the sensor sampling frequency to 100Hz.

[0143] Step 3.2: Data collection and processing

[0144] Data collection: continuously collect three-axis acceleration data on the watch, and extract features from the data in each time window;

[0145] Motion pattern recognition: extract three-axis acceleration signals every 3s.

[0146] Step 3.3: Motion pattern classification

[0147] Feature extraction: extract multi-dimensional features such as arm acceleration mean, direction change, standard deviation, etc. without filtering, and perform normalization:

[0148]

[0149] Where y represents the extracted features, and this processing ensures that the feature values are uniformly processed in different dimensions.

[0150] LDA classification: input the extracted features into the LDA model, through the formula:

[0151]

[0152] Where W is the coefficient matrix, X is the motion feature value, and L is the intercept vector, further optimizing the classification accuracy. Unlike traditional probability conversion, each output value is directly judged, and rule-based classification is performed.

[0153] Step 3.4: Positive and negative correlation judgment and mode switching

[0154] Adaptive judgment of positive and negative correlation: judge each element of each output value, greater than 0 is considered positive correlation, less than 0 is considered negative correlation, and combined with specific judgment rules of motion patterns to screen. For example, when more than 2 positive correlations or all negative correlation values are less than the set threshold, the output will be excluded from recognition, but its statistics will still be accumulated.

[0155] Mode Switching and Response: The system determines the current motion state based on 20 consecutive statistical outputs. This cumulative statistical approach enhances recognition stability, reduces short-term misjudgments, and improves the accuracy of the system in long-term motion recognition. The system records these changes and performs appropriate data processing, such as motion data recording and notifying the user of the current motion state.

[0156] Example 4: Sleep Assessment Function

[0157] Step 4.1: Hardware Preparation

[0158] Device: Uses an accelerometer to monitor the user's nighttime movements, with the sensor sampling frequency set to 50 Hz.

[0159] Step 4.2: Data collection and processing

[0160] Data Collection: During the night, the watch continuously collects triaxial acceleration data from the user's arm, primarily monitoring micro-movements and movement frequency. Sleep Assessment: The triaxial acceleration signal is extracted every 30 seconds.

[0161] Step 4.3: Feature extraction and sleep state judgment

[0162] Feature extraction: Extract low-frequency motion features of the arm, including the average square wavelength:

[0163]

[0164] And the standard deviation:

[0165]

[0166] Among them, y k Represents acceleration data, and n is the total number of data points within the sliding window. By analyzing the amplitude, frequency, and duration of movement, it determines whether the user is in deep sleep, light sleep, or awake.

[0167] Threshold judgment: The system determines the user's sleep stage based on the extracted low-frequency motion features and the set threshold. Combining squarewavelength and s, different thresholds are set to obtain a score for the current motion intensity. If the average score over a long period of time is lower than threshold 1, the user is considered to have fallen asleep, and the default judgment for falling asleep is light sleep. After a specified period of light sleep, if the average score over a period of time is lower than threshold 2, the user is considered to have entered deep sleep. After a period of deep sleep, if the average score over a period of time is greater than threshold 3, the user is considered to have entered light sleep. If the average score over a period of time is greater than threshold 4, the user is considered to have entered light sleep.

[0168] Step 4.4: Sleep report generation

[0169] • Data recording and analysis: The system records the user's overnight sleep status and generates a detailed sleep report, including deep sleep, light sleep, and wake-up time. The report can be used to analyze the user's sleep quality and provide personalized health recommendations.

[0170] Algorithm coordination and technical effect summary

[0171] The above four embodiments not only realize functions through independent algorithms respectively, but also embody the technical advantages of the invention through the cooperative work among algorithms:

[0172] 1. Anti-interference processing throughout the application: Whether it is wrist lifting, step counting, or motion pattern recognition, anti-interference processing (such as outlier rejection, frequency domain analysis, threshold judgment) is reflected in all functional modules, ensuring stability in complex environments.

[0173] 2. Unified application of LDA classification module: LDA model is widely used in wrist lifting, gait recognition, and motion pattern recognition. Through unified feature extraction and classification strategy, the system architecture is simplified, and the recognition accuracy and response speed are improved.

[0174] 3. Effective coordination of multi-level algorithms: Various algorithms such as feature extraction, frequency domain analysis, moving average filtering, and local maximum value detection work together to ensure the efficiency of data acquisition, processing, recognition, and control.

[0175] 4. Statistical-based misjudgment correction mechanism: Multiple statistics and cumulative output judgment methods (such as 20 times statistics rule) effectively reduce short-term misjudgments and ensure the stability of the system during long-term use.

[0176] Through the organic cooperation of these algorithms, the invention not only improves the multi-functional implementation capability of the smart watch, but also reduces power consumption and hardware complexity through a unified algorithm architecture, providing an ideal technical solution for multi-functional, low-power wearable devices.

[0177] The technical solution of the invention realizes multiple functions including step counting, wrist lifting, screen on and off, motion pattern recognition, and sleep evaluation through a single accelerometer, with the characteristics of simplified hardware design, low power consumption, and high precision. Through the organic cooperation of multiple algorithms such as anti-interference processing, LDA classification, feature extraction, and time threshold management, the system can operate stably in complex environments. The following are the key technical points and technical solutions that need to be protected of the invention:

[0178] I. Technical solution for realizing multiple functions based on a single accelerometer

[0179] Technical key points:

[0180] 1. Collect three-axis acceleration data using a single accelerometer, provide multiple function support (wrist-up screen-on, step counting, motion pattern recognition, sleep evaluation) through feature extraction and signal processing.

[0181] 2. Extract features from acceleration data through signal processing techniques (such as vector norm calculation, FFT frequency domain analysis, etc.).

[0182] 3. Combine linear discriminant analysis (LDA) model for action classification and recognition of multiple functions, achieve high-precision functions based on limited sensor input.

[0183] 4. Anti-interference algorithm (outlier rejection, frequency domain analysis) and feature extraction algorithm are suitable for all function modules, reducing the influence of noise and interference on system performance.

[0184] II. Precise identification and control method of wrist-up screen-on and screen-off

[0185] Technical key points:

[0186] 1. Extract Z-axis acceleration and Y-axis inclination features, combined with threshold judgment, to preliminarily identify wrist-up and wrist-down actions.

[0187] 2. Use LDA model to accurately classify wrist-up actions to ensure the accuracy of action recognition and reduce false touch phenomenon.

[0188] 3. Set a time threshold to avoid frequent switching between screen-on and screen-off.

[0189] 4. Reduce misjudgment and enhance system stability by continuously counting classification results (such as 20 times cumulative statistics).

[0190] III. High-precision step counting algorithm based on LDA model

[0191] Technical key points:

[0192] 1. Collect gait signals through the accelerometer and use FFT frequency domain analysis to extract gait features (such as step frequency, amplitude, etc.).

[0193] 2. Use LDA model to classify gait signals, combined with local maximum value detection algorithm to calculate the number of steps, reduce the influence of interference signals on step counting.

[0194] 3. Real-time gait classification and error correction to ensure the accuracy of step counting under various gaits (slow walking, fast walking, running).

[0195] 4. Moving average filter to distinguish between walking and running, combined with local maximum value detection algorithm to improve the accuracy of step counting.

[0196] IV. Motion pattern recognition method

[0197] Key technical points:

[0198] 1. Extract multi-dimensional features in the accelerometer (such as mean, standard deviation, direction change, etc.), combine with LDA model for motion pattern classification, distinguish walking, running, static and other motion states.

[0199] 2. Dynamically adapt to different users' motion behavior patterns, ensure accurate identification of different motion states in complex motion scenarios.

[0200] 3. Without using traditional probability conversion steps, directly judge each element of each output, positive and negative correlation combined with specific rules for classification. This method reduces computational complexity compared to probability conversion, improves real-time response speed and recognition accuracy.

[0201] 4. Based on LDA model motion classification and positive and negative correlation judgment, through continuous multiple statistical output, reduce short-term misjudgment, improve the stability of identification.

[0202] Five. Low-power sleep evaluation function

[0203] Key technical points:

[0204] 1. Extract small arm movement features through low-frequency acceleration signals to determine the user's sleep state (deep sleep, light sleep, wakefulness).

[0205] 2. Without relying on complex machine learning algorithms, set threshold and time window to accurately determine different sleep stages and generate sleep report.

[0206] 3. Low-power algorithm ensures long-term continuous monitoring, realizes efficient sleep evaluation and health advice generation.

[0207] Six. Time threshold and anti-misfire mechanism

[0208] Key technical points:

[0209] 1. By setting time threshold, avoid frequent triggering in various functions (such as lifting wrist to bright screen, screen off, step calculation), ensure system stability.

[0210] 2. Combine motion classification and time window management to prevent system from overreacting to short or invalid user actions, reduce misoperation.

[0211] 3. Time threshold and motion classification combined, ensure that misfire is reduced while improving system response speed and stability.

[0212] Seven. Integration and implementation of the overall system

[0213] Key technical points:

[0214] 1. The overall system integrates a single accelerometer, a multifunction signal processing module, an LDA classification model, and control logic, forming a multifunction wearable device with a simplified hardware structure.

[0215] 2. The system architecture design combines the cooperative work of multiple functions, reduces the dependence on multiple sensors, and achieves a balance between function and performance.

[0216] 3. By sharing accelerometer data, efficient data utilization and flexible switching between different functional modules are achieved, improving the overall performance of the system.

[0217] Conclusion

[0218] The technical key points and protection points of the present application mainly focus on using a single accelerometer to achieve multifunctional applications such as step counting, wrist lifting screen-on and screen-off, motion pattern recognition, and sleep evaluation through signal processing and LDA model. By simplifying the hardware structure, reducing power consumption, introducing anti-interference processing and feature extraction strategies, and combining multi-dimensional feature extraction and machine learning classification model, a portable smart watch system with high precision, low power consumption, and intelligence is provided.

[0219] Obviously, the above-described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should fall within the scope of protection of the present application.

[0220] It should be noted that the terms used herein are only intended to describe specific embodiments, and are not intended to limit the exemplary embodiments according to the present application. As used herein, the singular form is intended to include the plural form unless the context clearly indicates otherwise, and it should be understood that when the terms "comprise" and / or "include" are used in the specification, they indicate the presence of the features, steps, operations, devices, components, and / or combinations thereof.

[0221] It should be noted that the terms "first", "second", and the like used in the specification and claims of the present application and the above-described drawings are used to distinguish similar objects, and do not necessarily indicate a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein.

[0222] The above only describes the preferred embodiments of the present application and is not intended to limit the present application. For those skilled in the art, the present application can have various modifications and changes. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A single-accelerometer-based multi-functional system for wearable devices, characterized by, The application relates to a smart watch with a function of recognizing and responding to different arm movements, and belongs to the field of wearable devices. The application comprises: An accelerometer module: used for collecting three-axis acceleration data of a user's arm; A data collection and preprocessing module: used for performing vector norm calculation and data standardization processing on the collected three-axis acceleration data, and extracting relevant features according to different functions; A threshold test module: used for performing threshold judgment according to the acceleration change and angle characteristics of arm movement, and filtering out invalid actions; An LDA classification module: used for classifying the data that has undergone preprocessing and threshold test through a linear discriminant analysis model, and identifying different arm movements; A function module: including the functions of lifting the wrist to turn on the screen and turn off the screen, counting steps, recognizing a movement mode and evaluating sleep, and responding to corresponding control according to the classification result; The function module identifies whether the current is a gait action through the LDA model, and specifically comprises the following steps: Step 2.1: the smart watch adopts a three-axis accelerometer with a frequency of 100Hz; Step 2.2: three-axis acceleration data are collected and transmitted to a data processing module in real time, the acceleration signal is analyzed through fast Fourier transform, and frequency characteristics are extracted; Step 2.3: the frequency characteristics extracted in step 2.2 are input into the LDA classification module to identify whether the current is a gait action; Step 2.4: after identifying the gait action, the peak value of each step is identified through a local maximum value detection algorithm, and after removing outliers, frequency domain analysis is performed for anti-interference processing, and a moving average filter is adopted to distinguish running and walking; The function module judges the movement mode classification as follows: Step 3.1: the smart watch adopts a three-axis accelerometer with a frequency of 100Hz; Step 3.3: Normalize the features extracted in step 3.

2. The formula is as follows: ; Step 3.2: three-axis acceleration data are collected once every three seconds, and features are extracted in each time window; ； wherein is a coefficient matrix, motion characteristic values, is an intercept vector; Wherein y represents the extracted features; the features are normalized to ensure that the feature values are uniformly processed in different dimensions; then the extracted features are input into the LDA model to optimize the classification accuracy, and the formula is as follows: Step 3.4: each element of each output value is judged, and greater than 0 is regarded as positive correlation and less than 0 is regarded as negative correlation, and the output is screened in combination with the judgment rule of the movement mode; the current movement state is judged based on the statistical output of 20 times to realize mode switching and response; 2. The multi-functional system of claim 1, wherein: The judgment rule of the movement mode is as follows: when more than 2 positive correlations or all negative correlation values are less than a set threshold value, the output is excluded and is not identified, but the statistics are still accumulated.

3. The multi-functional system of claim 1, wherein: The data collection and preprocessing module comprises a data processing module, the data processing module collects acceleration data in three-axis directions, calculates the vector norm of three-axis acceleration in each time window, then normalizes the collected data to ensure that the data in different dimensions can be compared in the same range, and then extracts features from the data standardized in each time window. The threshold test module comprises a data window for time threshold test, valid data are obtained after the data after feature extraction are filtered through the data window, and the acceleration change of the Z-axis and the inclination change of the Y-axis are detected to realize preliminary judgment of the lifting and lowering actions of the arm.

4. The multi-functional system of claim 1, wherein: The LDA classification module is based on multi-dimensional features in acceleration data, according to the feature vectors of different actions, through linear discriminant analysis model for classification training, generate classification model, for real-time prediction of user's arm movement.

5. The multi-functional system of claim 1, wherein: The function module judges whether the current arm movement is wrist lifting or wrist lowering to judge the screen-on or screen-off control through the LDA model, and specifically includes the following steps: Step 1.1, the smart watch adopts a three-axis accelerometer with a frequency of 100Hz; Step 1.2, the accelerometer collects 100 times of three-axis acceleration data of the arm per second, which is transmitted to the data processing module in real time to obtain the vector norm of acceleration through vector norm calculation, and the normalized characteristic value is obtained through the normalization processing formula as follows: ; wherein is the first original feature, is the mean of the feature values, is the standard deviation of the feature values, is the normalized feature value; Step 1.3, in each time window, the angle and acceleration features of the arm are extracted; when the Z-axis acceleration is negative and the Y-axis inclination is less than 0.6, the wrist lifting action is judged; Step 1.4, input the features extracted in step 1.3 into the LDA classification module, and judge whether it is a wrist lifting screen-on action or a wrist lowering screen-off action according to the LDA classification result.

6. The multi-functional system of claim 1, wherein: The function module performs the sleep evaluation function as follows: Step 4.1, the smart watch uses an accelerometer to monitor the user's night movement, and the sampling frequency of the sensor is set to 50Hz; Step 4.2, collect the three-axis acceleration data of the user's arm every 30 seconds, and monitor the movement and movement frequency of the arm; Step 4.3, average square wavelength and standard deviation calculation are performed on the low frequency motion features extracted in step 3.2, and then by analyzing the motion amplitude, frequency and duration, it is judged whether the user is in deep sleep, light sleep or wake state; wherein the formula of average square wavelength is as follows: ; and the standard deviation formula is as follows: ; wherein, represents acceleration data, is the total number of data in the sliding window; Step 4.4, record and analyze the data obtained in step 4.3 to generate a sleep report.

7. The multi-functional system of claim 6, wherein: The threshold judgment standard in step 4.3 is as follows: according to the extracted low-frequency motion features and the set threshold, wherein different thresholds are set according to the square wavelength and s, the score of the current motion intensity is obtained, and the average value of the score is used to judge the light sleep, deep sleep or wake state.

Images