Method and system for steady-state measurement of time drift by intelligent device

Abstract

The invention provides a steady-state measurement method and system for time drift by intelligent equipment, and is applied to the field of data processing. An auxiliary time observation source is established in the initialization stage of intelligent equipment, a multi-source time reference system is constructed through an environment periodic signal, a human body physiological periodic signal and an equipment internal hardware counting signal, and after a stable periodic reference signal is detected, a main drift observation source is dynamically constructed according to different signal types; therefore, a stable and reliable time reference can be selected for drift observation in different operation scenes, a time drift value is calculated by detecting the operation state of equipment and combining a periodic reference signal, then steady-state condition judgment is carried out on the drift value, and a steady-state drift average value is generated. And filtering and stabilizing treatment on instantaneous fluctuation drifting are realized.

Description

A method and system for steady-state measurement of time drift in intelligent devices Technical Field

[0001] This invention relates to the field of data processing, and in particular to a method and system for steady-state measurement of time drift by an intelligent device. Background Technology

[0002] In smart health wearable devices, such as smartwatches, health monitoring bracelets, patch-type physiological monitoring devices, and multi-sensor human body monitoring systems, the devices typically need to continuously collect physiological signals (such as electrocardiogram, heart rate, blood oxygen, acceleration of movement, respiration, etc.) and mark the sampled data with timestamps to ensure the continuity and analyzability of the data on the timeline.

[0003] To maintain the device's internal time, wearable devices typically use a low-power real-time clock (RTC) or a quartz crystal oscillator as the clock source. However, crystal oscillators are susceptible to factors such as temperature variations, power supply voltage fluctuations, device aging, and changes in environmental conditions during actual operation, leading to clock frequency deviations and time drift. This time drift accumulates over long periods of operation, causing a discrepancy between the device's internal time and the actual time. Summary of the Invention

[0004] This invention aims to solve the problem that the time drift of a device fluctuates with changes in temperature, power supply status, and device operating load, resulting in unstable drift measurement results and affecting the accuracy of correction. It provides a steady-state measurement method and system for time drift of intelligent devices.

[0005] To solve the technical problem, this invention employs the following technical means: This invention provides a steady-state measurement method for time drift in intelligent devices, comprising: based on a preset initialization phase of the intelligent device, acquiring the initial time difference between the local clock and a reference time of the intelligent device; establishing an auxiliary time observation source for the intelligent device based on the initial time difference, wherein the auxiliary time observation source specifically includes environmental periodic signals, human physiological periodic signals, and internal hardware counting signals of the device; determining whether the auxiliary time observation source possesses a stable periodic reference signal; if so, identifying the signal type corresponding to the stable periodic reference signal; dynamically constructing the main drift observation source for the intelligent device based on the signal type; detecting the operating state of the intelligent device; and through... The operating state includes calculating the time drift value of the smart device, wherein the signal type specifically includes human gait cycle and environmental power frequency cycle; determining whether the time drift value meets the preset steady-state conditions; if it does, generating the steady-state drift average value of the time drift value; dynamically compensating the local clock of the smart device based on the steady-state drift average value; collecting the predicted drift value pre-monitored by the smart device; detecting the drift abnormal state of the smart device based on the predicted drift value; and triggering the compensation strategy of the smart device based on the drift abnormal state, wherein the drift abnormal state specifically includes strong electromagnetic interference, sensor abnormality, and clock hardware failure; and the compensation strategy specifically includes compensation period, compensation amplitude, and compensation rate.

[0006] Furthermore, before the step of identifying the signal type corresponding to the stable periodic reference signal and dynamically constructing the main drift observation source of the smart device based on the signal type, the method further includes: collecting signal attributes when the user wears the smart device based on the sensors preset by the smart device, wherein the signal attributes specifically include acceleration signals, environmental electromagnetic signals, ambient light signals, and physiological rhythm signals; determining whether the signal attributes detect non-periodic interference; if not, extracting periodic features from the signal attributes through the periodic analysis preset by the smart device, identifying the time interval between adjacent periods based on the periodic features, and calculating the periodic change rate of the signal attributes, wherein the periodic analysis specifically includes time domain periodic analysis, frequency domain periodic analysis, and correlation periodic analysis.

[0007] Furthermore, the step of detecting the operating state of the smart device and calculating the time drift value of the smart device based on the operating state further includes: identifying a disturbance vector of the operating state based on task parameters executed by the smart device within a preset time window, wherein the task parameters specifically include task type, task duration, and number of task switching; determining whether the disturbance vector reaches a preset disturbance threshold; if so, performing change detection on the disturbance vector to obtain a clock stability disturbance event of the smart device; using the clock stability disturbance event as a boundary to divide the continuous operating time into several drift observation time segments; collecting the cumulative time difference of the smart device's clock based on the drift observation time segments; and calculating the drift gradient of the smart device based on the cumulative time difference of the clock.

[0008] Furthermore, the step of collecting the predicted drift value pre-monitored by the intelligent device and detecting the drift anomaly state of the intelligent device based on the predicted drift value further includes: generating a corresponding drift residual sequence based on the difference between the predicted drift value and the actual drift observation value; determining whether there are preset abnormal structural features in the drift residual sequence, wherein the abnormal structural features specifically include abrupt change points, continuous offset intervals, and periodic residual fluctuations; if so, identifying the time point where the residual change rate exceeds the preset abrupt change threshold based on the residual change rate of the drift residual sequence, dynamically marking the time point as a drift abrupt change candidate point, using the drift abrupt change candidate point as the starting boundary, expanding forward and backward in the time series to obtain continuous residual deviation intervals, and constructing the corresponding drift anomaly time period.

[0009] Furthermore, the step of determining whether the auxiliary time observation source has a stable periodic reference signal further includes: performing a sliding window analysis on the continuous periodic interval sequence pre-generated by the intelligent device to mark the corresponding candidate stable periodic intervals; determining whether the candidate stable periodic intervals have multiple consecutive periodic features; if so, acquiring the periodic candidate signals of the candidate stable periodic intervals, identifying the waveform similarity between adjacent periods based on the similar structure of the periodic candidate signals, and dynamically generating a confidence score for the periodic candidate signals based on the waveform similarity.

[0010] Furthermore, the step of determining whether the time drift value meets the preset steady-state conditions further includes: obtaining the convergence characteristics of the drift evolution trajectory generated by the time drift value within a preset time period; determining whether the convergence characteristics match the preset steady-state convergence; if so, detecting the periodic oscillation amplitude of the drift evolution trajectory around the center value, calculating the energy distribution of the time drift value according to the periodic oscillation amplitude, collecting the distribution data of the time drift value in different frequency ranges, and dynamically mapping the time drift value to the corresponding phase change sequence according to the distribution data.

[0011] Furthermore, the step of obtaining the initial time difference between the local clock and the reference time of the smart device based on the initialization phase preset by the smart device, and establishing an auxiliary time observation source for the smart device based on the initial time difference, further includes: constructing a corresponding time difference sampling sequence in chronological order within a preset time window based on the preset time reference marker point of the smart device; determining whether the time difference change amplitude of the time difference sampling sequence is lower than a preset threshold; if so, establishing a mapping relationship between the local clock and the reference time based on the time difference sampling sequence, and generating the initial time offset structure of the smart device based on the mapping relationship.

[0012] This invention also provides a steady-state measurement system for time drift of a smart device, comprising: an establishment module, configured to acquire an initial time difference between the local clock and a reference time of the smart device based on a preset initialization phase of the smart device, and establish an auxiliary time observation source for the smart device based on the initial time difference, wherein the auxiliary time observation source specifically includes an environmental periodic signal, a human physiological periodic signal, and a hardware counting signal inside the device; a judgment module, configured to determine whether the auxiliary time observation source has a stable periodic reference signal; and an execution module, configured to, if so, identify the signal type corresponding to the stable periodic reference signal, dynamically construct a main drift observation source for the smart device based on the signal type, detect the operating state of the smart device, and calculate the drift based on the operating state. The system calculates the time drift value of the smart device, wherein the signal type specifically includes human gait cycle and environmental power frequency cycle; a second judgment module is used to determine whether the time drift value meets the preset steady-state conditions; a second execution module is used to generate the steady-state drift average value of the time drift value if it meets the conditions, dynamically compensate the local clock of the smart device based on the steady-state drift average value, collect the predicted drift value pre-monitored by the smart device, detect the drift abnormal state of the smart device according to the predicted drift value, and trigger the compensation strategy of the smart device according to the drift abnormal state, wherein the drift abnormal state specifically includes strong electromagnetic interference, sensor abnormality and clock hardware failure, and the compensation strategy specifically includes compensation period, compensation amplitude and compensation rate.

[0013] Furthermore, it also includes: a data acquisition module, used to acquire signal attributes when the user wears the smart device based on sensors preset in the smart device, wherein the signal attributes specifically include acceleration signals, environmental electromagnetic signals, ambient light signals, and physiological rhythm signals; a third judgment module, used to determine whether the signal attributes detect non-periodic interference; and a third execution module, used to, if not, extract periodic features from the signal attributes through periodic analysis preset in the smart device, identify the time interval between adjacent periods based on the periodic features, and calculate the periodic change rate of the signal attributes, wherein the periodic analysis specifically includes time-domain periodic analysis, frequency-domain periodic analysis, and correlation periodic analysis.

[0014] Furthermore, the execution module further includes: an identification unit, used to identify the disturbance vector of the running state based on the task parameters executed by the smart device within a preset time window, wherein the task parameters specifically include task type, task duration, and task switching count; a judgment unit, used to judge whether the disturbance vector reaches a preset disturbance threshold; and an execution unit, used to, if yes, perform change detection on the disturbance vector, obtain the clock stability disturbance event of the smart device, use the clock stability disturbance event as a boundary to divide the continuous running time into several drift observation time segments, collect the cumulative clock time difference of the smart device according to the drift observation time segments, and calculate the drift gradient of the smart device based on the cumulative clock time difference.

[0015] This invention provides a steady-state measurement method and system for time drift in intelligent devices, which has the following beneficial effects: In the initialization phase of the intelligent device, an auxiliary time observation source is established, and a multi-source time reference system is constructed using environmental periodic signals, human physiological periodic signals, and internal hardware counting signals. After detecting a stable periodic reference signal, the main drift observation source is dynamically constructed according to different signal types, thereby enabling the selection of a stable and reliable time reference for drift observation under different operating scenarios. Simultaneously, by detecting the device's operating status and calculating the time drift value in conjunction with the periodic reference signal, and then judging the steady-state condition of the drift value and generating a steady-state drift average, the instantaneous fluctuation drift is filtered and stabilized. Furthermore, by dynamically compensating the local clock based on the steady-state drift average and detecting abnormal drift states in conjunction with predicted drift values, abnormal factors such as strong electromagnetic interference, sensor anomalies, or clock hardware failures can be identified in a timely manner, triggering corresponding compensation strategies. Attached Figure Description

[0016] Figure 1 is a flowchart illustrating an embodiment of the steady-state measurement method for time drift using an intelligent device according to the present invention; Figure 2 is a structural block diagram illustrating an embodiment of the steady-state measurement system for time drift using an intelligent device according to the present invention. Detailed Implementation

[0017] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. The realization of the purpose, functional features and advantages of the invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings.

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Referring to Figure 1, a steady-state measurement method for time drift of a smart device according to an embodiment of the present invention is shown, comprising: S1: Based on a preset initialization phase of the smart device, obtaining the initial time difference between the local clock and the reference time of the smart device, and establishing an auxiliary time observation source for the smart device according to the initial time difference, wherein the auxiliary time observation source specifically includes environmental periodic signals, human physiological periodic signals, and internal hardware counting signals of the device; S2: Determining whether the auxiliary time observation source has a stable periodic reference signal; S3: If so, identifying the signal type corresponding to the stable periodic reference signal, dynamically constructing the main drift observation source of the smart device according to the signal type, detecting the operating state of the smart device, and through the operating state... S4: Determine whether the time drift value meets the preset steady-state conditions; S5: If it does, generate the steady-state drift average value of the time drift value, dynamically compensate the local clock of the smart device based on the steady-state drift average value, collect the predicted drift value of the smart device, detect the drift abnormal state of the smart device according to the predicted drift value, and trigger the compensation strategy of the smart device according to the drift abnormal state, wherein the drift abnormal state specifically includes strong electromagnetic interference, sensor abnormality and clock hardware failure, and the compensation strategy specifically includes compensation period, compensation amplitude and compensation rate.

[0020] In this embodiment, the system, based on a pre-set initialization phase of the smart device, obtains the initial time difference between the smart device's local clock and a reference time. Based on this initial time difference, auxiliary time observation sources for the smart device are established. These auxiliary time observation sources specifically include environmental periodic signals, human physiological periodic signals, and internal hardware counting signals. The system then determines whether these auxiliary time observation sources possess stable periodic reference signals to execute corresponding steps. For example, if the system determines that these auxiliary time observation sources do not possess stable periodic reference signals, the system considers the current environment or device state to lack a periodic signal that can serve as a reliable time reference. The system continues to continuously collect environmental signals, human physiological signals, and internal hardware counting signals within a preset time window, re-detects the periodic characteristics of candidate observation signals, and dynamically updates the candidate set of auxiliary time observation sources. Simultaneously, it temporarily uses the historical drift model of the device's local clock or the steady-state drift average of the previous stable phase as a transitional reference to perform small-amplitude predictive compensation on the local clock to maintain the continuity of device time. Conversely, if the system determines that these auxiliary time observation sources possess stable periodic reference signals, the system considers the current environment or device state to possess a periodic signal that serves as a reliable time reference. The system identifies the signal type corresponding to the stable periodic reference signal, specifically including human gait cycle and environmental power frequency cycle. Based on different signal types, it dynamically constructs the main drift observation source for intelligent devices, detects the operating status of intelligent devices, and calculates the time drift value of intelligent devices based on the operating status. By identifying and determining the signal type of the stable periodic reference signal, such as human gait cycle or environmental power frequency cycle, the system can utilize the inherent stable cycle of the natural environment or human activities as a time reference, thereby reducing the impact of local clock errors of individual devices on time measurement and improving the reliability and stability of the time reference. Simultaneously, the system dynamically constructs the main drift observation source based on different signal types, enabling the device to select the most suitable time reference according to the actual operating scenario. For example, in human motion scenarios, the human gait cycle is prioritized, while in stable indoor environments, the environmental power frequency cycle is used. This achieves dynamic switching and adaptive construction of the drift observation source. Furthermore, after determining the main drift observation source, the system detects the operating status of the intelligent device and calculates the time drift value by combining it with the periodic reference signal. This effectively eliminates the interference of changes in the device's operating status on clock drift. Then, the system determines whether the time drift value meets the preset steady-state conditions and executes the corresponding steps.For example, when the system determines that the time drift value cannot meet the preset steady-state conditions, it considers the current device clock drift state to be in a dynamic fluctuation stage, and the obtained time drift value has not yet formed a stable trend. The system will then temporarily suspend the local clock compensation operation and enter a continuous drift observation and stability detection stage. Within a new time window, it will continue to collect time drift data, track and analyze drift changes over continuous time periods, and continuously monitor the device's operating status to determine if there are influencing factors such as temperature changes, power fluctuations, or task load changes. By continuously updating the drift observation data, it will gradually identify drift trends, enabling the system to obtain more stable and reliable drift estimation results. Conversely, when the system determines that the time drift value meets the preset steady-state conditions, it considers the current device clock drift state to be in a stable stage. The system will generate a steady-state drift average value, dynamically compensate the smart device's local clock based on this average value, collect the predicted drift value pre-monitored by the smart device, and detect abnormal drift states of the smart device according to different predicted drift values. Normal states specifically include strong electromagnetic interference, sensor malfunctions, and clock hardware failures. Based on different drift anomaly states, compensation strategies are triggered for the intelligent devices. These strategies include compensation period, compensation amplitude, and compensation rate. By collecting predicted drift values ​​pre-monitored by the intelligent devices, the system can anticipate the development trend of clock drift and compare the predicted results with the current drift state, thereby enabling early identification of potential drift anomalies. This monitoring method based on predicted drift values ​​can detect drift anomalies before they significantly affect the device's time accuracy, improving the system's perception and response speed to clock drift changes. Furthermore, by detecting the presence of drift anomalies such as strong electromagnetic interference, sensor malfunctions, or clock hardware failures based on different predicted drift values ​​and triggering corresponding compensation strategies for different anomaly states, targeted handling of different types of anomalies can be achieved. Moreover, by dynamically adjusting the compensation period, compensation amplitude, and compensation rate, the time compensation strategy can adaptively adjust according to the actual operating environment, thereby improving the time stability of the intelligent devices and the system reliability in complex operating environments.

[0021] It should be noted that, in identifying the signal type corresponding to the stable periodic reference signal, and dynamically constructing the main drift observation source of the intelligent device based on the signal type, the system detects the operating status of the intelligent device and calculates the time drift value of the intelligent device based on the operating status. Specifically, after detecting that the auxiliary time observation source has a stable periodic reference signal, the system first performs feature recognition on the stable periodic signal to determine its corresponding signal type. Specifically, the system analyzes the period length, frequency range, and signal source characteristics of the signal and matches them with a preset signal type library to identify the signal type. The signal type may include human gait cycle and environmental power frequency cycle, etc. Subsequently, the system dynamically constructs the main drift observation source of the intelligent device based on different signal types, that is, selects the corresponding periodic signal as the time reference benchmark for the device clock drift observation. After completing the construction of the drift observation source, the system further detects the operating status of the intelligent device, such as the internal temperature change, battery power status, processor load, or sensor working status, and analyzes the time relationship between the device's local clock and the periodic reference signal in combination with the operating status to calculate the current time drift value of the intelligent device. Joint analysis with equipment operating status information can reduce the impact of environmental changes or equipment operating fluctuations on drift calculations, making the calculated time drift values ​​more stable and reliable. A specific example is as follows: When a user wears a smartwatch for daily walking, the device detects periodic vibration signals generated during human walking through an accelerometer and identifies the period of this signal as approximately 0.8 seconds. The system then determines this stable periodic signal as the human gait cycle and uses the time interval between consecutive gait peaks as the primary drift observation source. While recording the gait cycle, the system also monitors the current operating status of the equipment, such as... The system measures processor load changes and battery power status, and calculates time drift based on time changes recorded by the device's local clock over several gait cycles. For example, in an indoor environment, when the device detects a stable 50Hz power grid frequency signal, the system uses this frequency cycle as a drift observation source. By counting the number of detected frequency cycles over a certain period and combining this with changes in the device's operating status, the system calculates the time deviation between the device's local clock and the power grid time reference, thus obtaining the device's time drift value. In this way, the system can use stable periodic signals as a time reference in different scenarios, improving the accuracy of time drift measurement.

[0022] It should be added that the system generates a steady-state drift average value for the time drift value, dynamically compensates the local clock of the smart device based on this steady-state drift average value, collects the predicted drift value pre-monitored by the smart device, detects abnormal drift states of the smart device based on the predicted drift value, and triggers the compensation strategy of the smart device according to the abnormal drift state. Specifically, when the system determines that the time drift value meets the preset steady-state condition, it indicates that the current device clock drift change has tended to stabilize. At this time, the system will perform statistical processing on multiple time drift values ​​within a continuous time window, such as generating a steady-state drift average value through averaging or smoothing. This steady-state drift average value can effectively reflect... The system analyzes the stable drift trend of the device under its current operating state and reduces the impact of instantaneous fluctuations on the drift calculation results. Subsequently, based on this steady-state drift average, the system dynamically compensates the local clock of the smart device by adjusting the timing rate or time accumulation value of the device's local clock, gradually bringing the device's time closer to the reference time. While completing clock compensation, the system continuously collects the predicted drift values ​​pre-monitored by the smart device, monitors and analyzes potential future drift trends, and determines whether there is abnormal drift based on the difference between the predicted drift value and the current drift state. When abnormal drift is detected, the system further identifies the source of the anomaly, such as strong electromagnetic interference or sensor malfunctions. In case of anomalies or clock hardware failures, the system triggers corresponding compensation strategies based on the type of anomaly. This involves dynamically adjusting the compensation period, magnitude, and rate to ensure the stability and accuracy of the device's timekeeping. For example, during prolonged operation of a smartwatch, the system continuously acquires multiple time drift values ​​over a period of time. If these drift values ​​show relatively small fluctuations, satisfying steady-state conditions, the system averages these drift values ​​to obtain a steady-state drift average. Subsequently, the system compensates the local clock based on this average, for example, by making minor adjustments to the local clock at regular intervals to reduce the deviation between the device's time and the reference time. Simultaneously, the system also... Historical data generates predicted drift values. When the prediction results indicate that a significant drift change may occur in the future, the system will further detect whether there are any abnormal factors. For example, when the equipment is in a strong electromagnetic environment, the system may detect a sudden increase in the predicted drift value and determine that it is a drift anomaly caused by strong electromagnetic interference. At this time, the system can shorten the compensation cycle and reduce the compensation amplitude to avoid frequent and large clock adjustments. For another example, when abnormal sensor output or abnormal clock hardware status is detected, the system can appropriately increase the compensation rate or trigger a backup compensation strategy to maintain stable operation of the equipment time. Through the above methods, dynamic compensation and anomaly handling of the equipment clock can be achieved under different operating environments.

[0023] In this embodiment, before step S3, which identifies the signal type corresponding to the stable periodic reference signal and dynamically constructs the main drift observation source of the smart device based on the signal type, the method further includes: S301: Based on the sensors preset by the smart device, collecting signal attributes when the user wears the smart device, wherein the signal attributes specifically include acceleration signals, environmental electromagnetic signals, ambient light signals, and physiological rhythm signals; S302: Determining whether the signal attributes detect non-periodic interference; S303: If not, extracting periodic features from the signal attributes through the periodic analysis preset by the smart device, identifying the time interval between adjacent periods based on the periodic features, and calculating the periodic change rate of the signal attributes, wherein the periodic analysis specifically includes time-domain periodic analysis, frequency-domain periodic analysis, and correlation periodic analysis.

[0024] In this embodiment, the system collects signal attributes when the user wears the smart device, based on sensors pre-installed on the smart device. These signal attributes specifically include acceleration signals, environmental electromagnetic signals, ambient light signals, and physiological rhythm signals. The system then determines whether these signal attributes detect non-periodic interference and executes corresponding steps accordingly. For example, if the system determines that these signal attributes can detect non-periodic interference, it considers that the currently collected signals exhibit random or sudden changes, possibly influenced by changes in the external environment or the device's operating state, thus disrupting the original periodic structure of the signal. The system further suppresses noise and filters interference in the collected signals, such as through filtering, outlier removal, or signal smoothing to reduce the impact of interference. Simultaneously, it can continue to collect signal data within a new time window, re-detecting the signal periodic characteristics to determine whether a stable periodic structure has been restored. Furthermore, the system can dynamically adjust signal weights based on the source of interference. For example, when the acceleration signal is affected by irregular motion, its weight as a periodic reference is reduced, and other more stable signals such as environmental electromagnetic signals or physiological rhythm signals are used as a priority reference. Conversely, if the system determines that these signal attributes do not detect non-periodic interference... When interference occurs, the system will consider that the currently acquired signal contains random or sudden changes, possibly influenced by changes in the external environment or equipment operating status, thus disrupting the original periodic structure of the signal. The system will further suppress noise and filter interference in the acquired signal, such as through filtering, outlier removal, or signal smoothing to reduce the impact of interference. Simultaneously, the system can continue to acquire signal data within a new time window, re-detecting the signal periodic characteristics to determine if a stable periodic structure has been restored. Furthermore, the system can dynamically adjust signal weights based on the source of interference. For example, when an acceleration signal is affected by irregular motion, its weight as a periodic reference is reduced, and other more stable signals such as environmental electromagnetic signals or physiological rhythm signals are prioritized as references. For instance, when the system detects non-periodic interference in these signal attributes, it will consider that the currently acquired signal only contains fixed changes. The system will then use pre-set periodic analysis by the intelligent device, specifically including time-domain periodic analysis, frequency-domain periodic analysis, and correlation periodic analysis, to extract periodic characteristics from these signal attributes. Based on different periodic characteristics, the system will identify the time interval between adjacent periods and calculate the periodic change rate of these signal attributes.By combining time-domain periodic analysis, frequency-domain periodic analysis, and correlation periodic analysis, the system can identify and verify the periodicity of signals from different perspectives. This filters out the impact of random interference or sudden noise on the signal structure to a certain extent, allowing the system to still extract valuable periodic information from the interfered signal and improving the reliability of periodic identification. Furthermore, by calculating the periodic change rate of signal attributes, the system can further understand the trend of periodic signals over time and determine whether the period is stable enough to serve as a time reference signal. A small periodic change rate indicates relatively stable signal periodicity, which can be used for subsequent time drift observation or as a time reference; a large periodic change rate indicates poor signal stability, requiring further screening or re-acquisition of the signal.

[0025] It should be noted that, through the pre-set periodic analysis of the intelligent device, the periodic features in the signal attributes are extracted. Based on the periodic features, the time interval between adjacent periods is identified, and the periodic change rate of the signal attributes is calculated. Specifically, after collecting various signal attributes from the intelligent device, the system processes the signal using a pre-set periodic analysis method to extract the periodic features. Specifically, the system can perform segmented analysis on the signal data and combine time-domain periodic analysis, frequency-domain periodic analysis, and correlation periodic analysis to identify the repetitive change structure in the signal. For example, by detecting feature points such as signal peaks, troughs, or zero-crossing points, the periodic changes of the signal can be identified in the time domain; by performing spectral analysis on the signal, the dominant frequency component with obvious energy concentration can be identified in the frequency domain; at the same time, by performing autocorrelation calculation on the signal, the existence of a stable repetitive structure in the signal can be further verified. After the system extracts the periodic features of the signal, it marks the continuous periods and identifies the time interval between adjacent periods, thereby constructing a periodic time series. Subsequently, the system calculates the periodic change rate based on the changes in the time interval between multiple continuous periods, that is, by comparing the periodic features... The difference between adjacent time intervals is analyzed to determine the degree of periodicity change over time, thus assessing the stability of the signal's periodic structure. Specific examples include: For instance, when a user is walking while wearing a smart bracelet, the device collects acceleration signals generated by human movement via an accelerometer. After periodic analysis of this acceleration signal, the system can detect the peak acceleration generated with each step and identify the time interval between two adjacent peaks as a gait cycle. For example, if the system detects time intervals of 0.80 seconds, 0.82 seconds, and 0.81 seconds between several consecutive peaks, a gait cycle sequence can be constructed based on these time intervals, and the rate of change of the cycle can be further calculated to assess whether the gait cycle remains stable. Another example is in an indoor environment where the device collects environmental electromagnetic signals and detects a power grid frequency component of approximately 50Hz through frequency domain analysis. In this case, the system can identify the time intervals between adjacent power frequency cycles and calculate the rate of change of the cycle by statistically analyzing the changes across multiple cycles. If the time intervals between these cycles are found to be largely consistent, it indicates that the signal has high periodic stability and can serve as an important basis for subsequent time drift observations or time references.

[0026] In this embodiment, step S3, which detects the operating state of the smart device and calculates the time drift value of the smart device based on the operating state, further includes: S31: identifying the disturbance vector of the operating state based on the task parameters executed by the smart device within a preset time window, wherein the task parameters specifically include task type, task duration, and task switching count; S32: determining whether the disturbance vector reaches a preset disturbance threshold; S33: if so, performing change detection on the disturbance vector to obtain the clock stability disturbance event of the smart device, using the clock stability disturbance event as a boundary to divide the continuous operating time into several drift observation time segments, collecting the cumulative clock time difference of the smart device based on the drift observation time segments, and calculating the drift gradient of the smart device based on the cumulative clock time difference.

[0027] In this embodiment, the system identifies disturbance vectors in the operating state based on task parameters executed by the intelligent device within a pre-set time window. These parameters include task type, task duration, and number of task switches. The system then determines whether these disturbance vectors reach a pre-set disturbance threshold and executes corresponding steps accordingly. For example, if the system determines that the disturbance vectors in the operating state have not reached the pre-set threshold, it considers the overall operating state of the intelligent device to be relatively stable within the current time window, and the impact of the tasks executed by the device on the system clock or operating environment is minimal. The system continues to maintain the current operating state monitoring and drift observation process, continuously collecting device task parameters based on the current time window and periodically updating the disturbance vectors in the operating state to further confirm the stability of the device's operating state. Simultaneously, the time drift data acquired within the current time window is used as valid observation data for subsequent steady-state drift calculations or clock compensation analysis. Conversely, if the system determines that the disturbance vectors in the operating state have reached the pre-set threshold, it considers the operating state of the intelligent device to be relatively stable within the current time window. When an anomaly occurs, the system detects changes in these disturbance vectors, acquires clock stability disturbance events of the intelligent device, and uses these clock stability disturbance events as boundaries to divide the continuous running time into several drift observation time segments. Based on these drift observation time segments, the system collects the cumulative time difference of the intelligent device's clock, and calculates the drift gradient of the intelligent device based on different cumulative time differences. By detecting changes in disturbance vectors and identifying clock stability disturbance events, the system can promptly identify key operating state changes that affect the stability of the device's clock, thereby avoiding the direct use of time data affected by operating state interference for drift calculation, improving the accuracy of clock stability analysis. At the same time, this method allows drift observation and data acquisition within a relatively stable operating range, thereby reducing the interference of task switching or system load changes on drift measurement results, improving the stability and data comparability of time drift observation, and through the analysis of drift gradients, the system can identify clock drift acceleration or deceleration, providing a basis for subsequent time compensation strategy adjustments, thereby improving the time synchronization stability and correction accuracy of intelligent devices in complex operating environments.

[0028] It should be noted that when the disturbance vector of the operating state reaches a preset disturbance threshold, the disturbance vector is changed and a clock stability disturbance event of the smart device is obtained. Using this clock stability disturbance event as a boundary, the continuous operating time is divided into several drift observation time segments. Based on these drift observation time segments, the cumulative clock time difference of the smart device is collected. Based on the cumulative clock time difference, the drift gradient of the smart device is calculated. Specifically, when the system detects that the disturbance vector of the operating state reaches a preset disturbance threshold, it indicates that the operating state of the smart device has fluctuated significantly within the current time window, such as frequent task switching, sudden changes in task execution time, or system computational burden. Significant load variations can impact the stability of the device's clock. Therefore, the system needs to further detect changes in the disturbance vector to identify key events affecting clock stability, i.e., clock stability disturbance events. After identifying these disturbance events, the system uses them as time boundaries to divide the continuous operating timeline of the device into multiple drift observation time segments, ensuring a relatively consistent operating state within each time segment. Subsequently, the system collects the cumulative time difference between the device's local clock and the reference time within each drift observation time segment. By statistically analyzing the changes in the time difference within different time segments, the system calculates the device's drift gradient within each time segment. The rate of change of clock drift per unit time; this method can more precisely reflect the drift trend of the device clock under different operating states, thus providing a more accurate basis for subsequent clock correction and compensation; a specific example is as follows: For example, during the operation of a smartwatch, the system detects frequent changes in task type within a certain time window, such as switching from a low-power standby task to a data synchronization task, and then to a motion data processing task, while the processor load increases significantly. At this time, the operating state disturbance vector calculated by the system exceeds the preset disturbance threshold; after detecting these disturbance changes, the system can identify the task switching moments as clock stability disturbance events and use these moments as boundaries to control the device's operation. The system divides the time frame into multiple drift observation time segments; for example, one time segment is formed during the standby phase, another during the data synchronization phase, and a new time segment is formed during the motion data processing phase. Subsequently, the system calculates the cumulative time difference between the device's local clock and the reference time within each time segment. For example, the cumulative time difference is 2 milliseconds during the standby phase, 5 milliseconds during the data synchronization phase, and 9 milliseconds during the motion data processing phase. By comparing the changes in the cumulative time difference of different time segments and combining them with the duration of each segment, the system can calculate the drift gradient of different stages, thereby analyzing the drift changes of the device clock under different operating load conditions and providing a basis for the formulation of subsequent dynamic compensation strategies.

[0029] In this embodiment, step S5, which involves collecting the predicted drift value pre-monitored by the intelligent device and detecting the drift anomaly state of the intelligent device based on the predicted drift value, further includes: S51: generating a corresponding drift residual sequence based on the difference between the predicted drift value and the actual drift observation value; S52: determining whether there are preset abnormal structural features in the drift residual sequence, wherein the abnormal structural features specifically include abrupt change points, continuous offset intervals, and periodic residual fluctuations; S53: if so, identifying the time point where the residual change rate exceeds the preset abrupt change threshold based on the residual change rate of the drift residual sequence, dynamically marking the time point as a drift abrupt change candidate point, using the drift abrupt change candidate point as the starting boundary, expanding forward and backward in the time series to obtain continuous residual deviation intervals, and constructing the corresponding drift anomaly time period.

[0030] In this embodiment, the system generates a corresponding drift residual sequence based on the difference between a pre-set drift value and the actual drift observation value. The system then determines whether these drift residual sequences contain pre-defined abnormal structural features, specifically including abrupt change points, continuous offset intervals, and periodic residual fluctuations, and executes corresponding steps accordingly. For example, if the system determines that these drift residual sequences do not contain pre-defined abnormal structural features, it considers the difference between the current predicted drift value and the actual drift observation value to be within the normal fluctuation range, and the system's drift prediction model can well reflect the actual drift changes of the device clock. The system maintains the current drift monitoring and prediction mechanism, continues to predict subsequent drift changes based on the existing drift prediction model, and continuously updates the drift residual sequence for real-time monitoring. Simultaneously, it uses the drift observation data within the current time period as valid samples to further optimize or update the drift prediction model parameters, thereby improving the model's ability to fit the device clock drift characteristics. For example, if the system determines that these drift residual sequences contain pre-defined abnormal structural features, the system considers the difference between the current predicted drift value and the actual drift observation value to be within the normal fluctuation range, and executes corresponding steps accordingly. When the difference between shifted observations falls within an abnormal range, the system identifies time points where the residual change rate exceeds a pre-set abrupt change threshold based on the residual change rate of the drift residual sequence. These time points are dynamically marked as candidate drift abrupt change points. Using these candidate points as starting boundaries, the system expands forward and backward within the time series to obtain continuous residual deviation intervals and construct corresponding drift anomaly time periods. After identifying candidate drift abrupt change points, the system uses these candidate points as time boundaries to expand forward and backward, identifying continuous residual deviation intervals and constructing corresponding drift anomaly time periods. In this way, the originally scattered abnormal residual data can be structurally analyzed, clarifying the duration and impact range of abnormal drift. This avoids misjudgments caused by relying solely on a single abnormal point, making drift anomaly identification more accurate and complete. Furthermore, by constructing drift anomaly time periods, the system can more clearly analyze the clock drift characteristics of the equipment during the abnormal period and use this time period as an important reference for subsequent anomaly handling. For example, the system can combine the operating status or environmental information within the abnormal time period to determine the cause of the anomaly and adjust the clock compensation strategy or perform anomaly correction accordingly.

[0031] It should be noted that, based on the residual change rate of the drift residual sequence, the time points where the residual change rate exceeds a preset mutation threshold are identified, and these time points are dynamically marked as candidate drift mutation points. Using these candidate drift mutation points as starting boundaries, the system expands forward and backward in the time series to obtain continuous residual deviation intervals and construct corresponding drift anomaly time periods. Specifically, after generating the drift residual sequence, the system first performs a rate of change analysis on the residual sequence to identify the trend of residual changes over time. Specifically, the system obtains the residual change rate by calculating the difference between adjacent residual values ​​or the change amplitude per unit time. The system detects a residual change rate exceeding a pre-set abrupt change threshold at a certain time point, indicating a possible sudden drift change near that time point, such as a sudden change in equipment operating status or external environmental interference. Therefore, the system dynamically marks this time point as a candidate point for a drift abrupt change. Subsequently, using this candidate point as the starting boundary, the system expands forward and backward within the time series to detect residual changes over consecutive time periods. When the system finds that the residual value continuously deviates from the normal fluctuation range over a period of time, it identifies that continuous time interval as a residual deviation. The system defines an interval and further constructs this interval as a corresponding drift anomaly time period. In this way, the system can identify the complete anomalous drift process starting from a single abrupt change point, thus more accurately describing the range of occurrence of device clock drift anomalies. A specific example is as follows: For instance, during the operation of a smartwatch, the system continuously calculates the difference between the predicted drift value and the actual drift observation value, and generates a drift residual sequence. Suppose that at a certain point in time, the residual value suddenly jumps from a stable range of ±1 milliseconds to 5 milliseconds, and the rate of change between adjacent residuals significantly exceeds the system's set abrupt change threshold. At this time, the system will... The time point is marked as a candidate point for drift mutation. Subsequently, the system expands the analysis of the residual sequence forward and backward from this time point and finds that the residual value remains in a high deviation range for a period of time starting from this time point. For example, the residual value remains between 4 and 6 milliseconds for the next 30 seconds. Based on this, the system determines this time interval as a continuous residual deviation interval and constructs it as a drift anomaly time period. Through this process, the system can not only identify the starting time of the anomaly, but also determine the duration of the anomaly, providing a basis for subsequent analysis of the cause of the anomaly or adjustment of the clock compensation strategy.

[0032] In this embodiment, step 2, which determines whether the auxiliary time observation source has a stable periodic reference signal, further includes: S21: performing sliding window analysis on the continuous periodic interval sequence pre-generated by the intelligent device to mark the corresponding candidate stable periodic intervals; S22: determining whether the candidate stable periodic intervals have multiple consecutive periodic features; S23: if so, acquiring the periodic candidate signals of the candidate stable periodic intervals, identifying the waveform similarity between adjacent periods based on the similar structure of the periodic candidate signals, and dynamically generating a confidence score for the periodic candidate signals based on the waveform similarity.

[0033] In this embodiment, the system performs sliding window analysis on the pre-generated continuous periodic interval sequences from the intelligent device to identify corresponding candidate stable periodic intervals. The system then determines whether these candidate stable periodic intervals contain multiple consecutive periodic features, and executes corresponding steps accordingly. For example, if the system determines that these candidate stable periodic intervals do not contain multiple consecutive periodic features, it considers that although some potentially periodic signal intervals have been detected in the current time period, these periodic features have not formed a continuous and stable periodic structure. The system continues to collect new periodic interval data in subsequent time windows and reconstructs the continuous periodic interval sequences. It further detects periodic features by expanding the observation time range or adjusting the sliding window parameters, and performs further preprocessing operations on the original signal, such as noise filtering, abnormal interval removal, or signal smoothing, to reduce the impact of random interference on periodic detection. Conversely, if the system determines that these candidate stable periodic intervals contain multiple consecutive periodic features, it considers that periodic signal intervals have been detected in the current time period, and that these intervals can form a continuous and stable periodic structure. The system collects candidate periodic signals from these candidate stable periodic intervals. Based on the similar structure of these candidate periodic signals, it identifies the waveform similarity between adjacent periods and dynamically generates a reliability score for the candidate periodic signals according to different waveform similarity. By comparing the waveform similarity between adjacent periods, the system can quantitatively analyze the structural consistency between periodic signals. When adjacent periodic signals maintain high consistency in waveform shape, amplitude variation, or peak position, it indicates that the periodic signal has strong stability. Conversely, it may be affected by noise or environmental changes. Through this waveform similarity analysis method, the system can more accurately identify periodic signals with truly stable characteristics, thereby improving the accuracy of periodic feature extraction. At the same time, by dynamically generating a reliability score for the candidate periodic signals based on different waveform similarity, the system can quantitatively evaluate the reliability of different periodic signals and prioritize the periodic signals with higher reliability as time reference sources. This screening mechanism based on reliability scores can reduce the impact of low-quality periodic signals on subsequent time drift calculations, thereby providing a more stable and reliable data foundation for device clock drift observation and time compensation.

[0034] It should be noted that the process involves acquiring periodic candidate signals within the candidate stable period interval, identifying waveform similarity between adjacent periods based on the similarity structure of the periodic candidate signals, and dynamically generating a reliability score for the periodic candidate signals based on the waveform similarity. Specifically, after the system detects the existence of multiple consecutive periodic features within the candidate stable period interval, it first acquires signals within that interval to form a corresponding set of periodic candidate signals. Subsequently, the system performs structural feature analysis on these periodic candidate signals and identifies their waveform similarity by comparing the waveform structures between adjacent periods. For example, the system can extract key feature points from the periodic signals, including peak positions, trough positions, and wave characteristics. The system analyzes waveform amplitude changes and periodic contour structures, and determines the similarity between adjacent periods through correlation calculation or feature point matching. After obtaining waveform similarity, the system further integrates periodic interval consistency, periodic duration, and periodic structural integrity to comprehensively evaluate the periodic signal. Periodic interval consistency determines whether the time interval between adjacent periods remains stable; periodic duration assesses whether the periodic structure can be continuously maintained in the time series; and periodic structural integrity determines whether a single periodic waveform possesses a complete characteristic structure. Based on these multiple indicators, the system comprehensively calculates the periodic candidate signal, generating a corresponding periodic signal credibility score, thereby quantifying the signal's reliability. The stability of this periodic signal provides a reliable basis for the selection of subsequent periodic reference signals and the observation of time drift. A specific example is as follows: When a user wears a smart bracelet and walks, the device collects human gait signals through an accelerometer and detects multiple consecutive gait cycles within a certain time period. The system first extracts the waveforms of these periodic signals and compares the waveform structures between adjacent cycles, finding that the peak positions and waveform shapes of the consecutive gait cycles are basically consistent, indicating a high degree of waveform similarity. Subsequently, the system further analyzes the periodic intervals of these periodic signals; for example, the time intervals for detecting consecutive gait cycles are 0.80 seconds, 0.81 seconds, 0.79 seconds, and 0.8 seconds, respectively. A time interval of 0 seconds indicates high consistency in the period intervals. Furthermore, the continuous occurrence of these periods over a relatively long period suggests good period continuity. Each period waveform also exhibits distinct peaks and troughs, demonstrating high structural integrity. The system comprehensively calculates a reliability score based on the consistency of the period intervals, period continuity, and structural integrity, for example, generating a high reliability score for this set of gait period signals. Conversely, if significant changes in the period interval occur due to sudden acceleration or stopping by the user in certain periods, or if some period waveforms are missing or distorted, the system will lower the reliability score to avoid using unstable period signals for subsequent time drift observations or time reference establishment.

[0035] In this embodiment, step S4, which determines whether the time drift value meets the preset steady-state condition, further includes: S41: obtaining the convergence characteristics of the drift evolution trajectory generated by the time drift value within a preset time period; S42: determining whether the convergence characteristics match the preset steady-state convergence; S43: if so, detecting the periodic oscillation amplitude of the drift evolution trajectory around the center value, calculating the energy distribution of the time drift value according to the periodic oscillation amplitude, collecting the distribution data of the time drift value in different frequency ranges, and dynamically mapping the time drift value to the corresponding phase change sequence according to the distribution data.

[0036] In this embodiment, the system obtains the convergence characteristics of the drift evolution trajectory generated within a preset time period based on the time drift value. The system then determines whether these convergence characteristics match a preset steady-state convergence and executes corresponding steps accordingly. For example, if the system determines that the convergence characteristics of the drift evolution trajectory do not match the preset steady-state convergence, the system considers that the time drift change of the smart device has not yet formed a stable convergence trend within the current preset time period. The system will then extend the drift observation time window and continuously collect and update the drift data in subsequent time periods to obtain... A more complete drift trajectory is obtained, and trend analysis is performed on the current drift data, such as analyzing the drift direction, amplitude, and fluctuation range, to determine whether there is drift instability caused by environmental disturbances or changes in operating status. For example, when the system determines that the convergence characteristics of the drift evolution trajectory match the pre-set steady-state convergence, the system will consider that the time drift of the smart device has formed a stable convergence trend within the current preset time period. The system will detect the periodic oscillation amplitude of the drift evolution trajectory around the center value, and perform analysis on the time drift value based on different periodic oscillation amplitudes. The system calculates energy distribution and collects distribution data of time drift values ​​across different frequency ranges. Based on this data, it dynamically maps the time drift value into a corresponding phase change sequence. By performing energy distribution calculations on the time drift value and collecting distribution data across different frequency ranges, the system can analyze the energy distribution of drift changes within different frequency ranges. This allows for the identification of the main frequency components of drift fluctuations. Some drift changes may be concentrated in the low-frequency range, reflecting the influence of slow-changing factors such as temperature changes or power fluctuations, while high-frequency fluctuations may originate from changes in equipment operating load or environmental interference. Through this frequency structure analysis, the causes and patterns of time drift can be understood more accurately. Furthermore, by dynamically mapping the time drift value into a phase change sequence based on the energy distribution across different frequency ranges, the system can convert the original time drift data into a phase form more suitable for periodic change analysis. This approach provides a more intuitive description of the drift's change process on the time axis and offers a more flexible data representation for subsequent periodic feature analysis, anomaly detection, or dynamic compensation strategies, thereby improving the accuracy and adaptability of intelligent devices' time drift monitoring and adjustment processes.

[0037] It should be noted that the system detects the periodic oscillation amplitude of the drift evolution trajectory around the center value, calculates the energy distribution of the time drift value based on the periodic oscillation amplitude, collects the distribution data of the time drift value in different frequency ranges, and dynamically maps the time drift value to the corresponding phase change sequence based on the distribution data. Specifically, after the time drift of the smart device enters the steady-state convergence stage, the system not only focuses on the overall stable trend of the drift value, but also further analyzes the small periodic oscillation characteristics of the drift value around the center value. In specific operation, the system first uses the average value or center value of the time drift sequence as a reference benchmark, and then calculates the deviation of each sampling point relative to the center value to form a periodic oscillation amplitude sequence. These oscillation amplitudes reflect the periodic fluctuations of the drift under steady-state conditions, such as fluctuations in the device's operating load, slow temperature changes, or small periodic effects caused by external environmental interference. Next, the system calculates the energy distribution of this oscillation amplitude sequence. Frequency domain analysis methods, such as Fast Fourier Transform (FFT), are typically used to map the time series to the frequency domain, thereby obtaining the energy distribution in different frequency ranges. For example, the low-frequency range reflects the slow trend of drift or environmental interference, while the high-frequency range reflects the energy distribution in different frequency ranges. The frequency range reflects the impact of internal task switching or instantaneous load changes on drift. By quantifying the energy within the frequency range, the system can assess the main periodic components of the drift signal and their contribution, thereby clarifying the source and characteristics of drift fluctuations. Finally, based on these frequency energy distributions, the system dynamically maps the time drift values ​​into a phase change sequence. The mapping process converts the amplitude of each periodic oscillation into a corresponding phase value, making the original time drift sequence exhibit periodic changes in the phase domain, thus facilitating periodic feature analysis, anomaly detection, and dynamic compensation. The phase sequence provides a more intuitive understanding than the original time sequence. The periodic information allows the system to more accurately identify drift trends and their periodic patterns, supporting the execution of subsequent steady-state compensation strategies. A specific example is as follows: For instance, when a smartwatch is worn for an extended period, the local clock drift enters a steady-state convergence state, with an average drift center of 50 milliseconds. The system first calculates the deviation of each sampling point from the center value and finds that the drift exhibits minute periodic fluctuations within a range of ±2 milliseconds, forming a periodic oscillation amplitude sequence. Subsequently, the system performs frequency domain analysis on the oscillation amplitude sequence, dividing the frequency into low frequency (<0.01Hz), mid frequency (0.01–0.1Hz), and high frequency (>0.01Hz).The analysis results show that low-frequency energy accounts for 80%, indicating that drift is mainly affected by slow temperature changes; mid-frequency energy accounts for 15%, corresponding to medium-frequency fluctuations caused by user gait or daily activities; high-frequency energy accounts for only 5%, indicating that short-term task switching has little impact on drift. Based on these frequency energy data, the system maps drift values ​​to phase change sequences. For example, each ±2 millisecond periodic fluctuation corresponds to a phase change cycle between 0 and 360 degrees, with low-frequency components producing slow phase changes and high-frequency components producing rapid phase fine-tuning. Through the phase change sequence, the system can clearly identify the periodic structure and dynamic changes of the drift signal, which can be further used for steady-state drift compensation or abnormal drift detection. For example, if an abnormal jump occurs in the phase sequence, the system can quickly trigger a drift anomaly handling strategy, thereby improving the accuracy and stability of smart device time management.

[0038] In this embodiment, the step S1, which involves obtaining the initial time difference between the local clock and the reference time of the smart device based on the initialization phase of the smart device, and establishing an auxiliary time observation source for the smart device based on the initial time difference, further includes: S11: constructing a corresponding time difference sampling sequence in chronological order within a preset time window based on the preset time reference marker point of the smart device; S12: determining whether the time difference change amplitude of the time difference sampling sequence is lower than a preset threshold; S13: if so, establishing a mapping relationship between the local clock and the reference time based on the time difference sampling sequence, and generating the initial time offset structure of the smart device based on the mapping relationship.

[0039] In this embodiment, the system constructs corresponding time difference sampling sequences in chronological order within a pre-defined time window based on pre-set time reference markers on the smart device. The system then determines whether the time difference variation amplitude of these sampling sequences is lower than a pre-set threshold to execute corresponding steps. For example, if the system determines that the time difference variation amplitude of these sampling sequences is not lower than the pre-set threshold, the system considers that the time drift of the smart device still fluctuates significantly within the current time window, and the clock has not yet reached a stable state. The system extends the observation time window and continues to collect more time difference sampling data to obtain a more complete time drift trajectory. Simultaneously, it analyzes the possible causes of abnormal fluctuations by combining device operating status or environmental monitoring data, such as identifying short-term task load fluctuations or sensor anomalies, and filters or smooths the sampling data when necessary to reduce the impact of extreme fluctuations on drift analysis. For example, if the system determines that the time difference variation amplitude of these sampling sequences is lower than the pre-set threshold, the system will... Assuming that the time drift of the smart device does not fluctuate significantly within the current time window and the clock can reach a stable state, the system establishes a mapping relationship between the local clock and the reference time based on these time difference sampling sequences. Based on this mapping relationship, it generates the initial time offset structure of the smart device. After confirming clock stability, the system can use the time difference sampling sequences to establish a mapping relationship between the local clock and the reference time. This mapping relationship accurately describes the correspondence between the device's local clock and the reference time, enabling the system to quickly and accurately convert the local clock time to the reference time or perform corrections in subsequent time synchronization or drift calculations. This mapping relationship not only provides an initial time reference but also provides a data foundation for dynamic drift monitoring and clock compensation strategies. Furthermore, based on the established mapping relationship, the system can generate the initial time offset structure of the smart device, recording the initial offset between the local clock and the reference time and its changing characteristics. Through this initial structure, the system can quickly determine the drift trend in subsequent operation and perform dynamic compensation or correction based on the initial offset.

[0040] It should be noted that, based on the time difference sampling sequence, a mapping relationship between the local clock and the reference time is established. Based on this mapping relationship, the initial time offset structure of the smart device is generated. Specifically, after the smart device clock is determined to be stable, the system uses the time difference sampling sequence to accurately model the relationship between the local clock and the reference time. Specifically, the system first associates each collected time difference sample with its corresponding local clock time point, forming a time difference-local clock mapping sequence. Based on this mapping sequence, the system can analyze the offset characteristics of the local clock, including the overall offset, drift rate, and possible small periodic fluctuations. Subsequently, the system generates the initial time offset structure of the smart device according to the mapping relationship, representing the initial deviation and its changes between the local clock time and the reference time. Features are recorded in a structured manner; this offset structure not only reflects the initial time state of the device, but also provides basic data for subsequent drift observation, time compensation, and anomaly detection. By establishing this mapping relationship, the system can quantify the correspondence between the local clock and the reference time, transforming the raw drift data into structured information that can be used for algorithm processing. For example, the offset structure can record the cumulative deviation, drift gradient, and periodic fluctuation characteristics at each time point, enabling subsequent steady-state drift calculation, predicted drift value generation, and dynamic correction operations to be performed based on real and reliable data, thereby improving the accuracy and stability of time management. A specific example is as follows: For instance, during the initialization phase of a smartwatch, the system collects the time difference between the local clock and the reference time once per second within a 10-minute time window, forming a 10-minute × 60 seconds = 600 time difference sampling points; assuming the system finds that the local clock is on average 1.2 milliseconds ahead of the reference time, and the drift fluctuation is within ±0.05 milliseconds, the system maps these sampling points to a correspondence between the local clock time and the reference time difference; based on this mapping relationship, the system generates an initial time offset structure, including an initial average offset of 1.2 milliseconds, a drift gradient of 0.01 milliseconds / second, and periodic micro-amplitude fluctuation characteristics; this structure can then be used for subsequent time drift observation. For example, in daily use, the system can compare the newly collected time difference with the initial offset structure to determine whether there is a drift anomaly, and dynamically adjust the local clock when necessary, thereby ensuring that the smartwatch's time display is highly consistent with the reference time.

[0041] Referring to Figure 2, a steady-state measurement system for time drift of an intelligent device according to an embodiment of the present invention is shown, comprising: an establishment module 10, used to obtain the initial time difference between the local clock and the reference time of the intelligent device based on a preset initialization phase of the intelligent device, and to establish an auxiliary time observation source for the intelligent device according to the initial time difference, wherein the auxiliary time observation source specifically includes environmental periodic signals, human physiological periodic signals, and internal hardware counting signals of the device; a judgment module 20, used to determine whether the auxiliary time observation source has a stable periodic reference signal; and an execution module 30, used to, if yes, identify the signal type corresponding to the stable periodic reference signal, dynamically construct the main drift observation source of the intelligent device according to the signal type, detect the operating state of the intelligent device, and through the operation... The system performs the following steps: First, it calculates the time drift value of the smart device, where the signal type specifically includes the human gait cycle and the environmental power frequency cycle. Second, it determines whether the time drift value meets a preset steady-state condition. Third, it executes the following steps: If the condition is met, it generates the steady-state drift average value of the time drift value, dynamically compensates the local clock of the smart device based on the steady-state drift average value, collects the predicted drift value pre-monitored by the smart device, detects the drift anomaly state of the smart device based on the predicted drift value, and triggers the compensation strategy of the smart device based on the drift anomaly state. Specifically, the drift anomaly state includes strong electromagnetic interference, sensor anomaly, and clock hardware failure, and the compensation strategy specifically includes a compensation period, compensation amplitude, and compensation rate.

[0042] In this embodiment, the establishment module 10, based on the pre-set initialization phase of the smart device, obtains the initial time difference between the local clock and the reference time of the smart device. Based on this initial time difference, it establishes auxiliary time observation sources for the smart device. These auxiliary time observation sources specifically include environmental periodic signals, human physiological periodic signals, and internal hardware counting signals. Then, the judgment module 20 determines whether these auxiliary time observation sources possess a stable periodic reference signal, and executes the corresponding steps accordingly. For example, if the system determines that these auxiliary time observation sources do not possess a stable periodic reference signal, the system will consider that the current environment or device state lacks a reliable time reference. For the reference periodic signal, the system will continue to collect environmental signals, human physiological signals, and internal hardware counting signals within a preset time window. It will re-detect the periodic characteristics of candidate observation signals and dynamically update the candidate set of auxiliary time observation sources. Simultaneously, it will temporarily use the historical drift model of the device's local clock or the steady-state drift average of the previous stable phase as a transitional reference to perform small-amplitude predictive compensation on the local clock to maintain the continuity of device time. For example, when the system determines that these auxiliary time observation sources have stable periodic reference signals, the execution module 30 will consider that the current environment or device state has a reliable time reference periodicity. The system identifies the signal type corresponding to the stable periodic reference signal. Specific signal types include human gait cycles and environmental power frequency cycles. Based on different signal types, it dynamically constructs the main drift observation sources for intelligent devices, detects the operating status of the intelligent devices, and calculates the time drift value of the intelligent devices based on the operating status. By identifying and determining the signal type of the stable periodic reference signal, such as the human gait cycle or environmental power frequency cycle, the system can utilize the inherent stable cycles of the natural environment or human activities as a time reference, thereby reducing the impact of local clock errors of individual devices on time measurement and improving the reliability and stability of the time reference. Simultaneously, the main drift observation source is dynamically constructed based on different signal types, enabling the device to select the most suitable time reference according to the actual operating scenario. For example, the human gait cycle is prioritized in human movement scenarios, while the environmental power frequency cycle is used in stable indoor environments. This achieves dynamic switching and adaptive construction of the drift observation source. Furthermore, after determining the main drift observation source, the time drift value is calculated by detecting the operating status of the intelligent device and combining it with the periodic reference signal, which can effectively eliminate the interference of changes in the device's operating status on clock drift. Then, the second judgment module 40 judges whether the time drift value meets the preset steady-state conditions to execute the corresponding steps.For example, when the system determines that the time drift value cannot meet the preset steady-state conditions, the system will consider that the current device clock drift state is still in a dynamic fluctuation stage, and the obtained time drift value has not yet formed a stable trend. The system will temporarily suspend the execution of local clock compensation operation and enter the continuous drift observation and stability detection stage. Within a new time window, it will continue to collect time drift data, track and analyze drift changes over continuous time periods, and continuously monitor the device's operating status to determine whether there are influencing factors such as temperature changes, power fluctuations, or task load changes. Furthermore, by continuously updating the drift observation data, it will gradually identify the drift change trend, enabling the system to obtain more stable and reliable drift estimation results. Conversely, when the system determines that the time drift value can meet the preset steady-state conditions, the second execution module 50 will consider the current device clock drift state to be in a stable stage. The system will generate a steady-state drift average value of the time drift value. Based on this steady-state drift average value, it will dynamically compensate the local clock of the intelligent device, collect the predicted drift value pre-monitored by the intelligent device, and detect abnormal drift states of the intelligent device according to different predicted drift values. Drift anomalies specifically include strong electromagnetic interference, sensor malfunctions, and clock hardware failures. Based on different drift anomaly states, compensation strategies are triggered for the intelligent devices. These strategies include compensation period, compensation amplitude, and compensation rate. By collecting predicted drift values ​​pre-monitored by the intelligent devices, the system can anticipate the development trend of clock drift and compare the predicted results with the current drift state, thereby enabling early identification of potential drift anomalies. This monitoring method based on predicted drift values ​​can detect drift anomalies before they significantly affect the device's time accuracy, improving the system's perception and response speed to clock drift changes. Furthermore, by detecting the presence of strong electromagnetic interference, sensor malfunctions, or clock hardware failures based on different predicted drift values ​​and triggering corresponding compensation strategies for different anomalies, targeted handling of different types of anomalies can be achieved. Moreover, by dynamically adjusting the compensation period, compensation amplitude, and compensation rate, the time compensation strategy can adaptively adjust according to the actual operating environment, thereby improving the time stability of the intelligent devices and the system reliability in complex operating environments.

[0043] In this embodiment, the device further includes: a data acquisition module, used to acquire signal attributes of the user wearing the smart device based on sensors preset in the smart device, wherein the signal attributes specifically include acceleration signals, environmental electromagnetic signals, ambient light signals, and physiological rhythm signals; a third judgment module, used to determine whether the signal attributes detect non-periodic interference; and a third execution module, used to, if not, extract periodic features from the signal attributes through periodic analysis preset in the smart device, identify the time interval between adjacent periods based on the periodic features, and calculate the periodic change rate of the signal attributes, wherein the periodic analysis specifically includes time-domain periodic analysis, frequency-domain periodic analysis, and correlation periodic analysis.

[0044] In this embodiment, the system collects signal attributes when the user wears the smart device, based on sensors pre-installed on the smart device. These signal attributes specifically include acceleration signals, environmental electromagnetic signals, ambient light signals, and physiological rhythm signals. The system then determines whether these signal attributes detect non-periodic interference and executes corresponding steps accordingly. For example, if the system determines that these signal attributes can detect non-periodic interference, it considers that the currently collected signals exhibit random or sudden changes, possibly influenced by changes in the external environment or the device's operating state, thus disrupting the original periodic structure of the signal. The system further suppresses noise and filters interference in the collected signals, such as through filtering, outlier removal, or signal smoothing to reduce the impact of interference. Simultaneously, it can continue to collect signal data within a new time window, re-detecting the signal periodic characteristics to determine whether a stable periodic structure has been restored. Furthermore, the system can dynamically adjust signal weights based on the source of interference. For example, when the acceleration signal is affected by irregular motion, its weight as a periodic reference is reduced, and other more stable signals such as environmental electromagnetic signals or physiological rhythm signals are used as a priority reference. Conversely, if the system determines that these signal attributes do not detect non-periodic interference... When interference occurs, the system will consider that the currently acquired signal contains random or sudden changes, possibly influenced by changes in the external environment or equipment operating status, thus disrupting the original periodic structure of the signal. The system will further suppress noise and filter interference in the acquired signal, such as through filtering, outlier removal, or signal smoothing to reduce the impact of interference. Simultaneously, the system can continue to acquire signal data within a new time window, re-detecting the signal periodic characteristics to determine if a stable periodic structure has been restored. Furthermore, the system can dynamically adjust signal weights based on the source of interference. For example, when an acceleration signal is affected by irregular motion, its weight as a periodic reference is reduced, and other more stable signals such as environmental electromagnetic signals or physiological rhythm signals are prioritized as references. For instance, when the system detects non-periodic interference in these signal attributes, it will consider that the currently acquired signal only contains fixed changes. The system will then use pre-set periodic analysis by the intelligent device, specifically including time-domain periodic analysis, frequency-domain periodic analysis, and correlation periodic analysis, to extract periodic characteristics from these signal attributes. Based on different periodic characteristics, the system will identify the time interval between adjacent periods and calculate the periodic change rate of these signal attributes.By combining time-domain periodic analysis, frequency-domain periodic analysis, and correlation periodic analysis, the system can identify and verify the periodicity of signals from different perspectives. This filters out the impact of random interference or sudden noise on the signal structure to a certain extent, allowing the system to still extract valuable periodic information from the interfered signal and improving the reliability of periodic identification. Furthermore, by calculating the periodic change rate of signal attributes, the system can further understand the trend of periodic signals over time and determine whether the period is stable enough to serve as a time reference signal. A small periodic change rate indicates relatively stable signal periodicity, which can be used for subsequent time drift observation or as a time reference; a large periodic change rate indicates poor signal stability, requiring further screening or re-acquisition of the signal.

[0045] In this embodiment, the execution module further includes: an identification unit, used to identify the disturbance vector of the running state based on the task parameters executed by the smart device within a preset time window, wherein the task parameters specifically include task type, task duration, and task switching count; a judgment unit, used to judge whether the disturbance vector reaches a preset disturbance threshold; and an execution unit, used to perform change detection on the disturbance vector if so, obtain the clock stability disturbance event of the smart device, divide the continuous running time into several drift observation time segments using the clock stability disturbance event as a boundary, collect the cumulative clock time difference of the smart device according to the drift observation time segments, and calculate the drift gradient of the smart device based on the cumulative clock time difference.

[0046] In this embodiment, the system identifies disturbance vectors in the operating state based on task parameters executed by the intelligent device within a pre-set time window. These parameters include task type, task duration, and number of task switches. The system then determines whether these disturbance vectors reach a pre-set disturbance threshold and executes corresponding steps accordingly. For example, if the system determines that the disturbance vectors in the operating state have not reached the pre-set threshold, it considers the overall operating state of the intelligent device to be relatively stable within the current time window, and the impact of the tasks executed by the device on the system clock or operating environment is minimal. The system continues to maintain the current operating state monitoring and drift observation process, continuously collecting device task parameters based on the current time window and periodically updating the disturbance vectors in the operating state to further confirm the stability of the device's operating state. Simultaneously, the time drift data acquired within the current time window is used as valid observation data for subsequent steady-state drift calculations or clock compensation analysis. Conversely, if the system determines that the disturbance vectors in the operating state have reached the pre-set threshold, it considers the operating state of the intelligent device to be relatively stable within the current time window. When an anomaly occurs, the system detects changes in these disturbance vectors, acquires clock stability disturbance events of the intelligent device, and uses these clock stability disturbance events as boundaries to divide the continuous running time into several drift observation time segments. Based on these drift observation time segments, the system collects the cumulative time difference of the intelligent device's clock, and calculates the drift gradient of the intelligent device based on different cumulative time differences. By detecting changes in disturbance vectors and identifying clock stability disturbance events, the system can promptly identify key operating state changes that affect the stability of the device's clock, thereby avoiding the direct use of time data affected by operating state interference for drift calculation, improving the accuracy of clock stability analysis. At the same time, this method allows drift observation and data acquisition within a relatively stable operating range, thereby reducing the interference of task switching or system load changes on drift measurement results, improving the stability and data comparability of time drift observation, and through the analysis of drift gradients, the system can identify clock drift acceleration or deceleration, providing a basis for subsequent time compensation strategy adjustments, thereby improving the time synchronization stability and correction accuracy of intelligent devices in complex operating environments.

[0047] In this embodiment, the second execution module further includes: a generation unit, used to generate a corresponding drift residual sequence based on the difference between the predicted drift value and the actual drift observation value; a second judgment unit, used to judge whether there are preset abnormal structural features in the drift residual sequence, wherein the abnormal structural features specifically include abrupt change points, continuous offset intervals, and periodic residual fluctuations; and a second execution unit, used to, if so, identify the time point where the residual change rate exceeds a preset abrupt change threshold based on the residual change rate of the drift residual sequence, dynamically mark the time point as a drift abrupt change candidate point, use the drift abrupt change candidate point as the starting boundary, expand forward and backward in the time series, obtain continuous residual deviation intervals, and construct the corresponding drift abrupt change time period.

[0048] In this embodiment, the system generates a corresponding drift residual sequence based on the difference between a pre-set drift value and the actual drift observation value. The system then determines whether these drift residual sequences contain pre-defined abnormal structural features, specifically including abrupt change points, continuous offset intervals, and periodic residual fluctuations, and executes corresponding steps accordingly. For example, if the system determines that these drift residual sequences do not contain pre-defined abnormal structural features, it considers the difference between the current predicted drift value and the actual drift observation value to be within the normal fluctuation range, and the system's drift prediction model can well reflect the actual drift changes of the device clock. The system maintains the current drift monitoring and prediction mechanism, continues to predict subsequent drift changes based on the existing drift prediction model, and continuously updates the drift residual sequence for real-time monitoring. Simultaneously, it uses the drift observation data within the current time period as valid samples to further optimize or update the drift prediction model parameters, thereby improving the model's ability to fit the device clock drift characteristics. For example, if the system determines that these drift residual sequences contain pre-defined abnormal structural features, the system considers the difference between the current predicted drift value and the actual drift observation value to be within the normal fluctuation range, and executes corresponding steps accordingly. When the difference between shifted observations falls within an abnormal range, the system identifies time points where the residual change rate exceeds a pre-set abrupt change threshold based on the residual change rate of the drift residual sequence. These time points are dynamically marked as candidate drift abrupt change points. Using these candidate points as starting boundaries, the system expands forward and backward within the time series to obtain continuous residual deviation intervals and construct corresponding drift anomaly time periods. After identifying candidate drift abrupt change points, the system uses these candidate points as time boundaries to expand forward and backward, identifying continuous residual deviation intervals and constructing corresponding drift anomaly time periods. In this way, the originally scattered abnormal residual data can be structurally analyzed, clarifying the duration and impact range of abnormal drift. This avoids misjudgments caused by relying solely on a single abnormal point, making drift anomaly identification more accurate and complete. Furthermore, by constructing drift anomaly time periods, the system can more clearly analyze the clock drift characteristics of the equipment during the abnormal period and use this time period as an important reference for subsequent anomaly handling. For example, the system can combine the operating status or environmental information within the abnormal time period to determine the cause of the anomaly and adjust the clock compensation strategy or perform anomaly correction accordingly.

[0049] In this embodiment, the judgment module further includes: a marking unit, used to perform sliding window analysis on the continuous periodic interval sequence pre-generated by the smart device, and mark the corresponding candidate stable periodic interval; a third judgment unit, used to determine whether the candidate stable periodic interval has multiple consecutive periodic features; and a third execution unit, used to, if so, collect the periodic candidate signal of the candidate stable periodic interval, identify the waveform similarity between adjacent periods according to the similar structure of the periodic candidate signal, and dynamically generate a credibility score of the periodic candidate signal based on the waveform similarity.

[0050] In this embodiment, the system performs sliding window analysis on the pre-generated continuous periodic interval sequences from the intelligent device to identify corresponding candidate stable periodic intervals. The system then determines whether these candidate stable periodic intervals contain multiple consecutive periodic features, and executes corresponding steps accordingly. For example, if the system determines that these candidate stable periodic intervals do not contain multiple consecutive periodic features, it considers that although some potentially periodic signal intervals have been detected in the current time period, these periodic features have not formed a continuous and stable periodic structure. The system continues to collect new periodic interval data in subsequent time windows and reconstructs the continuous periodic interval sequences. It further detects periodic features by expanding the observation time range or adjusting the sliding window parameters, and performs further preprocessing operations on the original signal, such as noise filtering, abnormal interval removal, or signal smoothing, to reduce the impact of random interference on periodic detection. Conversely, if the system determines that these candidate stable periodic intervals contain multiple consecutive periodic features, it considers that periodic signal intervals have been detected in the current time period, and that these intervals can form a continuous and stable periodic structure. The system collects candidate periodic signals from these candidate stable periodic intervals. Based on the similar structure of these candidate periodic signals, it identifies the waveform similarity between adjacent periods and dynamically generates a reliability score for the candidate periodic signals according to different waveform similarity. By comparing the waveform similarity between adjacent periods, the system can quantitatively analyze the structural consistency between periodic signals. When adjacent periodic signals maintain high consistency in waveform shape, amplitude variation, or peak position, it indicates that the periodic signal has strong stability. Conversely, it may be affected by noise or environmental changes. Through this waveform similarity analysis method, the system can more accurately identify periodic signals with truly stable characteristics, thereby improving the accuracy of periodic feature extraction. At the same time, by dynamically generating a reliability score for the candidate periodic signals based on different waveform similarity, the system can quantitatively evaluate the reliability of different periodic signals and prioritize the periodic signals with higher reliability as time reference sources. This screening mechanism based on reliability scores can reduce the impact of low-quality periodic signals on subsequent time drift calculations, thereby providing a more stable and reliable data foundation for device clock drift observation and time compensation.

[0051] In this embodiment, the second judgment module further includes: an acquisition unit, used to acquire the convergence characteristics of the drift evolution trajectory generated by the time drift value within a preset time period; a fourth judgment unit, used to determine whether the convergence characteristics match a preset steady-state convergence; and a fourth execution unit, used to detect the periodic oscillation amplitude of the drift evolution trajectory around the center value if yes, perform energy distribution calculation on the time drift value according to the periodic oscillation amplitude, collect distribution data of the time drift value in different frequency ranges, and dynamically map the time drift value into a corresponding phase change sequence based on the distribution data.

[0052] In this embodiment, the system obtains the convergence characteristics of the drift evolution trajectory generated within a preset time period based on the time drift value. The system then determines whether these convergence characteristics match a preset steady-state convergence and executes corresponding steps accordingly. For example, if the system determines that the convergence characteristics of the drift evolution trajectory do not match the preset steady-state convergence, the system considers that the time drift change of the smart device has not yet formed a stable convergence trend within the current preset time period. The system will then extend the drift observation time window and continuously collect and update the drift data in subsequent time periods to obtain... A more complete drift trajectory is obtained, and trend analysis is performed on the current drift data, such as analyzing the drift direction, amplitude, and fluctuation range, to determine whether there is drift instability caused by environmental disturbances or changes in operating status. For example, when the system determines that the convergence characteristics of the drift evolution trajectory match the pre-set steady-state convergence, the system will consider that the time drift of the smart device has formed a stable convergence trend within the current preset time period. The system will detect the periodic oscillation amplitude of the drift evolution trajectory around the center value, and perform analysis on the time drift value based on different periodic oscillation amplitudes. The system calculates energy distribution and collects distribution data of time drift values ​​across different frequency ranges. Based on this data, it dynamically maps the time drift value into a corresponding phase change sequence. By performing energy distribution calculations on the time drift value and collecting distribution data across different frequency ranges, the system can analyze the energy distribution of drift changes within different frequency ranges. This allows for the identification of the main frequency components of drift fluctuations. Some drift changes may be concentrated in the low-frequency range, reflecting the influence of slow-changing factors such as temperature changes or power fluctuations, while high-frequency fluctuations may originate from changes in equipment operating load or environmental interference. Through this frequency structure analysis, the causes and patterns of time drift can be understood more accurately. Furthermore, by dynamically mapping the time drift value into a phase change sequence based on the energy distribution across different frequency ranges, the system can convert the original time drift data into a phase form more suitable for periodic change analysis. This approach provides a more intuitive description of the drift's change process on the time axis and offers a more flexible data representation for subsequent periodic feature analysis, anomaly detection, or dynamic compensation strategies, thereby improving the accuracy and adaptability of intelligent devices' time drift monitoring and adjustment processes.

[0053] In this embodiment, the establishment module further includes: a construction unit, used to construct a corresponding time difference sampling sequence in chronological order within a preset time window based on a preset time reference marker point of the smart device; a fifth judgment unit, used to determine whether the time difference change amplitude of the time difference sampling sequence is lower than a preset threshold; and a fifth execution unit, used to, if so, establish a mapping relationship between the local clock and the reference time according to the time difference sampling sequence, and generate the initial time offset structure of the smart device according to the mapping relationship.

[0054] In this embodiment, the system constructs corresponding time difference sampling sequences in chronological order within a pre-defined time window based on pre-set time reference markers on the smart device. The system then determines whether the time difference variation amplitude of these sampling sequences is lower than a pre-set threshold to execute corresponding steps. For example, if the system determines that the time difference variation amplitude of these sampling sequences is not lower than the pre-set threshold, the system considers that the time drift of the smart device still fluctuates significantly within the current time window, and the clock has not yet reached a stable state. The system extends the observation time window and continues to collect more time difference sampling data to obtain a more complete time drift trajectory. Simultaneously, it analyzes the possible causes of abnormal fluctuations by combining device operating status or environmental monitoring data, such as identifying short-term task load fluctuations or sensor anomalies, and filters or smooths the sampling data when necessary to reduce the impact of extreme fluctuations on drift analysis. For example, if the system determines that the time difference variation amplitude of these sampling sequences is lower than the pre-set threshold, the system will... Assuming that the time drift of the smart device does not fluctuate significantly within the current time window and the clock can reach a stable state, the system establishes a mapping relationship between the local clock and the reference time based on these time difference sampling sequences. Based on this mapping relationship, it generates the initial time offset structure of the smart device. After confirming clock stability, the system can use the time difference sampling sequences to establish a mapping relationship between the local clock and the reference time. This mapping relationship accurately describes the correspondence between the device's local clock and the reference time, enabling the system to quickly and accurately convert the local clock time to the reference time or perform corrections in subsequent time synchronization or drift calculations. This mapping relationship not only provides an initial time reference but also provides a data foundation for dynamic drift monitoring and clock compensation strategies. Furthermore, based on the established mapping relationship, the system can generate the initial time offset structure of the smart device, recording the initial offset between the local clock and the reference time and its changing characteristics. Through this initial structure, the system can quickly determine the drift trend in subsequent operation and perform dynamic compensation or correction based on the initial offset.

[0055] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A steady-state measurement method for time drift of an intelligent device, characterized in that, Includes the following steps: Based on the preset initialization phase of the smart device, the initial time difference between the local clock and the reference time of the smart device is obtained. Based on this initial time difference, an auxiliary time observation source for the smart device is established. Specifically, the auxiliary time observation source includes environmental periodic signals, human physiological periodic signals, and internal hardware counting signals. It is determined whether the auxiliary time observation source possesses a stable periodic reference signal. If so, the signal type corresponding to the stable periodic reference signal is identified. Based on the signal type, the main drift observation source for the smart device is dynamically constructed. The operating state of the smart device is detected, and the time drift value of the smart device is calculated based on the operating state. In this process, the signal types specifically include human gait cycles and environmental power frequency cycles; it is determined whether the time drift value meets preset steady-state conditions; if so, a steady-state drift average value of the time drift value is generated; based on the steady-state drift average value, the local clock of the smart device is dynamically compensated; the predicted drift value pre-monitored by the smart device is collected; based on the predicted drift value, the drift abnormal state of the smart device is detected; and based on the drift abnormal state, the compensation strategy of the smart device is triggered. Specifically, the drift abnormal state includes strong electromagnetic interference, sensor abnormality, and clock hardware failure; and the compensation strategy specifically includes compensation period, compensation amplitude, and compensation rate.

2. The steady-state measurement method for time drift of an intelligent device according to claim 1, characterized in that, Before the step of identifying the signal type corresponding to the stable periodic reference signal and dynamically constructing the main drift observation source of the smart device based on the signal type, the method further includes: collecting signal attributes when the user wears the smart device based on the sensors preset by the smart device, wherein the signal attributes specifically include acceleration signals, environmental electromagnetic signals, ambient light signals, and physiological rhythm signals; determining whether the signal attributes detect non-periodic interference; if not, extracting periodic features from the signal attributes through the periodic analysis preset by the smart device, identifying the time interval between adjacent periods based on the periodic features, and calculating the periodic change rate of the signal attributes, wherein the periodic analysis specifically includes time domain periodic analysis, frequency domain periodic analysis, and correlation periodic analysis.

3. The steady-state measurement method for time drift of an intelligent device according to claim 1, characterized in that, The step of detecting the operating status of the smart device and calculating the time drift value of the smart device based on the operating status further includes: identifying a disturbance vector of the operating status based on task parameters executed by the smart device within a preset time window, wherein the task parameters specifically include task type, task duration, and number of task switching; determining whether the disturbance vector reaches a preset disturbance threshold; if so, performing change detection on the disturbance vector to obtain a clock stability disturbance event of the smart device; using the clock stability disturbance event as a boundary to divide the continuous operating time into several drift observation time segments; collecting the cumulative time difference of the smart device's clock based on the drift observation time segments; and calculating the drift gradient of the smart device based on the cumulative time difference of the clock.

4. The steady-state measurement method for time drift of an intelligent device according to claim 1, characterized in that, The step of collecting the predicted drift value pre-monitored by the intelligent device and detecting the drift anomaly state of the intelligent device based on the predicted drift value further includes: generating a corresponding drift residual sequence based on the difference between the predicted drift value and the actual drift observation value; determining whether there are preset abnormal structural features in the drift residual sequence, wherein the abnormal structural features specifically include abrupt change points, continuous offset intervals, and periodic residual fluctuations; if so, identifying the time point where the residual change rate exceeds the preset abrupt change threshold based on the residual change rate of the drift residual sequence, dynamically marking the time point as a drift abrupt change candidate point, using the drift abrupt change candidate point as the starting boundary, expanding forward and backward in the time series to obtain continuous residual deviation intervals, and constructing the corresponding drift anomaly time period.

5. The steady-state measurement method for time drift of an intelligent device according to claim 1, characterized in that, The step of determining whether the auxiliary time observation source has a stable periodic reference signal further includes: performing a sliding window analysis on the continuous periodic interval sequence pre-generated by the intelligent device to mark the corresponding candidate stable periodic intervals; determining whether the candidate stable periodic interval has multiple consecutive periodic features; if so, collecting the periodic candidate signals of the candidate stable periodic intervals, identifying the waveform similarity between adjacent periods based on the similar structure of the periodic candidate signals, and dynamically generating a confidence score for the periodic candidate signals based on the waveform similarity.

6. The steady-state measurement method for time drift of an intelligent device according to claim 1, characterized in that, The step of determining whether the time drift value meets the preset steady-state conditions further includes: obtaining the convergence characteristics of the drift evolution trajectory generated by the time drift value within a preset time period; determining whether the convergence characteristics match the preset steady-state convergence; if so, detecting the periodic oscillation amplitude of the drift evolution trajectory around the center value, calculating the energy distribution of the time drift value according to the periodic oscillation amplitude, collecting the distribution data of the time drift value in different frequency ranges, and dynamically mapping the time drift value to the corresponding phase change sequence according to the distribution data.

7. The steady-state measurement method for time drift of an intelligent device according to claim 1, characterized in that, The step of obtaining the initial time difference between the local clock and the reference time of the smart device in the initialization phase based on the smart device, and establishing an auxiliary time observation source for the smart device based on the initial time difference, further includes: constructing a corresponding time difference sampling sequence in chronological order within a preset time window based on the preset time reference marker point of the smart device; determining whether the time difference change amplitude of the time difference sampling sequence is lower than a preset threshold; if so, establishing a mapping relationship between the local clock and the reference time based on the time difference sampling sequence, and generating the initial time offset structure of the smart device based on the mapping relationship.

8. A steady-state measurement system for time drift of an intelligent device, characterized in that, include: A module is established to acquire the initial time difference between the local clock and the reference time of the smart device based on a preset initialization phase. Based on this initial time difference, an auxiliary time observation source for the smart device is established, specifically including environmental periodic signals, human physiological periodic signals, and internal hardware counting signals. A judgment module is used to determine whether the auxiliary time observation source possesses a stable periodic reference signal. An execution module is used to, if so, identify the signal type corresponding to the stable periodic reference signal, dynamically construct the main drift observation source for the smart device based on the signal type, detect the operating status of the smart device, and calculate the time drift value of the smart device based on the operating status. The signal types specifically include human gait cycles and environmental power frequency cycles; the second judgment module is used to determine whether the time drift value meets the preset steady-state conditions; the second execution module is used to generate the steady-state drift average value of the time drift value if the conditions are met, dynamically compensate the local clock of the smart device based on the steady-state drift average value, collect the predicted drift value pre-monitored by the smart device, detect the drift abnormal state of the smart device according to the predicted drift value, and trigger the compensation strategy of the smart device according to the drift abnormal state. The drift abnormal state specifically includes strong electromagnetic interference, sensor abnormality, and clock hardware failure, and the compensation strategy specifically includes compensation period, compensation amplitude, and compensation rate.

9. The steady-state measurement system for time drift of an intelligent device according to claim 8, characterized in that, Also includes: The acquisition module is used to acquire signal attributes when the user wears the smart device based on the sensors preset in the smart device. The signal attributes specifically include acceleration signals, environmental electromagnetic signals, ambient light signals, and physiological rhythm signals. The third judgment module is used to determine whether the signal attributes detect non-periodic interference. The third execution module is used to extract periodic features from the signal attributes through the periodic analysis preset in the smart device if no interference is detected. Based on the periodic features, the time interval between adjacent periods is identified, and the periodic change rate of the signal attributes is calculated. The periodic analysis specifically includes time-domain periodic analysis, frequency-domain periodic analysis, and correlation periodic analysis.

10. The steady-state measurement system for time drift of an intelligent device according to claim 8, characterized in that, The execution module further includes: an identification unit, used to identify the disturbance vector of the running state based on the task parameters executed by the smart device within a preset time window, wherein the task parameters specifically include task type, task duration, and task switching count; a judgment unit, used to judge whether the disturbance vector reaches a preset disturbance threshold; and an execution unit, used to perform change detection on the disturbance vector if so, obtain the clock stability disturbance event of the smart device, divide the continuous running time into several drift observation time segments using the clock stability disturbance event as a boundary, collect the cumulative clock time difference of the smart device according to the drift observation time segments, and calculate the drift gradient of the smart device based on the cumulative clock time difference.

Images