A sensor error calibration method and system for a smart watch

Abstract

The application provides a sensor error calibration method and system of a smart watch, which is applied to the field of data processing; the application realizes real-time monitoring of a sensor by the smart watch, collects trend sensing data, detects whether the data deviates from the trend based on a preset standard, thereby improving the accuracy of abnormal detection of the sensor, when the data deviates, identifies abnormal values by standard deviation calculation and classifies them, including short-term sudden abnormality, long-term drift and outliers, to accurately analyze the error source, then judges whether the abnormal values can be repaired, if yes, compensates by linear, nonlinear or periodic drift according to the abnormal type, and dynamically adjusts the sensing error coefficient in combination with environmental factors, realizes self-adaptive calibration, can effectively detect sensor drift, prevents data error accumulation caused by long-term use, ensures that the smart watch can maintain high-precision measurement in various environments, and improves user experience and equipment reliability.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of data processing, in particular to a sensor error calibration method and system for a smart watch. BACKGROUND

[0002] The sensors in the smart watch (such as accelerometers, gyroscopes, heart rate sensors, temperature and humidity sensors, etc.) are used to detect and record various physiological and motion data of the user. These sensors capture physical signals and convert them into electrical signals, which are then processed by the processing unit of the smart watch and ultimately provide feedback to the user.

[0003] However, over time and with changes in the use environment, the performance of the sensors may change, which is usually manifested as "drift" phenomenon, because the micro components inside the sensor may be physically damaged or aged, causing the output value to drift, for example, the drift of the accelerometer will cause the inaccuracy of the direction calculation, and the drift of the accelerometer will cause the error of the gait monitoring. SUMMARY

[0004] The present application aims to solve the problem of how to effectively detect sensor drift and compensate in time to ensure that the data is accurate after a long time of use, and provides a sensor error calibration method and system for a smart watch.

[0005] To solve the technical problems, the present application adopts the following technical means:

[0006] The present application provides a sensor error calibration method for a smart watch, comprising:

[0007] Based on the preset use time of the smart watch, the preset sensors are monitored in real time by the smart watch, and corresponding trend sensor data is collected;

[0008] Determine whether the trend sensor data presents a preset deviation trend;

[0009] If yes, select a preset data point from the trend sensor data, calculate the standard deviation of the data point, mark the data point that exceeds the standard deviation range as an outlier in the trend sensor data, and classify the outlier, wherein the outlier classification specifically includes short-term sudden abnormality, long-term drift and outlier;

[0010] Determine whether the outlier can be repaired;

[0011] If yes, drift compensation is performed on the abnormal values according to the abnormal classification, the sensing error coefficient of the smart watch is dynamically calibrated, the environment factors pre-collected by the smart watch are taken as compensation input of the sensing error coefficient, and the sensing error coefficient is adaptively corrected, wherein the drift compensation specifically includes linear drift compensation, nonlinear drift compensation and periodic drift compensation, and the environment factors specifically include temperature, humidity and air pressure.

[0012] Further, the step of marking the data points beyond the standard deviation range as abnormal values in the trend sensing data further includes:

[0013] Based on the number of abnormal values, the proportion data of the number of abnormal values in the trend sensing data is collected, and the distribution information of the abnormal values is identified according to the proportion data;

[0014] It is judged whether the distribution information matches a preset specific condition, wherein the specific condition specifically includes that abnormal values are concentrated in abnormal fluctuations caused by a specific time period or a specific environment factor;

[0015] If yes, the working environment of the sensor is detected, the precision error of the sensor is calculated according to the working environment, and the corresponding abnormal rule is obtained from the distribution information.

[0016] Further, the step of classifying the abnormal values further includes:

[0017] Data abnormal features of the abnormal values are extracted, and output data of different sensors in the smart watch are compared, wherein the data abnormal features specifically include fluctuation amplitude, change speed and change direction;

[0018] It is judged whether the output data detects a preset associated abnormality;

[0019] If yes, the communication protocol interface between the different sensors is identified, the bus communication state of the smart watch is obtained based on the communication protocol interface, and the network communication quality of the smart watch is generated in real time according to the bus communication state, wherein the bus communication state specifically includes noise interference, data transmission delay and connection loss, and the network communication quality specifically is the communication quality when data transmission is performed between different sensors.

[0020] Further, the step of taking the environment factors pre-collected by the smart watch as compensation input of the sensing error coefficient further includes:

[0021] Based on the fluctuation change of the environment parameter, the compensation demand of the sensing error coefficient is identified, wherein the fluctuation change specifically includes temperature and humidity fluctuation range, air pressure change and electromagnetic interference;

[0022] determining whether the compensation requirement matches a preset compensation strategy;

[0023] If not, obtaining a use scenario of the smart watch by a user, adaptively optimizing the compensation strategy according to the use scenario, and dynamically adjusting an influence coefficient of a single sensor on the compensation strategy when an error occurs in the single sensor according to a preset complementarity between different sensors, wherein the use scenario specifically includes a sports scenario, an office life scenario, and an extreme weather scenario.

[0024] Further, the step of determining whether the trend sensor data presents a preset deviating trend further includes:

[0025] detecting a corresponding instantaneous abnormal event from the trend sensor data based on a preset trend period, wherein the trend period specifically includes a global trend and a local trend;

[0026] determining whether the instantaneous abnormal event presents a preset periodic fluctuation;

[0027] If yes, obtaining regular change information corresponding to the trend sensor data through the instantaneous abnormal event, and constructing a trend change point corresponding to the periodic fluctuation according to the regular change information.

[0028] Further, the step of determining whether the abnormal value can be repaired further includes:

[0029] calculating a data missing rate of a time window corresponding to the abnormal value based on the time window;

[0030] determining whether the data missing rate exceeds a preset missing value;

[0031] If yes, identifying influence content of missing data on final sensor data through the abnormal value, guiding a user to re-measure the final sensor data according to the influence content, and recording a corresponding abnormal reason in the smart watch according to an abnormal data segment of the abnormal value, wherein the abnormal reason specifically includes a sensor falling off, signal loss, and external environmental factors.

[0032] Further, the step of collecting corresponding trend sensor data by real-time monitoring of a preset sensor in the smart watch based on a preset use duration of the smart watch further includes:

[0033] detecting a time window in which an abnormal value occurs based on a preset statistical characteristic of sensor data, wherein the statistical characteristic specifically includes a mean value, a variance, and a standard deviation;

[0034] determining whether a number of the time windows reaches a preset threshold.

[0035] If yes, according to the use duration, a self-calibration mode of the sensor is triggered periodically, through the self-calibration mode, the error accumulation of the sensor is reduced, a calibration tutorial of the smart watch is activated, and the user is guided to calibrate the smart watch manually according to the calibration tutorial.

[0036] The application further provides a sensor error calibration system of a smart watch, comprising:

[0037] The acquisition module is configured to acquire corresponding trend sensor data by monitoring the preset sensor of the smart watch in real time based on the use duration of the smart watch.

[0038] The judgment module is configured to judge whether the trend sensor data presents a preset deviation trend.

[0039] The execution module is configured to select a preset data point from the trend sensor data, calculate the standard deviation of the data point, mark the data point that exceeds the standard deviation range as an outlier in the trend sensor data, and perform an abnormality classification on the outlier, if yes.

[0040] The second judgment module is configured to judge whether the outlier can be repaired.

[0041] The second execution module is configured to perform drift compensation on the outlier according to the abnormality classification, dynamically calibrate the sensor error coefficient of the smart watch, use the environment factors pre-acquired by the smart watch as the compensation input of the sensor error coefficient, and adaptively correct the sensor error coefficient, if yes.

[0042] Further, the execution module further comprises:

[0043] The identification unit is configured to acquire proportion data of the number of outliers in the trend sensor data based on the number of outliers, and identify distribution information of the outliers according to the proportion data.

[0044] The judgment unit is configured to judge whether the distribution information matches a preset specific condition, wherein the specific condition specifically includes that the outliers are concentrated in abnormal fluctuations caused by a specific time period or a specific environment factor.

[0045] The execution unit is configured to detect the working environment of the sensor, calculate the precision error of the sensor according to the working environment, and obtain corresponding abnormal rules from the distribution information, if yes.

[0046] Further comprising:

[0047] The extraction module is configured to extract data anomaly features of the abnormal values, and compare output data of different sensors in the smart watch, wherein the data anomaly features specifically include fluctuation amplitude, change speed and change direction.

[0048] The third judgment module is configured to judge whether the output data detects a preset accompanying anomaly.

[0049] The third execution module is configured to, if yes, identify a communication protocol interface between the different sensors, acquire a bus communication state of the smart watch based on the communication protocol interface, and generate a network communication quality of the smart watch in real time according to the bus communication state, wherein the bus communication state specifically includes noise interference, data transmission delay and connection loss, and the network communication quality specifically is a communication quality when data transmission is performed between the different sensors.

[0050] The present application provides a sensor error calibration method and system for a smart watch, which has the following beneficial effects:

[0051] The present application can improve the accuracy of sensor anomaly detection by real-time monitoring of the sensors by the smart watch, collecting trend sensor data, and detecting whether the data deviates from the trend based on a preset standard. When the data deviates, the abnormal values are identified by standard deviation calculation and classified into short-term sudden anomalies, long-term drifts and outliers to accurately analyze the error sources. Then, it is judged whether the abnormal values are repairable. If they are repairable, linear, nonlinear or periodic drift compensation is adopted according to the abnormal type, and the sensor error coefficient is dynamically adjusted in combination with environmental factors (such as temperature, humidity and air pressure) to realize adaptive calibration. This can effectively detect sensor drift, prevent data error accumulation caused by long-term use, ensure that the smart watch can maintain high-precision measurement in various environments, and improve user experience and device reliability. BRIEF DESCRIPTION OF DRAWINGS

[0052] Figure 1 It is a flowchart of an embodiment of the sensor error calibration method for the smart watch of the present application.

[0053] Figure 2 It is a structural block diagram of an embodiment of the sensor error calibration system for the smart watch of the present application. DETAILED DESCRIPTION

[0054] It should be understood that the specific embodiments described herein are only used to explain the present application, and are not used to limit the purpose, functional features and advantages of the present application. The present application will be further described with reference to the embodiments and the accompanying drawings.

[0055] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only 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 work fall within the protection scope of the present application.

[0056] Reference is made to the accompanying drawings Figure 1 For the sensor error calibration method of the smart watch in an embodiment of the present application, comprising:

[0057] S1: based on the preset use time of the smart watch, the preset sensor is monitored in real time by the smart watch, and corresponding trend sensing data is collected;

[0058] S2: judging whether the trend sensing data presents a preset deviation trend;

[0059] S3: if yes, selecting a preset data point from the trend sensing data, calculating the standard deviation of the data point, marking the data point that exceeds the standard deviation range as an outlier in the trend sensing data, and classifying the outlier, wherein the outlier classification specifically includes short-term sudden abnormality, long-term drift and outlier;

[0060] S4: judging whether the outlier can be repaired;

[0061] S5: if yes, according to the outlier classification, performing drift compensation on the outlier, dynamically calibrating the sensor error coefficient of the smart watch, taking the environment factors pre-collected by the smart watch as the compensation input of the sensor error coefficient, and adaptively correcting the sensor error coefficient, wherein the drift compensation specifically includes linear drift compensation, nonlinear drift compensation and periodic drift compensation, and the environment factors specifically include temperature, humidity and air pressure.

[0062] In the embodiment, the system collects the corresponding trend sensor data by real-time monitoring of the pre-set sensor through the smart watch based on the pre-set used time length of the smart watch, and then the system judges whether the trend sensor data presents the pre-set deviation trend to execute the corresponding steps; for example, when the system determines that the trend sensor data collected by the sensor does not present the pre-set deviation trend, the system considers that the measurement accuracy and stability of the current sensor still meet the expectation, no obvious drift or abnormal situation occurs, the system continues to record and store the measurement data according to the existing sensor data processing logic, ensures the continuity and integrity of the data, maintains the existing sensor error compensation strategy, and does not perform additional compensation adjustment to avoid excessive correction affecting the measurement result, and stores the current measurement environment (such as temperature, humidity, air pressure) and device state to facilitate comparison with the current state when an abnormality is detected in the future, analyze possible influencing factors, and set a regular calibration interval; even if no drift occurs, the baseline comparison can be performed according to the pre-set period to find potential subtle drift trend in advance; for example, when the system determines that the trend sensor data collected by the sensor presents the pre-set deviation trend, the system considers that the measurement accuracy of the current sensor may have drift or abnormal situation, the system selects the pre-set data points from the trend sensor data, calculates the standard deviation of the data points, marks the data points exceeding the standard deviation range as abnormal values in the trend sensor data, classifies the abnormal values, and the abnormal classification specifically includes short-term sudden abnormality, long-term drift and outlier; the system can identify the sensor measurement error in time by judging whether the trend sensor data deviates from the set standard, prevent the error data from affecting the final measurement result, adopt the standard deviation calculation and abnormal value marking method to help exclude incidental interference and improve the stability and accuracy of the data, and divide the abnormal values into short-term sudden abnormality, long-term drift and outlier, so that the system can take corresponding processing strategies for different types of abnormalities, the short-term sudden abnormality can be processed by smoothing filtering, the long-term drift can trigger an automatic compensation mechanism, and the outlier can be used for early warning of possible hardware failure of the sensor to improve the fault management capability of the smart watch, and based on the abnormal classification, the system can adjust the sensor compensation strategy according to the abnormal type, for example, the trend compensation algorithm is adopted to correct the long-term drift, so that the sensor can maintain high-precision measurement capability even after long-time use, thereby improving the reliability and user experience of the smart watch in complex environment; then the system judges whether the abnormal values in the trend sensor data can be repaired to execute the corresponding steps;For example, when the system determines that outliers in trend sensing data cannot be repaired, it will assume that the cause is a sensor hardware failure, severe long-term drift, environmental interference exceeding the compensable range, or excessive data loss preventing effective compensation. The system will store the unrepairable data separately and mark it in the log or data record to avoid affecting the calculation of normal data. If there are few outliers, the system can use historical trend data, neighboring data interpolation, or preset default values ​​for compensation to maintain measurement continuity. If outliers persist, the system can trigger a self-check mechanism to check the sensor's operating status and send a notification to the user. If the system detects an anomaly in the trend sensing data and determines that the deviation is minor, it will perform drift compensation based on the anomaly classification. This drift compensation includes linear drift compensation, nonlinear drift compensation, and periodic drift compensation, dynamically calibrating the smartwatch's sensors. For example, if the system suggests recalibrating, checking the hardware, or replacing the device, and if the smartwatch has multiple built-in sensors, it can reduce the weight of data from the faulty sensor, prioritizing data from other sensors for calculations, or switch to a backup measurement mode (e.g., using a different algorithm or data source). For instance, if the system determines that anomalies in trend sensing data can be corrected, it will consider the data deviation minor and compensate accordingly. The system will perform drift compensation based on different anomaly classifications, including linear drift compensation, nonlinear drift compensation, and periodic drift compensation. The error coefficient is calculated by using environmental factors pre-collected by the smartwatch, specifically temperature, humidity, and air pressure, as compensation inputs. This adaptively corrects the sensor error coefficient. The system compensates for outliers by correcting for drift, allowing for timely correction even with small data deviations. This prevents erroneous data from affecting the final measurement results. Linear drift compensation is suitable for long-term stable offsets, non-linear drift compensation corrects complex error variations, and periodic drift compensation adjusts for periodically fluctuating errors, ensuring long-term data stability and accuracy. Simultaneously, using pre-collected environmental factors (such as temperature, humidity, and air pressure) as compensation inputs allows the system to dynamically adjust its error calibration strategy under different environmental conditions. This mechanism effectively reduces measurement errors caused by external environmental changes, ensuring the smartwatch maintains accurate measurement performance in various scenarios such as high temperature, low temperature, and high humidity. Furthermore, the adaptive correction mechanism allows the sensor error coefficient to dynamically adjust with real-time data, preventing the sensor from accumulating errors over time and affecting measurement results. This not only improves long-term data stability but also reduces the need for frequent calibration or device replacement due to sensor inaccuracy, thereby extending the smartwatch's lifespan and improving user experience and device reliability.

[0063] It should be noted that preset data points are selected from the trend sensing data, the standard deviation of the data points is calculated, and data points that exceed the standard deviation range are marked as outliers in the trend sensing data. The outliers are then classified as anomalies. A specific example is as follows:

[0064] Suppose a user wears a smartwatch, and the heart rate sensor of the watch collects data every second to record the user's heart rate changes. The system sets the analysis window to the last 10 minutes of data (collecting data every second, there are 600 data points), but for the sake of illustration, we will use only 10 data points as an example.

[0065] The normal heart rate range of a user in a quiet state is 60-80 bpm, but due to certain external or internal factors, abnormal heart rate data may occur; for example, the watch is too loose, the sensor is aging, the user suddenly exercises vigorously, or signal interference, etc., which may cause abnormal fluctuations in heart rate data.

[0066] Suppose the heart rate data (unit: bpm) is as follows:

[0067] 70, 72, 71, 73, 150, 74, 70, 72, 71, 73;

[0068] First, calculate the mean,

[0069] The mean (Mean, μ) calculation formula is:

[0070] Where ∑X represents the sum of all data points, and N represents the number of data points.

[0071] Then calculate the mean:

[0072] Then calculate the standard deviation,

[0073] The standard deviation (Standard Deviation, σ) calculation formula is:

[0074] Calculate the square of the difference between each data point and the mean:

[0075] (70-79.6)²=92.16;

[0076] (72-79.6)²=57.76;

[0077] (71-79.6)²=73.96;

[0078] (73-79.6)²=43.56;

[0079] (150-79.6)²=4920.16;

[0080] (74-79.6)²=31.36;

[0081] (70-79.6)²=92.16;

[0082] (72-79.6)2=57.76;

[0083] (71-79.6)2=73.96;

[0084] (73-79.6)2=43.56;

[0085] Sum:

[0086] 92.16+57.76+73.96+43.56+4920.16+31.36+92.16+57.76+73.96+43.56=5486.24;

[0087] Calculate the standard deviation:

[0088] Then set the threshold for outlier detection,

[0089] The set abnormal detection range is:

[0090] Then calculate the range:

[0091] Lower limit = 79.6-46.84=32.76 bpm

[0092] Upper limit = 79.6+46.84=126.44 bpm

[0093] That is, any data outside the range of 32.76-126.44 bpm is marked as an outlier;

[0094] Finally, mark the outliers, in the data set, only 150 bpm is outside the set range (greater than 126.44 bpm), so it is marked as an outlier, classify the outlier 150 bpm:

[0095] Short-term sudden anomaly: if only one data point is abnormal, and the data before and after it is normal, it means that it may be a short-term anomaly caused by movement, emotional fluctuations or signal interference; The system can record but not compensate to avoid interfering with normal trend analysis;

[0096] Long-term drift: if the next several data are continuously higher than 126.44 bpm, for example, the subsequent data become 140, 145, 150, 155, it means that the sensor may have drifted or accumulated errors, and the system needs to dynamically calibrate the sensor error coefficient;

[0097] Outlier: if the data point 150 bpm is far beyond the physiological reasonable range (such as 300 bpm or more), it may be a sensor failure or improper wearing, the system will directly ignore the data point, and may prompt the user to check the equipment;

[0098] In summary, through the above method, the system can accurately identify abnormal values, set reasonable thresholds through standard deviation to prevent false positives, and classify different types of abnormal situations, so that the system can take targeted compensation strategies to improve the reliability of measurement data. If it is a short-term sudden abnormality, the system can not compensate temporarily to avoid affecting the overall trend judgment; if it is a long-term drift, the system can dynamically calibrate the sensor error coefficient to improve data accuracy, and outliers can be directly removed to prevent false data from affecting the overall analysis results, ensuring the data quality and user experience of the smart watch.

[0099] It should be noted that according to the abnormal classification, the drift compensation is performed on the abnormal value, and the sensor error coefficient of the smart watch is dynamically calibrated, and specific examples are as follows:

[0100] Suppose a smart watch has a built-in temperature sensor to monitor the user's body temperature to provide health management recommendations; however, after a long period of use, the system finds that the sensor measures a body temperature that is 0.5°C higher in a high-temperature environment (such as hot summer) and 0.4°C lower in a low-temperature environment (such as winter outdoors); this indicates that the sensor has a long-term drift phenomenon, resulting in a large error in body temperature data, which may mislead health assessment;

[0101] First, abnormal detection and classification are performed, and the system finds that the measurement value of the temperature sensor continuously increases when the ambient temperature increases and continuously decreases when the ambient temperature decreases, and this drift trend repeats with seasonal changes; therefore, the system marks this drift as **“long-term drift”**, not an accidental short-term sudden abnormality or outlier;

[0102] Then, a drift compensation strategy is developed, and according to the characteristics of long-term drift, the system uses a periodic drift compensation method combined with environmental factors (temperature, humidity) to calibrate the error coefficient of the sensor:

[0103] Establish a temperature-error curve:

[0104] Through historical data regression analysis, the error range of the sensor under different environmental temperature conditions is calculated; for example:

[0105] At an ambient temperature of 10°C, the body temperature measurement value is 0.4°C lower

[0106] At an ambient temperature of 35°C, the body temperature measurement value is 0.5°C higher

[0107] At an ambient temperature of 22°C, the measurement error is close to 0°C (i.e., no deviation)

[0108] Adaptive correction of sensor error coefficient:

[0109] In combination with the temperature and humidity sensor built-in the smartwatch, each time the body temperature is measured, the system will:

[0110] read the current ambient temperature;

[0111] find the corresponding deviation value in the temperature-error curve;

[0112] automatically compensate the measurement result of the calibrated sensor;

[0113] For example, at 35°C, the measured body temperature of 37.5°C is automatically corrected to 37.0°C;

[0114] Finally, check the running effect, after drift compensation, the body temperature measured by the smartwatch under different ambient temperatures is closer to the true body temperature, improving the accuracy of health monitoring; Because the system can dynamically calibrate the sensor error, even after long-term use, the measurement deviation caused by sensor aging can be effectively compensated, prolonging the service life of the device;

[0115] In summary, the above content classifies abnormal values, combines historical trend analysis, and the system can use linear, nonlinear or periodic drift compensation methods to correct the sensor error coefficient; For example, for the long-term drift problem of the temperature sensor in high or low temperature environment, the system constructs a temperature-error curve to adaptively adjust the measurement data, so that the smartwatch can maintain high accuracy in body temperature measurement under different environmental conditions; This compensation mechanism improves the stability of the data and the long-term reliability of the device.

[0116] In this embodiment, the data points exceeding the standard deviation range are marked as abnormal values in the trend sensor data in step S3, and the step S3 further includes:

[0117] S31: Based on the number of abnormal values, collect the proportion data of the number of abnormal values in the trend sensor data, and according to the proportion data, identify the distribution information of the abnormal values;

[0118] S32: Determine whether the distribution information matches a preset specific condition, wherein the specific condition specifically includes that the abnormal values are concentrated in abnormal fluctuations caused by a specific time period or a specific environmental factor;

[0119] S33: If yes, detect the working environment of the sensor, calculate the accuracy error of the sensor according to the working environment, and obtain the corresponding abnormal rule from the distribution information.

[0120] In this embodiment, the system collects the proportion data of the number of abnormal values in the trend sensing data based on the number of abnormal values, identifies the distribution information of the abnormal values according to different proportion data, and then judges whether the distribution information matches the pre-set specific condition, and the specific condition specifically includes that the abnormal values are concentrated in abnormal fluctuations caused by a specific time period or a specific environmental factor, so as to execute the corresponding steps; for example, when the system determines that the distribution information of the abnormal values cannot match the pre-set specific condition, the system considers that the occurrence of the abnormal values is not caused by known patterns or predictable external environmental factors (such as a specific time period or a specific environmental factor), and appropriately increases the time window or the number of data points to improve the accuracy of system judgment, for example, if the current analysis is based on 10 minutes of data, it can be changed to 30 minutes or 1 hour of data analysis to capture a wider trend, and if the distribution of abnormal values is relatively scattered, the weight of standard deviation calculation or the threshold of abnormal value judgment may need to be adjusted to prevent over-filtering or missing abnormal values, and if the proportion of abnormal values continues to increase but the reason is unknown, the system can enter a warning mode, reduce the data reliability, or only be used for user reference, without directly affecting the key decision (such as health warning); for example, when the system determines that the distribution information of the abnormal values can match the pre-set specific condition, the system considers that the occurrence of the abnormal values is caused by known factors, detects the working environment of the sensor, calculates the precision error of the sensor according to different environmental conditions, and obtains the corresponding abnormal rules from the distribution information; the system detects the working environment of the sensor, calculates the precision error according to different environmental conditions, can dynamically adjust the calibration parameters of the sensor, for example, if the system detects that the increase of 5°C in the environmental temperature causes the measurement error to increase by 0.2%, the system can compensate for this error in real time, thereby ensuring that the measurement result is not affected by environmental fluctuations, and by analyzing the distribution rules of the abnormal values, the system can identify long-term trends and adjust the error compensation strategy, for example, if the system finds that a certain sensor measures high in a high humidity environment, it can automatically enable a specific humidity calibration algorithm, thereby maintaining the accuracy of the measurement data for a long time and prolonging the effective service life of the device, and if the system cannot determine the source of the abnormality, it may trigger the data repair or compensation mechanism, causing unnecessary calculation consumption and even data distortion, and based on the matching of the abnormal values with the known environmental factors, the system can execute compensation under the correct conditions to avoid interference with normal data, thereby improving the overall energy efficiency and reliability of the system.

[0121] It should be noted that detecting the working environment of the sensor, calculating the precision error of the sensor according to the working environment, and obtaining the corresponding abnormal rules from the distribution information are specifically as follows:

[0122] Assume that the smartwatch has a built-in barometric pressure sensor to measure the user's altitude; however, in some extreme environments (such as rainy or humid areas), the user's feedback on the altitude measurement shows abnormal drift; for example, a user is at an altitude of 100 meters, but the watch shows 120 meters, with an error of 20%, seriously affecting the watch's navigation and exercise monitoring functions;

[0123] Problem analysis, the system records the user's historical measurement data, and combines with the environmental sensor to detect the following situation:

[0124] The environmental humidity is abnormal, and through the built-in temperature and humidity sensor in the smartwatch, it is detected that the current environmental humidity is as high as 90% (usually when the humidity exceeds 85%, it may affect the reading of the barometric pressure sensor); in contrast, in a normal environment (for example, 50% humidity), the measurement error of the watch is small, and the data is stable;

[0125] Then calculate the sensor accuracy error, by analyzing the long-term trend data, the system finds that when the humidity exceeds 85%, the measured barometric pressure value is lower than the actual value, resulting in a higher altitude calculated by the watch; further analysis of historical data, the system fits the relationship between humidity and measurement error: error = 0.1 × (humidity−50%);

[0126] It is calculated that the humidity increases from 50% to 90%, resulting in an additional error of 0.1 × (90−50) = 4hPa;

[0127] Since the atmospheric pressure decreases by 1hPa, the measured altitude increases by 8.3 meters, so the 4hPa error makes the measured height 33.2 meters higher than the actual height, which is close to the user's feedback of 20% error (i.e. 20 meters);

[0128] Get the abnormal rule, by analyzing the trend of sensor data, the system finds that this abnormality only occurs when the humidity is higher than 85%, which meets the preset abnormal classification condition; the system further analyzes that the higher the humidity, the greater the measurement deviation, but this phenomenon has a stable trend and can be corrected through environmental compensation;

[0129] Then execute the compensation measures, the smartwatch identifies the abnormality and starts the humidity compensation mechanism:

[0130] According to the humidity value 90% measured by the environmental sensor, apply the compensation model: height correction = calculated height−(error × 8.3) height correction = calculated height−(error × 8.3) height correction = calculated height−(error × 8.3) where error = 4hPa, the corresponding height error is about 33.2 meters;

[0131] After compensation, the measured altitude of the watch is corrected from 120 meters to 100 meters, returning to normal;

[0132] Finally, dynamically adjust the sensor error coefficient, the system will record the current humidity information 90% to the error compensation database, and in the future similar environmental conditions, automatically apply the correction parameters, improve the compensation accuracy; if the future humidity reaches 95% or 100%, the system can predict the error may intensify in advance, and adaptively adjust the compensation parameters, further optimize the data accuracy;

[0133] That is, the smart watch can still obtain accurate altitude data in a high humidity environment, without affecting outdoor navigation, motion recording and other functions. At the same time, the compensation mechanism can adaptively correct sensor errors, avoid measurement drift accumulation caused by environmental changes, and does not require manual calibration by the user. The watch can automatically adjust the sensor data to enhance the intelligent level;

[0134] In summary, through the above content, the system detects the working environment of the sensor (such as humidity), calculates the error and extracts the abnormal rules. The system can accurately identify the source of the abnormality, and optimize the data quality through automatic compensation. This method is suitable for various environment-sensitive sensors (such as temperature, pressure, humidity, acceleration sensor, etc.), ensuring high-precision measurement under different environmental conditions, and improving the intelligent level and long-term stability of the device.

[0135] In this embodiment, before the step S3 of classifying the abnormal value, further comprising:

[0136] S301: Extract the data abnormality characteristics of the abnormal value, and compare the output data of different sensors in the smart watch, wherein the data abnormality characteristics specifically include fluctuation amplitude, change speed and change direction;

[0137] S302: Determine whether the output data detects a preset associated abnormality;

[0138] S303: If yes, identify the communication protocol interface between the different sensors, obtain the bus communication state of the smart watch based on the communication protocol interface, and generate the network communication quality of the smart watch in real time according to the bus communication state, wherein the bus communication state specifically includes noise interference, data transmission delay and connection loss, and the network communication quality specifically refers to the communication quality when different sensors transmit data.

[0139] In the present embodiment, the system extracts data anomaly features of the abnormal value, which specifically include fluctuation amplitude, change speed and change direction, compares the output data of different sensors in the smart watch, and then determines whether the output data detects the pre-set accompanying anomaly to execute corresponding steps; for example, when the system determines that the output data of different sensors in the smart watch does not detect the pre-set accompanying anomaly, the system considers that the current abnormal value is likely to be a short-time error or accidental fluctuation, rather than caused by environmental factors, equipment failure or systematic error, the system temporarily stores the current abnormal value, and continues to collect data in the subsequent time window to observe whether there is a persistent deviation; for example, if the heart rate sensor reading suddenly rises to 180 bpm in a certain measurement, but the acceleration sensor and skin resistance sensor are normal, it may be a false measurement, the system will verify whether the data returns to normal within the next 10 seconds, and if the abnormal point does not trigger the accompanying anomaly but occurs multiple times in a short time, the system will record these data and mark them as "low-priority anomaly" for review in subsequent data analysis; for example, when the system determines that the output data of different sensors in the smart watch detects the pre-set accompanying anomaly, the system considers that the current abnormal value is caused by environmental factors, equipment failure or systematic error, the system identifies the communication protocol interface between different sensors, obtains the bus communication state of the smart watch based on the communication protocol interface, and the bus communication state specifically includes noise interference, data transmission delay and connection loss, and generates the network communication quality of the smart watch in real time according to different bus communication states, and the network communication quality specifically refers to the communication quality when different sensors transmit data; by analyzing the communication protocol interface and bus communication state between different sensors, the system can determine whether the anomaly is caused by environmental factors, equipment failure or systematic error, which helps to accurately locate the source of the anomaly and avoid mistaking hardware failure or communication anomaly as sensor drift or short-time fluctuation, and can dynamically evaluate the integrity and accuracy of sensor data during transmission. If the system finds that the data is affected due to poor communication quality, it can take measures such as redundancy check, automatic retransmission or signal enhancement to ensure data stability; for example, if the data transmission of a certain sensor has high delay, the system can preferentially use other sensor data or perform data compensation to reduce the influence of the anomaly, and can automatically adjust the data processing strategy based on real-time communication quality evaluation; for example, in a high signal noise environment, the system can reduce the sensor sampling rate to reduce false positives, or enable redundant sensors for data compensation, thereby improving measurement accuracy. This adaptive optimization capability enables the smart watch to maintain high precision operation in complex environments and improve user experience.

[0140] In the embodiment, the step S5 of taking the pre-acquired environmental factors of the smart watch as the compensation input of the sensing error coefficient further includes:

[0141] S51: identifying the compensation demand of the sensing error coefficient based on the fluctuation change of the environmental parameter, wherein the fluctuation change specifically includes the temperature and humidity fluctuation range, the air pressure change and the electromagnetic interference;

[0142] S52: judging whether the compensation demand matches the preset compensation strategy;

[0143] S53: if not, acquiring the use scenario of the smart watch by the user, adaptively optimizing the compensation strategy according to the use scenario, and dynamically adjusting the influence coefficient of the compensation strategy when a single sensor appears an error according to the preset complementarity between different sensors, wherein the use scenario specifically includes the sports scenario, the office and life scenario and the extreme weather scenario.

[0144] In the embodiment, the system identifies the compensation demand of the sensing error coefficient based on the fluctuation change of the environmental parameters, and the fluctuation change specifically includes the temperature and humidity fluctuation range, the air pressure change and the electromagnetic interference. Then, the system determines whether the compensation demand matches the pre-set compensation strategy to execute the corresponding steps. For example, when the system determines that the compensation demand of the sensing error coefficient matches the pre-set compensation strategy, the system considers that the source of the current error is known, and the system has the corresponding compensation scheme to correct the error. The system calculates the optimal compensation parameter according to the matched compensation strategy and the current environmental parameter fluctuation. For example, if the temperature rise causes the sensor measurement value to be too high, the system can introduce a temperature compensation factor to dynamically adjust the measurement value. If the electromagnetic interference increases, the system can use a filtering algorithm to reduce the error or temporarily increase the signal sampling times to improve the data accuracy. At the same time, one or more of linear compensation, nonlinear compensation or periodic drift compensation is selected for adjustment. For example, it is suitable for the case that the error changes linearly with the environmental parameter, such as the error drift caused by temperature change. It is suitable for the case that the error changes nonlinearly with the environment, such as the unstable fluctuation of the sensor output in the high humidity environment. When the environmental parameter change has periodic characteristics (such as the diurnal temperature difference), the error trend is predicted through the historical data and the compensation value is dynamically adjusted. After the compensation strategy is executed, the sensing error coefficient is dynamically updated to ensure that the subsequent data acquisition meets the corrected measurement standard. For example, the relationship between the current temperature and humidity and the error is recorded to optimize the compensation model. When the electromagnetic interference is strong, the signal processing algorithm is adjusted to ensure stable data acquisition. For example, when the system determines that the compensation demand of the sensing error coefficient cannot match the pre-set compensation strategy, the system considers that the source of the current error is unknown. The system obtains the use scenarios of the smart watch, and the use scenarios specifically include the sports scenario, the office and life scenario and the extreme weather scenario. According to different use scenarios, the compensation strategy is adaptively optimized, and the influence coefficient of the compensation strategy when a single sensor error occurs is dynamically adjusted according to the complementarity between different sensors.The system can narrow down the possible causes of errors by obtaining the user's usage scenarios, such as sports, office life, or extreme weather. For example, in a sports scenario, errors may be caused by intense shaking or posture changes, while in extreme weather, errors may be affected by environmental factors such as temperature, humidity, etc. This way, the system can more accurately determine the potential source of errors rather than relying on fixed compensation modes. At the same time, according to different usage scenarios, the compensation strategy is adaptively optimized, which can make the compensation scheme more flexible. For example, in a sports scenario, the system can perform more stringent filtering on inertial sensor data to reduce errors caused by intense exercise. In the extreme weather scenario, the system can increase the weight of environmental sensor data to ensure that external environmental changes are reasonably considered. This dynamic adjustment allows the system to maintain high accuracy and reliability in different environmental conditions. According to the complementarity between different sensors, the influence of single sensor error on the overall compensation strategy can be optimized. For example, when the acceleration sensor data deviates, the system can use the gyroscope data for cross-validation to avoid excessive influence of single sensor data error on the final measurement result. Through multi-sensor data fusion, the system can more accurately calculate the error source and reduce unnecessary compensation correction, improving the stability of the overall data.

[0145] It should be noted that the usage scenario of the smart watch is obtained, and the compensation strategy is adaptively optimized according to the usage scenario. According to the preset complementarity between different sensors, the influence coefficient of a single sensor error on the compensation strategy is dynamically adjusted. Specific examples are as follows:

[0146] Suppose a user wears a smart watch for high-intensity outdoor running training. During running, the acceleration sensor of the smart watch is affected by intense shaking, causing abnormal deviations in step count, motion trajectory, and speed measurement. For example, the acceleration data fluctuates greatly in a short period of time, causing the system to misjudge the step frequency, resulting in the user's step count being calculated too high or too low. In addition, due to environmental factors such as wind speed changes and electromagnetic interference, the GPS signal is affected, and the trajectory record may drift or have breakpoints, affecting the accuracy of the motion data.

[0147] In this case, the system of the smart watch will optimize the compensation according to the following steps:

[0148] The usage scenario information is obtained, and the heart rate sensor built-in the smart watch detects that the user's heart rate has significantly increased, indicating that the user is performing high-intensity exercise. The GPS sensor data is continuously updated, showing that the user is moving quickly. Combined with the acceleration sensor data, it can be confirmed that the user is in a running scenario. Combined with the temperature and humidity sensor data, the system determines that the current environmental temperature is high and the humidity is low, and there is no extreme weather interference, ruling out the possibility of errors caused by weather factors.

[0149] Sensor complementarity adjustment. Since the acceleration sensor is greatly affected by shaking, leading to increased data noise, the system reduces the weight of the acceleration sensor in step count calculation. Gyroscope data is relatively stable and can provide angular velocity information, so the system increases the weight of the gyroscope in step count calculation and trajectory correction to optimize the accuracy of step count calculation. The GPS sensor data drifts due to signal fluctuations. After the system detects trajectory abnormalities, it combines GPS and barometer to use pressure change trends to determine height changes, reducing the impact of GPS data drift on trajectory recording.

[0150] Dynamic adjustment of compensation strategy. In the step count calculation model of the smart watch, the system reduces the dependence on the acceleration sensor for large fluctuations in a short period of time, and instead uses a gyroscope angular velocity and step length model to estimate the number of steps. The calculation of the movement trajectory uses a multi-sensor fusion algorithm to filter GPS data and correct the trajectory with gyroscope data, making the trajectory curve smoother and reducing errors caused by signal drift. In terms of speed calculation, the system estimates speed based on the step frequency detected by the gyroscope and the step length model, and compares it with the speed calculated by GPS. When the error is too large, the system dynamically adjusts the trustworthiness of GPS data to make the final calculated speed more stable and accurate.

[0151] In summary, after the above optimization and compensation strategies, the smart watch can reduce the impact of sensor errors in high-intensity running scenarios, making step count more accurate and movement trajectory more realistic. It avoids data distortion caused by single sensor abnormalities, improves user trust in exercise data and use experience.

[0152] In this embodiment, the step S2 of judging whether the trend sensor data presents a preset deviation trend further includes:

[0153] S21: Detecting a corresponding instantaneous abnormal event from the trend sensor data based on a preset trend period, wherein the trend period specifically includes a global trend and a local trend;

[0154] S22: Judging whether the instantaneous abnormal event presents a preset periodic fluctuation;

[0155] S23: If yes, obtaining regular change information corresponding to the trend sensor data through the instantaneous abnormal event, and constructing a trend change point corresponding to the periodic fluctuation according to the regular change information.

[0156] In this embodiment, the system detects corresponding transient abnormal events from the trend sensing data based on a pre-set trend period, which specifically includes global trend and local trend, and then determines whether these transient abnormal events present pre-set periodic fluctuations to perform corresponding steps; for example, when the system determines that the transient abnormal events do not present pre-set periodic fluctuations, the system considers that these abnormal events are likely caused by accidental factors, environmental interference or device errors, rather than systematic or periodic abnormalities, and the system combines the trend sensing data to backtrack the environmental parameters (such as temperature and humidity, electromagnetic interference, pressure change, etc.) at the time of the abnormal event occurrence to determine whether there is a sudden external interference, detects the sensor state inside the smart watch, including battery power, transmission signal quality, hardware running state, to exclude the possibility of device failure, at the same time, if the abnormality involves multiple sensors and has a greater impact, further calibration of the sensor parameters of the smart watch may be needed, such as adjusting the sensor sensitivity, optimizing the data sampling frequency, etc., and a user prompt is provided to record the time, place and possible environmental factors of the abnormal event occurrence, allowing the user to input supplementary information to help the system optimize the abnormality detection strategy; for example, when the system determines that the transient abnormal events present pre-set periodic fluctuations, the system considers that these abnormal events are not accidental factors, but are likely to be systematic or periodic abnormalities, and the system obtains regular change information corresponding to the trend sensing data through these transient abnormal events, and constructs trend change points of periodic fluctuations according to different regular change information; the system can distinguish between accidental abnormalities and systematic abnormalities by determining the periodic fluctuations of the transient abnormal events, avoiding mistaking normal fluctuations for sudden abnormalities, which can improve the accuracy of the smart watch sensing data, make the system more reliable for data analysis and processing, and ensure that the user obtains high-quality health monitoring or exercise data feedback, at the same time, by extracting regular change information from the trend sensing data, the system can construct trend change points of periodic fluctuations, which enables the smart watch to predict possible future abnormal conditions, such as data fluctuations under specific time or specific conditions, so as to optimize the working mode of the sensor in advance, reduce errors, and improve the intelligence level of the device, and according to different regular change information, the system can optimize the compensation strategy of the sensing error, for example, when a long-term periodic fluctuation is detected, the smart watch can automatically adjust the sensor sensitivity, data filtering method or calibration parameters to reduce error accumulation and improve long-term stability, which is particularly important for high-precision scenarios (such as heart rate monitoring, step frequency calculation, etc.).

[0157] It should be noted that the regular change information corresponding to the trend sensing data is obtained through the transient abnormal events, and the trend change points corresponding to the periodic fluctuations are constructed according to the regular change information, and specific examples are as follows:

[0158] Assume a user wears a smartwatch for all-day health monitoring, and the watch has a built-in blood oxygen sensor that records SpO2 data every 5 minutes. Normally, the user's blood oxygen level is stable between 95%-98%. But in the past week, the system found that the user's blood oxygen data often drops temporarily between 2:00 and 3:00 in the morning, with the lowest value reaching 89%, and then returns to normal around 3:30;

[0159] First, data collection and anomaly detection are performed. The blood oxygen sensor continuously monitors the user's blood oxygen level and records all data points. The system found that the blood oxygen level often drops during the period from 2:00 to 3:00 in the morning, with some nights' minimum value even dropping to 89%. This phenomenon has lasted for several days, and the system determines that this abnormal data point is not accidental, but has some repeatability;

[0160] Then, regular change information extraction is performed. The system compares the data of the past week and finds that the occurrence time of this abnormal value is almost fixed between 2:00 and 3:00, and a temporary drop occurs every night, and then gradually recovers. Further analysis of environmental data (such as temperature, humidity, sleep posture, etc.) shows that the user may have a risk of sleep apnea during this time period. Accelerometer data shows that the user's turning over frequency increases during this period, which may be due to discomfort caused by the decrease in blood oxygen level;

[0161] Then, trend change points are constructed. After data analysis, the system marks this time period as "blood oxygen drop prone period" and establishes trend change points to improve the accuracy of anomaly recognition when similar situations are detected in the future. The system will combine historical data to predict future blood oxygen fluctuation patterns and may adjust the blood oxygen measurement frequency to increase data sampling density during this period to improve monitoring accuracy;

[0162] Then, intelligent optimization and user feedback are performed. If the anomaly is related to some adjustable factors, such as the user's sleep posture, the system can prompt the user to adjust the sleep posture or use a specific pillow to improve respiratory patency. If the anomaly may involve health risks, the system can suggest that the user undergo more in-depth health checks, such as sleep apnea monitoring, to confirm whether there is sleep apnea syndrome (SAS). In the future, the system can provide early warning during this period, such as reminding the user when they fall asleep, or working with the medical system to trigger a health alert when a serious drop in blood oxygen is detected;

[0163] In summary, by constructing trend change points, the system not only discovers instantaneous abnormal events, but also successfully extracts potential periodic health problems (such as nighttime low oxygen risk); this provides early warning to the user, avoiding potential health risks; at the same time, by analyzing periodic fluctuations, the system can optimize the blood oxygen monitoring algorithm, increasing the sampling frequency in a specific time period to obtain more accurate data; in the future, the system can even adaptively adjust the measurement mechanism, for example, when a downward trend in blood oxygen is found, automatically enable a higher precision measurement mode; and the system provides personalized health recommendations based on trend change points, for example, for users with nighttime blood oxygen decline, the system can recommend suitable sleep posture, breathing training or air quality optimization scheme; if the system finds that the abnormal pattern exists for a long time, it can also suggest the user to consult a doctor, or even connect to remote medical services through a smart watch; traditional anomaly detection methods may ignore short-term fluctuations or misjudge them as accidental errors, while the system can include these fluctuations in long-term analysis by constructing trend change points, reducing misjudgment and improving detection reliability; through this mechanism, the smart watch is not just a simple health monitoring device, but a smart system that can identify, learn and optimize user health management.

[0164] In the embodiment, the step S4 of judging whether the abnormal value can be repaired includes:

[0165] S41: Based on the time window corresponding to the abnormal value, calculate the data missing rate of the time window;

[0166] S42: Determine whether the data missing rate exceeds the preset missing value;

[0167] S43: If yes, identify the influence content of the missing data on the final sensor data through the abnormal value, guide the user to re-measure the additional sensor data according to the influence content, and record the corresponding abnormal reason in the smart watch according to the abnormal data segment of the abnormal value, wherein the abnormal reason specifically includes sensor falling off, signal loss and external environmental factors.

[0168] In this embodiment, the system calculates the data missing rate of the time window based on the abnormal value corresponding to the time window, and then the system determines whether the data missing rate exceeds the pre-set missing value to perform the corresponding steps; for example, when the system determines that the data missing rate of the time window does not exceed the pre-set missing value, the system considers that the degree of data loss is within an acceptable range and does not have a significant impact on the overall data analysis, prediction or compensation strategy. The system can still perform effective calculation and processing through existing data, and the system will perform calculation according to the original logic without triggering the abnormal processing mechanism. For example, in the sleep monitoring process, if the body movement data of an individual time point is lost, but the overall sleep trend is not affected, the system can still normally calculate the deep sleep and light sleep ratio and output a reliable sleep report. At the same time, the system continuously tracks the data missing situation of the time window to ensure that the missing rate does not continuously increase in a short period of time. For example, if the smart watch occasionally has less than 5% data loss in a day, and such situation does not occur frequently, the system can consider that the device sensor and communication module are still working stably and there is no need to trigger additional repair mechanism. If the missing rate is at a low level but shows an upward trend, the system can dynamically adjust the monitoring threshold to increase the attention to data integrity. For example, if the data missing rate is close to the set threshold (for example, the set threshold is 10%, but the current missing rate has reached 8%), the system can give a soft reminder in advance to prompt the user to keep the device connection stable, such as checking the Bluetooth connection or wearing position. For example, when the system determines that the data missing rate of the time window exceeds the pre-set missing value, the system considers that the data loss will have a significant impact. The system identifies the influence content of the missing data on the final sensing data through the abnormal value, guides the user to re-measure the final sensing data according to different influence content, and records the corresponding abnormal reason in the smart watch according to the abnormal data segment of the abnormal value. The abnormal reason specifically includes sensor falling off, signal loss and external environmental factors.The system can timely discover situations that may affect the accuracy of measurement by identifying outliers and assessing the impact of missing data on the final sensor data. When the data missing rate exceeds the preset threshold, the system can take proactive measures to reduce the spread of false data, thereby improving the data integrity recorded by the smartwatch and ensuring that users obtain more reliable health monitoring results. At the same time, by guiding users to take additional measurements, the system can minimize the impact of data loss on monitoring results. For example, in heart rate monitoring, if the data missing rate is too high, the system can remind the user to adjust the wearing position or take measurements again during a specific period to make up for the missing data, improve the accuracy of health analysis, and record the specific reasons for the abnormal data segment (such as sensor falling off, signal loss, or external environmental factors) to help the system analyze the root cause of data loss and take targeted measures. For example, if frequent signal loss is detected, the system can prompt the user to check the Bluetooth connection or adjust the wearing method of the smartwatch to avoid similar problems from recurring. By storing and analyzing historical abnormal records, the system can continuously optimize its data collection strategy. For example, if a user frequently experiences data loss due to environmental interference, the system can adjust the sampling frequency of the sensor or enhance the signal processing algorithm to adapt to data collection needs in different environments, improving the stability and adaptability of the smartwatch in complex environments.

[0169] In this embodiment, based on the preset use duration of the smartwatch, the step S1 of collecting corresponding trend sensor data by real-time monitoring of the preset sensor by the smartwatch further includes:

[0170] S11: detecting a time window in which outliers occur based on preset statistical characteristics of sensor data, wherein the statistical characteristics specifically include mean, variance, and standard deviation;

[0171] S12: determining whether the number of time windows reaches a preset threshold;

[0172] S13: if yes, triggering a preset self-calibration mode of the sensor regularly according to the use duration, reducing error accumulation of the sensor through the self-calibration mode, activating a preset calibration tutorial of the smartwatch, and guiding the user to manually calibrate the smartwatch according to the calibration tutorial.

[0173] In the present embodiment, the system detects time windows of abnormal value occurrence based on pre-set statistical characteristics of sensor data, which specifically include mean, variance and standard deviation, and then determines whether the number of time windows reaches a pre-set threshold to execute corresponding steps; for example, when the system determines that the number of time windows does not reach the pre-set threshold, the system considers that the current abnormal value occurrence is relatively sporadic, which may be an occasional error rather than a systematic or trend abnormality, and the system temporarily suspends the execution of compensation or correction measures to avoid unnecessary adjustment of normal data, which helps to reduce unnecessary intervention caused by misjudgment, ensures the stability of data processing, and continues to monitor subsequent data and temporarily stores the current abnormal data. If the frequency of abnormal value occurrence in subsequent data increases and gradually reaches the set threshold, the system will re-evaluate the situation to determine whether to take further processing measures, and perform local analysis on these sporadic abnormal values, such as checking whether they are concentrated in certain specific conditions (such as high temperature environment or loose wearing condition), if it is found that abnormal values mainly occur in certain specific environment or operation behavior, the system can prompt the user to adjust, such as re-wearing the smart watch or avoiding certain specific interference environment; for example, when the system determines that the number of time windows reaches the pre-set threshold, the system considers that the current abnormal value is relatively concentrated, the system triggers the pre-set self-calibration mode of the sensor according to the usage time of the smart watch, reduces the error accumulation of the sensor through the self-calibration mode, activates the pre-set calibration tutorial of the smart watch, and guides the user to manually calibrate the smart watch according to the calibration tutorial.The system can effectively reduce the error accumulation of the sensor caused by environmental changes, hardware aging or continuous use by triggering the self-calibration mode of the sensor regularly, ensuring that the measurement data of the smart watch always maintains high precision, which is particularly important for functions that rely on sensor data such as health monitoring and motion tracking, and can improve user experience and data reliability. At the same time, since the smart watch is used in different scenarios, its sensor may be affected by external factors such as temperature changes, humidity, electromagnetic interference, etc., causing measurement deviation. The system triggers self-calibration based on usage time, allowing the device to dynamically adjust based on actual usage, avoiding gradual distortion of measurement data due to long-term non-calibration, enhancing the device's adaptive ability. By activating the built-in calibration tutorial of the smart watch, the user can manually calibrate the device, allowing the user to more intuitively understand the device's status and participate in the device's maintenance process. This not only ensures data accuracy but also improves user trust in the device and enhances user interaction experience. Because the sensor has not been calibrated for a long time, it may cause large measurement errors, affecting the user's judgment of the smart watch data. For example, if the health monitoring data such as heart rate and blood oxygen deviates for a long time, it may mislead the user's health management. The system combines self-calibration and manual calibration to reduce the potential risks of data errors, ensuring reliable measurement results and improving the overall safety and practicality of the device.

[0174] Reference the attached Figure 2 The sensor error calibration system of the smart watch in an embodiment of the present application comprises:

[0175] The acquisition module 10 is configured to monitor the preset sensor of the smart watch in real time based on the preset usage time of the smart watch, and collect corresponding trend sensor data.

[0176] The judgment module 20 is configured to judge whether the trend sensor data presents a preset deviation trend.

[0177] The execution module 30 is configured to select a preset data point from the trend sensor data if the trend sensor data presents a preset deviation trend, calculate the standard deviation of the data point, mark the data point that exceeds the standard deviation range as an outlier in the trend sensor data, and perform an anomaly classification on the outlier, wherein the anomaly classification specifically includes short-term sudden anomaly, long-term drift and outlier.

[0178] The second judgment module 40 is configured to judge whether the outlier can be repaired.

[0179] The second execution module 50 is configured to, if possible, perform drift compensation on the abnormal value according to the abnormal classification, dynamically calibrate a sensing error coefficient of the smart watch, take an environment factor pre-collected by the smart watch as a compensation input of the sensing error coefficient, and adaptively correct the sensing error coefficient, wherein the drift compensation specifically includes linear drift compensation, nonlinear drift compensation, and periodic drift compensation, and the environment factor specifically includes temperature, humidity, and air pressure.

[0180] In the embodiment, the collection module 10 collects the corresponding trend sensor data by monitoring the pre-set sensor through the smart watch based on the pre-set used time length of the smart watch, and then the judgment module 20 judges whether the trend sensor data presents the pre-set deviation trend to execute the corresponding steps; for example, when the system determines that the trend sensor data collected by the sensor does not present the pre-set deviation trend, the system considers that the measurement accuracy and stability of the current sensor still meet the expectation, no obvious drift or abnormal situation occurs, the system continues to record and store the measurement data according to the existing sensor data processing logic, ensures the continuity and integrity of the data, maintains the existing sensor error compensation strategy, and does not perform additional compensation adjustment, so as to avoid over-correction affecting the measurement result, and stores the current measurement environment (such as temperature, humidity, air pressure) and device state, so as to compare with the current state when an abnormality is detected in the future, analyze the possible influencing factors, and set a periodic calibration interval; for example, when the system determines that the trend sensor data collected by the sensor presents the pre-set deviation trend, the execution module 30 considers that the measurement accuracy of the current sensor may have drift or abnormal situation, the system selects the pre-set data points from the trend sensor data, calculates the standard deviation of the data points, marks the data points exceeding the standard deviation range as abnormal values in the trend sensor data, classifies the abnormal values, and the abnormal classification specifically includes short-term sudden abnormality, long-term drift and outlier; the system can identify the sensor measurement error in time by judging whether the trend sensor data deviates from the set standard, prevent the error data from affecting the final measurement result, adopts the standard deviation calculation and abnormal value marking, which helps to exclude incidental interference and improve the stability and accuracy of the data, and divides the abnormal values into short-term sudden abnormality, long-term drift and outlier, so that the system can take corresponding processing strategies according to different types of abnormalities, the short-term sudden abnormality can be processed by smoothing filtering, the long-term drift can trigger the automatic compensation mechanism, and the outlier can be used for early warning of possible hardware failure of the sensor, improves the fault management ability of the smart watch, and on the basis of the abnormal classification, the system can adjust the sensor compensation strategy according to the abnormal type, for example, the trend compensation algorithm is adopted to correct the long-term drift, so that the sensor can maintain high-precision measurement ability even after long-time use, thereby improving the reliability and user experience of the smart watch in complex environment; and then the second judgment module 40 judges whether the abnormal values in the trend sensor data can be repaired to execute the corresponding steps;For example, when the system determines that the abnormal values in the trend sensor data cannot be repaired, the system considers that it may be caused by sensor hardware failure, serious long-term drift, environmental interference beyond the compensable range, or too much data missing to effectively compensate. The system will store the unrepaired data separately and mark it in the log or data record to avoid affecting the calculation of normal data. If there are few abnormal values, the system can use historical trend data, interpolation of adjacent data, or pre-set default values for compensation to maintain measurement continuity. At the same time, if abnormal data continues to appear, the system can trigger a self-checking mechanism to check the sensor working state and send a reminder to the user, such as suggesting recalibration, checking hardware, or replacing the device. If the smartwatch has multiple sensors, the system can reduce the data weight of the faulty sensor and preferentially use data from other sensors for calculation or switch to a backup measurement mode (such as using a different algorithm or data source). For example, when the system determines that the abnormal values in the trend sensor data cannot be repaired, the system considers that it may be caused by sensor hardware failure, serious long-term drift, environmental interference beyond the compensable range, or too much data missing to effectively compensate. The system will store the unrepaired data separately and mark it in the log or data record to avoid affecting the calculation of normal data. If there are few abnormal values, the system can use historical trend data, interpolation of adjacent data, or pre-set default values for compensation to maintain measurement continuity. At the same time, if abnormal data continues to appear, the system can trigger a self-checking mechanism to check the sensor working state and send a reminder to the user, such as suggesting recalibration, checking hardware, or replacing the device. If the smartwatch has multiple sensors, the system can reduce the data weight of the faulty sensor and preferentially use data from other sensors for calculation or switch to a backup measurement mode (such as using a different algorithm or data source). For example, when the system determines that the abnormal values in the trend sensor data cannot be repaired, the system considers that it may be caused by sensor hardware failure, serious long-term drift, environmental interference beyond the compensable range, or too much data missing to effectively compensate. The system will store the unrepaired data separately and mark it in the log or data record to avoid affecting the calculation of normal data. If there are few abnormal values, the system can use historical trend data, interpolation of adjacent data, or pre-set default values for compensation to maintain measurement continuity. At the same time, if abnormal data continues to appear, the system can trigger a self-checking mechanism to check the sensor working state and send a reminder to the user, such as suggesting recalibration, checking hardware, or replacing the device. If the smartwatch has multiple sensors, the system can reduce the data weight of the faulty sensor and preferentially use data from other sensors for calculation or switch to a backup measurement mode (such as using a different algorithm or data source).

[0181] In the embodiment, the execution module further includes:

[0182] The identification unit is configured to acquire proportion data of the abnormal number of the abnormal values in the trend sensor data based on the abnormal number of the abnormal values, and identify distribution information of the abnormal values according to the proportion data.

[0183] a judging unit configured to judge whether the distribution information matches a preset specific condition, wherein the specific condition specifically comprises that the abnormal values are concentrated in an abnormal fluctuation caused by a specific time period or a specific environmental factor;

[0184] an executing unit configured to, if yes, detect a working environment of the sensor, calculate an accuracy error of the sensor according to the working environment, and obtain a corresponding abnormal rule from the distribution information.

[0185] In this embodiment, the system collects the proportion data of the number of abnormal values in the trend sensing data based on the number of abnormal values, identifies the distribution information of the abnormal values according to different proportion data, and then judges whether the distribution information matches the pre-set specific condition, and the specific condition specifically includes that the abnormal values are concentrated in abnormal fluctuations caused by a specific time period or a specific environmental factor, so as to execute the corresponding steps; for example, when the system determines that the distribution information of the abnormal values cannot match the pre-set specific condition, the system considers that the occurrence of the abnormal values is not caused by known patterns or predictable external environmental factors (such as a specific time period or a specific environmental factor), and appropriately increases the time window or the number of data points to improve the accuracy of the system judgment, for example, if the current analysis is based on 10 minutes of data, it can be changed to 30 minutes or 1 hour of data analysis to capture a wider trend, and if the distribution of abnormal values is relatively scattered, the weight of standard deviation calculation or the threshold of abnormal value judgment may need to be adjusted to prevent over-filtering or missing abnormal values, and if the proportion of abnormal values continues to increase but the reason is unknown, the system can enter a warning mode, reduce the data reliability, or only be used for user reference, without directly affecting the key decision (such as health warning); for example, when the system determines that the distribution information of the abnormal values can match the pre-set specific condition, the system considers that the occurrence of the abnormal values is caused by known factors, detects the working environment of the sensor, calculates the precision error of the sensor according to different working environments, and obtains the corresponding abnormal rules from the distribution information; the system detects the working environment of the sensor, calculates the precision error according to different environmental conditions, can dynamically adjust the calibration parameters of the sensor, for example, if the system detects that the increase of 5°C in the environmental temperature causes the measurement error to increase by 0.2%, the system can compensate for this error in real time, so as to ensure that the measurement result is not affected by environmental fluctuations, and by analyzing the distribution rule of the abnormal values, the system can identify long-term trends and adjust the error compensation strategy, for example, if the system finds that a certain sensor measures high in a high humidity environment, a specific humidity calibration algorithm can be automatically enabled to maintain the accuracy of the measurement data for a long time and prolong the effective service life of the device, and if the system cannot determine the source of the abnormality, the data repair or compensation mechanism may be triggered, resulting in unnecessary calculation consumption or even data distortion, and based on the matching of the abnormal values with the known environmental factors, the system can execute compensation under the correct conditions to avoid interference with normal data, thereby improving the overall energy efficiency and reliability of the system.

[0186] In this embodiment, it also includes:

[0187] The extraction module is configured to extract data abnormal features of the abnormal values, and compare output data of different sensors in the smart watch, wherein the data abnormal features specifically include fluctuation amplitude, change speed and change direction.

[0188] A third judging module is configured to judge whether the output data detects a preset accompanying anomaly.

[0189] A third executing module is configured to, if yes, identify a communication protocol interface between the different sensors, acquire a bus communication state of the smart watch based on the communication protocol interface, and generate a network communication quality of the smart watch in real time according to the bus communication state, where the bus communication state specifically includes noise interference, data transmission delay and connection loss, and the network communication quality specifically is a communication quality when data transmission is performed between the different sensors.

[0190] In the present embodiment, the system extracts data anomaly features of the abnormal value, which specifically include fluctuation amplitude, change speed and change direction, compares the output data of different sensors in the smart watch, and then determines whether the output data detects the pre-set accompanying anomaly to execute corresponding steps; for example, when the system determines that the output data of different sensors in the smart watch does not detect the pre-set accompanying anomaly, the system considers that the current abnormal value is likely to be a short-time error or accidental fluctuation, rather than caused by environmental factors, equipment failure or systematic error, the system temporarily stores the current abnormal value, and continues to collect data in the subsequent time window to observe whether there is a persistent deviation; for example, if the heart rate sensor reading suddenly rises to 180 bpm in a certain measurement, but the acceleration sensor and skin resistance sensor are normal, it may be a false measurement, the system will verify whether the data returns to normal within the next 10 seconds, and if the abnormal point does not trigger the accompanying anomaly but occurs multiple times in a short time, the system will record these data and mark them as "low-priority anomaly" for review in subsequent data analysis; for example, when the system determines that the output data of different sensors in the smart watch detects the pre-set accompanying anomaly, the system considers that the current abnormal value is caused by environmental factors, equipment failure or systematic error, the system identifies the communication protocol interface between different sensors, obtains the bus communication state of the smart watch based on the communication protocol interface, and the bus communication state specifically includes noise interference, data transmission delay and connection loss, and generates the network communication quality of the smart watch in real time according to different bus communication states, and the network communication quality specifically refers to the communication quality when different sensors transmit data; by analyzing the communication protocol interface and bus communication state between different sensors, the system can determine whether the anomaly is caused by environmental factors, equipment failure or systematic error, which helps to accurately locate the source of the anomaly and avoid mistaking hardware failure or communication anomaly as sensor drift or short-time fluctuation, and can dynamically evaluate the integrity and accuracy of sensor data during transmission. If the system finds that the data is affected due to poor communication quality, it can take measures such as redundancy check, automatic retransmission or signal enhancement to ensure data stability; for example, if the data transmission of a certain sensor has high delay, the system can preferentially use other sensor data or perform data compensation to reduce the influence of the anomaly, and can automatically adjust the data processing strategy based on real-time communication quality evaluation; for example, in a high signal noise environment, the system can reduce the sensor sampling rate to reduce false positives, or enable redundant sensors for data compensation, thereby improving measurement accuracy. This adaptive optimization capability enables the smart watch to maintain high precision operation in complex environments and improve user experience.

[0191] In the embodiment, the second execution module further comprises:

[0192] A second identification unit is configured to identify a compensation demand of the sensing error coefficient based on fluctuation changes of the environmental parameters, wherein the fluctuation changes specifically include a temperature and humidity fluctuation range, air pressure changes and electromagnetic interference;

[0193] A second judgment unit is configured to judge whether the compensation demand matches a preset compensation strategy.

[0194] A second execution unit is configured to, if not, acquire a use scenario of the smart watch by a user, adaptively optimize the compensation strategy according to the use scenario, and dynamically adjust an influence coefficient of the compensation strategy when a single sensor has an error according to a preset complementarity between different sensors.

[0195] In the embodiment, the system identifies the compensation demand of the sensing error coefficient based on the fluctuation change of the environmental parameters, and the fluctuation change specifically includes the temperature and humidity fluctuation range, the air pressure change and the electromagnetic interference. Then, the system determines whether the compensation demand matches the pre-set compensation strategy to execute the corresponding steps. For example, when the system determines that the compensation demand of the sensing error coefficient matches the pre-set compensation strategy, the system considers that the source of the current error is known, and the system has the corresponding compensation scheme to correct the error. The system calculates the optimal compensation parameter according to the matched compensation strategy and the current environmental parameter fluctuation. For example, if the temperature rise causes the sensor measurement value to be too high, the system can introduce a temperature compensation factor to dynamically adjust the measurement value. If the electromagnetic interference increases, the system can use a filtering algorithm to reduce the error or temporarily increase the signal sampling times to improve the data accuracy. At the same time, one or more of linear compensation, nonlinear compensation or periodic drift compensation is selected for adjustment. For example, it is suitable for the case that the error changes linearly with the environmental parameter, such as the error drift caused by temperature change. It is suitable for the case that the error changes nonlinearly with the environment, such as the unstable fluctuation of the sensor output in the high humidity environment. When the environmental parameter change has periodic characteristics (such as the diurnal temperature difference), the error trend is predicted through the historical data and the compensation value is dynamically adjusted. After the compensation strategy is executed, the sensing error coefficient is dynamically updated to ensure that the subsequent data acquisition meets the corrected measurement standard. For example, the relationship between the current temperature and humidity and the error is recorded to optimize the compensation model. When the electromagnetic interference is strong, the signal processing algorithm is adjusted to ensure stable data acquisition. For example, when the system determines that the compensation demand of the sensing error coefficient cannot match the pre-set compensation strategy, the system considers that the source of the current error is unknown. The system obtains the use scenarios of the smart watch, and the use scenarios specifically include the sports scenario, the office and life scenario and the extreme weather scenario. According to different use scenarios, the compensation strategy is adaptively optimized, and the influence coefficient of the compensation strategy when a single sensor error occurs is dynamically adjusted according to the complementarity between different sensors.By obtaining the user's usage scenarios such as sports, office life or extreme weather, the system can narrow down the possible causes of errors, for example, in the sports scenario, errors may be caused by violent shaking or posture changes, while in extreme weather, errors may be affected by environmental factors such as temperature, humidity, etc. This way can help the system more accurately determine the potential source of error, rather than relying on fixed compensation patterns. At the same time, according to different usage scenarios, the compensation strategy is adaptively optimized, which can make the compensation scheme more flexible, for example, in the sports scenario, the system can perform more stringent filtering on the inertial sensor data to reduce errors caused by violent movement, while in the extreme weather scenario, the system can increase the weight of the environmental sensor data to ensure that the changes in the external environment are reasonably considered. This dynamic adjustment method enables the system to maintain high accuracy and reliability in different environmental conditions, and according to the complementarity between different sensors, it can optimize the influence of single sensor error on the overall compensation strategy, for example, when the data of the acceleration sensor deviates, the system can use the data of the gyroscope for cross-validation to avoid the data error of a single sensor affecting the final measurement results too much. Through multi-sensor data fusion, the system can more accurately calculate the error source and reduce unnecessary compensation correction, improving the stability of the overall data.

[0196] In this embodiment, the determining module further includes:

[0197] The detection unit is configured to detect a corresponding instantaneous abnormal event from the trend sensor data based on a preset trend period, wherein the trend period specifically includes a global trend and a local trend.

[0198] The third determination unit is configured to determine whether the instantaneous abnormal event presents a preset periodic fluctuation.

[0199] The third execution unit is configured to, if so, obtain regular change information corresponding to the trend sensor data through the instantaneous abnormal event, and construct a trend change point corresponding to the periodic fluctuation according to the regular change information.

[0200] In this embodiment, the system detects corresponding transient abnormal events from the trend sensing data based on a pre-set trend period, which specifically includes global trend and local trend, and then determines whether the transient abnormal events present pre-set periodic fluctuations to perform corresponding steps; for example, when the system determines that the transient abnormal events do not present pre-set periodic fluctuations, the system considers that these abnormal events are likely to be caused by accidental factors, environmental interference or device errors, rather than systematic or periodic abnormalities, and the system combines the trend sensing data to backtrack the environmental parameters (such as temperature and humidity, electromagnetic interference, pressure change, etc.) at the time of the abnormal event occurrence to determine whether there is a sudden external interference, detects the sensor state inside the smart watch, including battery power, transmission signal quality, hardware running state, to exclude the possibility of device failure, at the same time, if the abnormality involves multiple sensors and has a greater impact, further calibration of the sensor parameters of the smart watch may be needed, such as adjusting the sensor sensitivity, optimizing the data sampling frequency, etc., and a user prompt is provided to record the time, place and possible environmental factors of the abnormal event occurrence, allowing the user to input supplementary information to help the system optimize the abnormality detection strategy; for example, when the system determines that the transient abnormal events present pre-set periodic fluctuations, the system considers that these abnormal events are not accidental factors, but are likely to be systematic or periodic abnormalities, and the system obtains regular change information corresponding to the trend sensing data through these transient abnormal events to construct trend change points of periodic fluctuations according to different regular change information; the system can distinguish between accidental abnormalities and systematic abnormalities by determining the periodic fluctuations of the transient abnormal events, avoiding mistaking normal fluctuations for sudden abnormalities, which can improve the accuracy of the smart watch sensing data, make the system more reliably analyze and process data, and ensure that the user obtains high-quality health monitoring or motion data feedback, at the same time, by extracting regular change information from the trend sensing data, the system can construct trend change points of periodic fluctuations, which enables the smart watch to predict possible future abnormal conditions, such as data fluctuations under specific time or specific conditions, so as to optimize the working mode of the sensor in advance, reduce errors, and improve the intelligence level of the device, and according to different regular change information, the system can optimize the compensation strategy of the sensing error, for example, when a long-term periodic fluctuation is detected, the smart watch can automatically adjust the sensor sensitivity, data filtering method or calibration parameter to reduce error accumulation and improve long-term stability, which is particularly important for high-precision scenarios (such as heart rate monitoring, step frequency calculation, etc.).

[0201] In this embodiment, the second determination module further includes:

[0202] The calculation unit is configured to calculate a data missing rate of the time window based on the abnormal value corresponding to the time window.

[0203] A fourth judging unit is configured to judge whether the data missing rate exceeds a preset missing value.

[0204] A fourth executing unit is configured to, if yes, identify, by the abnormal value, influence content of the missing data on the final sensing data, guide a user to additionally measure the final sensing data according to the influence content, and record a corresponding abnormal reason in the smart watch according to the abnormal data segment of the abnormal value, where the abnormal reason specifically includes a sensor falling off, signal loss, and external environmental factors.

[0205] In this embodiment, the system calculates the data missing rate of the time window based on the abnormal value corresponding to the time window, and then the system determines whether the data missing rate exceeds the pre-set missing value to perform the corresponding steps; for example, when the system determines that the data missing rate of the time window does not exceed the pre-set missing value, the system considers that the degree of data loss is within an acceptable range and does not have a significant impact on the overall data analysis, prediction or compensation strategy. The system can still perform effective calculation and processing through existing data, and the system will perform calculation according to the original logic without triggering the abnormal processing mechanism. For example, in the sleep monitoring process, if the body movement data of an individual time point is lost, but the overall sleep trend is not affected, the system can still normally calculate the deep sleep and light sleep ratio and output a reliable sleep report. At the same time, the system continuously tracks the data missing situation of the time window to ensure that the missing rate does not continuously increase in a short period of time. For example, if the smart watch occasionally has less than 5% data loss in a day, and such situation does not occur frequently, the system can consider that the device sensor and communication module are still working stably and there is no need to trigger additional repair mechanism. If the missing rate is at a low level but shows an upward trend, the system can dynamically adjust the monitoring threshold to increase the attention to data integrity. For example, if the data missing rate is close to the set threshold (for example, the set threshold is 10%, but the current missing rate has reached 8%), the system can give a soft reminder in advance to prompt the user to keep the device connection stable, such as checking the Bluetooth connection or wearing position. For example, when the system determines that the data missing rate of the time window exceeds the pre-set missing value, the system considers that the data loss will have a significant impact. The system identifies the influence content of the missing data on the final sensing data through the abnormal value, guides the user to re-measure the final sensing data according to different influence content, and records the corresponding abnormal reason in the smart watch according to the abnormal data segment of the abnormal value. The abnormal reason specifically includes sensor falling off, signal loss and external environmental factors.The system can timely discover situations that may affect the accuracy of measurement by identifying outliers and assessing the impact of missing data on final sensor data. When the data missing rate exceeds the preset threshold, the system can take proactive measures to reduce the spread of false data, thereby improving the data integrity recorded by the smartwatch and ensuring that users obtain more reliable health monitoring results. At the same time, by guiding users to take additional measurements, the system can minimize the impact of data loss on monitoring results. For example, in heart rate monitoring, if the data missing rate is too high, the system can remind the user to adjust the wearing position or take measurements again during a specific period to make up for the missing data, improve the accuracy of health analysis, and record the specific reasons for the abnormal data segment (such as sensor falling off, signal loss, or external environmental factors) to help the system analyze the root cause of data loss and take targeted measures. For example, if frequent signal loss is detected, the system can prompt the user to check the Bluetooth connection or adjust the wearing method of the smartwatch to avoid similar problems from recurring. By storing and analyzing historical abnormal records, the system can continuously optimize its data collection strategy. For example, if a user frequently experiences data loss due to environmental interference, the system can adjust the sampling frequency of the sensor or enhance the signal processing algorithm to adapt to data collection needs in different environments, improving the stability and adaptability of the smartwatch in complex environments.

[0206] In this embodiment, the acquisition module further includes:

[0207] The second detection unit is configured to detect a time window in which outliers occur based on preset statistical characteristics of the sensor data, wherein the statistical characteristics specifically include mean, variance, and standard deviation.

[0208] The fifth judgment unit is configured to judge whether the number of the time windows reaches a preset threshold.

[0209] The fifth execution unit is configured to, if yes, periodically trigger a preset self-calibration mode of the sensor according to the usage duration, reduce error accumulation of the sensor through the self-calibration mode, activate a preset calibration tutorial of the smartwatch, and guide the user to perform manual calibration on the smartwatch according to the calibration tutorial.

[0210] In the present embodiment, the system detects time windows of abnormal value occurrence based on pre-set statistical characteristics of sensor data, which specifically include mean, variance and standard deviation, and then determines whether the number of time windows reaches a pre-set threshold to execute corresponding steps; for example, when the system determines that the number of time windows does not reach the pre-set threshold, the system considers that the current abnormal value occurrence is relatively sporadic, which may be an occasional error rather than a systematic or trend abnormality, and the system temporarily suspends the execution of compensation or correction measures to avoid unnecessary adjustment of normal data, which helps to reduce unnecessary intervention caused by misjudgment, ensures the stability of data processing, and continues to monitor subsequent data and temporarily stores the current abnormal data. If the frequency of abnormal value occurrence in subsequent data increases and gradually reaches the set threshold, the system will re-evaluate the situation to determine whether to take further processing measures, and perform local analysis on these sporadic abnormal values, such as checking whether they are concentrated in certain specific conditions (such as high temperature environment or loose wearing condition), if it is found that abnormal values mainly occur in certain specific environment or operation behavior, the system can prompt the user to adjust, such as re-wearing the smart watch or avoiding certain specific interference environment; for example, when the system determines that the number of time windows reaches the pre-set threshold, the system considers that the current abnormal value is relatively concentrated, the system triggers the pre-set self-calibration mode of the sensor according to the usage time of the smart watch, reduces the error accumulation of the sensor through the self-calibration mode, activates the pre-set calibration tutorial of the smart watch, and guides the user to manually calibrate the smart watch according to the calibration tutorial.The system can effectively reduce the error accumulation of the sensor caused by environmental changes, hardware aging or continuous use by triggering the self-calibration mode of the sensor regularly, ensuring that the measurement data of the smart watch always maintains high precision, which is particularly important for functions that rely on sensor data such as health monitoring and sports tracking, and can improve the user's experience and data reliability. At the same time, since the smart watch is used in different scenarios, its sensor may be affected by external factors such as temperature changes, humidity, electromagnetic interference, etc., which may cause measurement deviation. The system triggers self-calibration based on usage time, allowing the device to dynamically adjust according to actual usage, avoiding gradual distortion of measurement data due to long-term non-calibration, enhancing the device's adaptive ability. By activating the built-in calibration tutorial of the smart watch, the user can manually calibrate the device, allowing the user to more intuitively understand the device's status and participate in the device's maintenance process. This not only ensures the accuracy of the data, but also improves the user's trust in the device and enhances the user's interactive experience. Because the sensor has not been calibrated for a long time, it may cause large measurement errors, affecting the user's judgment of the smart watch data. For example, if the heart rate, blood oxygen and other health monitoring data deviate for a long time, it may mislead the user's health management. The system combines self-calibration and manual calibration to reduce the potential risks caused by data errors, ensuring the reliability of the measurement results, thereby improving the overall safety and practicality of the device.

[0211] Although embodiments of the present application have been shown and described, it is to be understood that various modifications, substitutions, replacements and changes can be made to these embodiments without departing from the principles and spirit of the present application, and the scope of the present application is defined by the appended claims and their equivalents.

Claims

1.A method for calibrating sensor errors of a smart watch, the method comprising: The method comprises the following steps: Based on the preset use time of the smart watch, the preset sensor is monitored in real time through the smart watch to collect corresponding trend sensor data; Determine whether the trend sensor data shows a preset deviation trend; If so, select a preset data point from the trend sensor data, calculate the standard deviation of the data point, mark the data points outside the standard deviation range as outliers in the trend sensor data, and classify the outliers, wherein the outlier classification specifically includes short-term sudden outliers, long-term drifts and outliers; Determine whether the outliers can be repaired; If so, according to the outlier classification, the outliers are compensated for drift, the sensor error coefficient of the smart watch is dynamically calibrated, the environmental factors pre-collected by the smart watch are used as compensation input of the sensor error coefficient, and the sensor error coefficient is adaptively corrected, wherein the drift compensation specifically includes linear drift compensation, nonlinear drift compensation and periodic drift compensation, and the environmental factors specifically include temperature, humidity and air pressure; In the step of marking the data points outside the standard deviation range as outliers in the trend sensor data, it further comprises: Based on the number of outliers, collect the proportion data of the number of outliers in the trend sensor data, and identify the distribution information of the outliers according to the proportion data; Determine whether the distribution information matches a preset specific condition, wherein the specific condition specifically includes outliers concentrated in a specific time period or abnormal fluctuations caused by a specific environmental factor; If so, detect the working environment of the sensor, calculate the accuracy error of the sensor according to the working environment, and obtain the corresponding abnormal rules from the distribution information; In the step of using the environmental factors pre-collected by the smart watch as compensation input of the sensor error coefficient, it further comprises: Based on the fluctuation of the environmental factors, identify the compensation demand of the sensor error coefficient, wherein the fluctuation specifically includes temperature and humidity fluctuation range, air pressure change and electromagnetic interference; Determine whether the compensation demand matches a preset compensation strategy; If not, obtain the use scenario of the smart watch by the user, adaptively optimize the compensation strategy according to the use scenario, dynamically adjust the influence coefficient of the compensation strategy when a single sensor has an error according to the preset complementarity between different sensors, wherein the use scenario specifically includes sports scenario, office and life scenario and extreme weather scenario. 2.The sensor error calibration method of a smart watch according to claim 1, characterized in that, Before the step of classifying the outliers, it further comprises: Extract the data anomaly features of the outliers, and compare the output data of different sensors in the smart watch, wherein the data anomaly features specifically include fluctuation amplitude, change speed and change direction; Determine whether the output data detects a preset associated anomaly; If yes, a communication protocol interface between the different sensors is identified, a bus communication state of the smart watch is acquired based on the communication protocol interface, and a network communication quality of the smart watch is generated in real time according to the bus communication state, wherein the bus communication state specifically includes noise interference, data transmission delay and connection loss, and the network communication quality specifically is a communication quality when data transmission is performed between different sensors. 3.The sensor error calibration method of a smart watch according to claim 1, characterized in that, The step of judging whether the trend sensor data presents a preset deviated trend further includes: detecting a corresponding instantaneous abnormal event from the trend sensor data based on a preset trend period, wherein the trend period specifically includes a global trend and a local trend; judging whether the instantaneous abnormal event presents a preset periodic fluctuation; if yes, acquiring regular change information corresponding to the trend sensor data through the instantaneous abnormal event, and constructing a trend change point corresponding to the periodic fluctuation according to the regular change information. 4.The sensor error calibration method of a smart watch of claim 1, wherein, The step of judging whether the abnormal value can be repaired further includes: calculating a data missing rate of a time window corresponding to the abnormal value based on the time window; judging whether the data missing rate exceeds a preset missing value; if yes, identifying influence content of missing data on final sensor data through the abnormal value, guiding a user to additionally measure the final sensor data according to the influence content, and recording a corresponding abnormal reason in the smart watch according to an abnormal data segment of the abnormal value, wherein the abnormal reason specifically includes sensor falling off, signal loss and external environmental factors. 5.The sensor error calibration method of a smart watch of claim 1, wherein, The step of collecting corresponding trend sensor data by monitoring a preset sensor in the smart watch based on a preset use duration of the smart watch further includes: detecting a time window in which an abnormal value occurs based on a preset statistical characteristic of sensor data, wherein the statistical characteristic specifically includes a mean value, a variance and a standard deviation; judging whether a number of the time windows reaches a preset threshold; if yes, periodically triggering a preset self-calibration mode of the sensor according to the use duration, reducing error accumulation of the sensor through the self-calibration mode, activating a preset calibration tutorial of the smart watch, and guiding a user to manually calibrate the smart watch according to the calibration tutorial. 6.A sensor error calibration system of a smart watch, characterized by, The method includes: collecting corresponding trend sensor data by monitoring a preset sensor in the smart watch based on a preset use duration of the smart watch; judging whether the trend sensor data presents a preset deviated trend; if yes, selecting a preset data point from the trend sensor data, calculating a standard deviation of the data point, marking the data point that exceeds a standard deviation range as an abnormal value in the trend sensor data, and performing abnormal classification on the abnormal value, wherein the abnormal classification specifically includes short-term sudden abnormality, long-term drift and outlier; judging whether the abnormal value can be repaired; The second execution module is configured to, if possible, perform drift compensation on the abnormal value according to the abnormal classification, dynamically calibrate a sensing error coefficient of the smart watch, take an environment factor pre-collected by the smart watch as a compensation input of the sensing error coefficient, and adaptively correct the sensing error coefficient, wherein the drift compensation specifically includes linear drift compensation, nonlinear drift compensation, and periodic drift compensation, and the environment factor specifically includes temperature, humidity, and air pressure. The execution module further includes: An identification unit configured to collect proportion data of the abnormal number in the trend sensing data based on the abnormal number of the abnormal value, and identify distribution information of the abnormal value according to the proportion data; A judgment unit configured to judge whether the distribution information matches a preset specific condition, wherein the specific condition specifically includes abnormal value concentration in abnormal fluctuation caused by a specific time period or a specific environment factor; An execution unit configured to, if yes, detect a working environment of the sensor, calculate an accuracy error of the sensor according to the working environment, and obtain a corresponding abnormal rule from the distribution information; The second execution module further includes: A second identification unit configured to identify compensation requirements of the sensing error coefficient based on fluctuation changes of the environment factor, wherein the fluctuation changes specifically include a temperature and humidity fluctuation range, air pressure changes, and electromagnetic interference; A second judgment unit configured to judge whether the compensation requirements match a preset compensation strategy; A second execution unit configured to, if no, obtain a use scenario of the smart watch by a user, adaptively optimize the compensation strategy according to the use scenario, and dynamically adjust an influence coefficient of the compensation strategy when an error occurs in a single sensor according to a preset complementarity between different sensors, wherein the use scenario specifically includes a sports scenario, an office and life scenario, and an extreme weather scenario. 7.The sensor error calibration system of a smart watch of claim 6, wherein, The execution module further includes: An extraction module configured to extract data abnormal features of the abnormal value, and compare output data of different sensors in the smart watch, wherein the data abnormal features specifically include fluctuation amplitude, change speed, and change direction; A third judgment module configured to judge whether a preset accompanying abnormality is detected from the output data; A third execution module configured to, if yes, identify a communication protocol interface between the different sensors, obtain a bus communication state of the smart watch based on the communication protocol interface, and generate a network communication quality of the smart watch in real time according to the bus communication state, wherein the bus communication state specifically includes noise interference, data transmission delay, and connection loss, and the network communication quality specifically is a communication quality when data transmission is performed between different sensors.

Images