Clock power saving scheduling method and system based on multi-source energy consumption data analysis

Abstract

The application relates to the technical field of data analysis, and particularly discloses a watch power-saving scheduling method and system based on multi-source energy consumption data analysis, which comprises the following steps: identifying smooth power consumption intervals and fluctuation power consumption intervals in adjacent time points and corresponding static background features and dynamic energy consumption mode labels through changes in power consumption in energy consumption data of a watch, so as to construct a time period energy consumption image of the watch; generating three-level power-saving scheduling parameters of the watch by combining real-time ambient brightness change trends and user interaction prediction probabilities of the watch within a preset short time window after the current time; applying the three-level power-saving scheduling parameters to the watch, and after a preset scheduling period, taking newly collected actual energy consumption data of the watch as feedback input to update the time period energy consumption image; and the application can improve the power-saving scheduling efficiency of the watch.

Description

Technical Field

[0001] This invention relates to the field of data analysis technology, and in particular to a method and system for energy-saving scheduling of clocks based on multi-source energy consumption data analysis. Background Technology

[0002] Traditional power-saving scheduling methods for clocks often employ fixed, single-dimensional strategies, such as reducing display brightness based on a fixed time threshold or simply relying on remaining battery power to enter sleep mode. These methods fail to comprehensively consider the multi-source energy consumption characteristics of clocks at different times of day, leading to premature entry into low-power states during peak user interaction periods, impacting user experience, or maintaining high refresh rates in low-light conditions, resulting in unnecessary power waste. Existing technologies are ineffective in terms of scheduling performance, failing to adapt to dynamic changes in time, ambient brightness, and user behavior.

[0003] Another existing approach attempts to trigger power-saving modes by detecting user interaction actions. However, this method only focuses on the current interaction event and lacks the ability to analyze historical energy consumption data or predict future energy consumption trends. When the clock is in a stable power consumption range, existing technology cannot identify the static background characteristics of this range and still uses a uniform strategy to schedule energy consumption. When the clock is in a fluctuating power consumption range, existing technology also struggles to dynamically adjust power-saving parameters based on abrupt changes and gradients in the fluctuation pattern, resulting in delayed or overly aggressive scheduling actions and a poor balance between overall power saving effect and user experience. Summary of the Invention

[0004] In view of the above problems, the purpose of this invention is to provide a clock power-saving scheduling method and system based on multi-source energy consumption data analysis, so as to solve the problem that the existing clock power-saving scheduling methods, due to the use of a single fixed strategy and the lack of time-domain analysis and pattern recognition of multi-source energy consumption data, cannot distinguish between stable power consumption ranges and fluctuating power consumption ranges, resulting in an imbalance between power-saving effect and user experience.

[0005] The present invention provides a clock energy-saving scheduling method based on multi-source energy consumption data analysis, comprising:

[0006] Step 1: By analyzing the changes in power consumption in the clock's energy consumption data, identify the stable power consumption range and fluctuating power consumption range within adjacent time points, as well as the corresponding static background features and dynamic energy consumption pattern labels, in order to construct a time period energy consumption profile of the clock.

[0007] Step 2: Based on the static background features of the clock at the current moment in the energy consumption profile of the time period, combined with the real-time ambient brightness change trend and user interaction prediction probability within a preset short window after the current moment, generate the three-level power-saving scheduling parameters of the clock.

[0008] Step 3: Apply the three-level power-saving scheduling parameters to the clock. After a preset scheduling cycle, use the newly collected actual energy consumption data of the clock as feedback input to update the energy consumption profile for the specified time period.

[0009] Preferably, the process of identifying stable power consumption intervals and fluctuating power consumption intervals within adjacent time points, along with corresponding static background features and dynamic energy consumption pattern labels, based on changes in power consumption data from the clock, is as follows:

[0010] Based on the multi-source energy consumption data of the clocks aligned by timestamps, the rate of change of power consumption rate at adjacent time points is calculated. The formula for calculating the rate of change of power consumption rate is as follows:

[0011] ;

[0012] In the formula, This represents the rate of change in the rate of electricity consumption at adjacent time points. This indicates the rate of power consumption at the previous time point in the adjacent time points. This indicates the rate of power consumption at the next time point in the adjacent time intervals. This represents the time interval between the adjacent time points;

[0013] When the rate of change of the power consumption rate is lower than the preset rate of change threshold, the corresponding time interval is marked as a stable power consumption interval, and the other time intervals are marked as fluctuating power consumption intervals.

[0014] The average ambient brightness value and average user interaction frequency within the stable power consumption range are used as the static background features of the clock.

[0015] The peak power consumption rate, the gradient of ambient brightness value change, and the abrupt change point of user interaction frequency within the fluctuating power consumption range are correlated and mapped to generate a dynamic energy consumption mode label for the clock.

[0016] Preferably, the process of constructing the time-period energy consumption profile of the clock is as follows:

[0017] Using the average ambient brightness value in the static background features as the base layer, and the average user interaction frequency as the weight coefficient of the base layer, the steady-state energy consumption base plate of the clock is laid out.

[0018] Each feature value in the dynamic energy consumption mode label is anchored to the corresponding position on the time axis of the steady-state energy consumption base plate to correct the steady-state energy consumption base plate.

[0019] The corrected steady-state energy consumption base plate, together with the embedded dynamic energy consumption mode label, is packaged and encapsulated into the time period energy consumption profile of the clock.

[0020] Preferably, the process of correcting the steady-state energy consumption base plate is as follows:

[0021] Using the time point corresponding to the peak power consumption rate as the anchor point, the area on the steady-state energy consumption base plate located within a preset range before and after the anchor point is expanded outward along the vertical axis in proportion to the amplitude of the peak power consumption rate.

[0022] Using the direction of the ambient brightness value change gradient as the distortion vector, a shear transformation proportional to the magnitude of the ambient brightness value change gradient is applied to the edge contour of the steady-state energy consumption base plate along the time axis direction of the steady-state energy consumption base plate.

[0023] Using the position of the user interaction frequency mutation point on the time axis as the cutting line, the steady-state energy consumption base plate is divided into multiple independent sub-blocks.

[0024] Preferably, the real-time ambient brightness change trend is as follows:

[0025] Within a preset short window after the current time of the clock, the ambient light sensor of the clock is woken up sequentially at a fixed sampling interval to read the original ambient brightness reading at each sampling time, so as to form a sequence of discrete brightness points scattered along the time axis of the clock.

[0026] The values ​​in the discrete brightness point sequence are compared sequentially. If the subsequent value is greater than the previous value, it is marked as an increase; if it is less than the previous value, it is marked as a decrease; if it is equal to the previous value, it is marked as a level.

[0027] The number of all markers in the discrete brightness point sequence is counted, and the direction of change corresponding to the marker type with the highest proportion is taken as the real-time ambient brightness change trend of the clock.

[0028] Preferably, the process for predicting the probability of user interaction is as follows:

[0029] Extract all historical interaction moments that belong to the same time period type as the current moment from the clock's historical operation log;

[0030] The time interval between every two historical interaction moments is mapped onto the time axis to form an interval distribution histogram with the time interval as the horizontal axis and the frequency of occurrence of the time interval as the vertical axis.

[0031] Using the time elapsed since the most recent actual user interaction action from the current time of the clock as the query index, the frequency of occurrence located in the interval distribution histogram is divided by the total number of time intervals to obtain the predicted probability of user interaction of the clock.

[0032] Preferably, the process of generating the three-level power-saving scheduling parameters for the clock is as follows:

[0033] The average ambient brightness value in the static background features is superimposed on the real-time ambient brightness change trend to obtain the backlight brightness attenuation coefficient of the clock.

[0034] The product of the average user interaction frequency in the static background features and the predicted probability of user interaction is used as the display refresh rate reduction ratio of the clock.

[0035] Using the larger of the backlight brightness attenuation coefficient and the display refresh rate reduction ratio as a benchmark, the benchmark is mapped in stages according to the current remaining power of the clock to obtain the sensor sampling interval extension factor of the clock.

[0036] The backlight brightness attenuation coefficient, the display refresh rate reduction ratio, and the sensor sampling interval extension factor are integrated into the three-level power-saving scheduling parameters of the clock.

[0037] Preferably, the process of applying the three-level power-saving scheduling parameters to the clock is as follows:

[0038] Based on the display refresh rate reduction ratio, remove the corresponding ratio of frame refresh instructions from the display driver instruction stream of the clock.

[0039] Using the backlight brightness attenuation coefficient as the duty cycle adjustment factor for the clock pulse width modulation signal, the output brightness of the clock is reduced.

[0040] According to the sensor sampling interval extension factor, the trigger time for the next data acquisition of the built-in ambient light sensor and accelerometer in the watch is postponed.

[0041] Preferably, after a preset scheduling period, the newly collected actual energy consumption data of the clock is used as feedback input to update the energy consumption profile for the specified time period. The process is as follows:

[0042] After a preset scheduling period, the actual energy consumption data within the scheduling period is collected again to construct the actual energy consumption feature vector of the clock.

[0043] The actual energy consumption feature vector is weighted and averaged with the static background features of the corresponding time period in the time period energy consumption profile, and the weighted average result is used to replace the original static background features.

[0044] The fluctuation portion in the actual energy consumption feature vector is compared with the original dynamic energy consumption pattern label. If the deviation exceeds a preset deviation threshold, the original label is replaced with the dynamic energy consumption pattern label extracted from the actual energy consumption feature vector.

[0045] This invention also provides a power-saving scheduling system for clocks based on multi-source energy consumption data analysis, the system comprising:

[0046] The profile building module is used to identify stable power consumption ranges and fluctuating power consumption ranges within adjacent time points, as well as corresponding static background features and dynamic energy consumption pattern labels, by analyzing the power consumption changes in the energy consumption data of the clock, so as to construct the time period energy consumption profile of the clock.

[0047] The parameter generation module is used to generate three-level power-saving scheduling parameters for the clock based on the static background features of the clock at the current moment in the energy consumption profile of the time period, combined with the real-time ambient brightness change trend and user interaction prediction probability of the clock within a preset short window after the current moment.

[0048] The feedback update module is used to apply the three-level power-saving scheduling parameters to the clock. After a preset scheduling cycle, the newly collected actual energy consumption data of the clock is used as feedback input to update the energy consumption profile of the time period.

[0049] As can be seen from the above technical solution, the clock power-saving scheduling method and system based on multi-source energy consumption data analysis provided by this invention identifies the stable power consumption range and fluctuating power consumption range in the clock energy consumption data and extracts static background features and dynamic energy consumption pattern labels respectively. It then integrates and constructs a time-period energy consumption profile that reflects the steady-state and dynamic attributes of the time period. This allows the clock to match the static background features in the profile according to the current time. At the same time, it combines the real-time ambient brightness change trend and user interaction prediction probability to generate three-level power-saving scheduling parameters. These three-level power-saving scheduling parameters jointly and adaptively adjust the display refresh rate, backlight brightness, and sensor sampling interval, thereby maintaining smooth response during periods of high user interaction or high ambient brightness, and significantly reducing unnecessary energy consumption during periods of low demand. Meanwhile, the actual energy consumption feedback after the preset scheduling cycle updates the profile to form a closed-loop optimization, which significantly improves the accuracy and timeliness of the power-saving strategy. Under the same usage scenario, it extends the battery life of the clock while maintaining a good user experience. Attached Figure Description

[0050] Other objects and results of the invention will become more apparent and readily understood by referring to the following description taken in conjunction with the accompanying drawings, and with a more complete understanding of the invention. In the drawings:

[0051] Figure 1 This is a flowchart illustrating the power-saving scheduling method for clocks based on multi-source energy consumption data analysis, according to an embodiment of the present invention.

[0052] Figure 2 This is a functional block diagram of a clock power-saving scheduling system based on multi-source energy consumption data analysis, according to an embodiment of the present invention. Detailed Implementation

[0053] Existing clock energy-saving scheduling methods employ a single fixed strategy and lack time-domain analysis and pattern recognition of multi-source energy consumption data, resulting in an inability to distinguish between stable and fluctuating power consumption ranges, thus causing an imbalance between energy-saving effects and user experience.

[0054] To address the aforementioned problems, this invention provides a method and system for energy-saving scheduling of clocks based on multi-source energy consumption data analysis. The specific embodiments of this invention will be described in detail below with reference to the accompanying drawings.

[0055] To illustrate the power-saving scheduling method and system for clocks based on multi-source energy consumption data analysis provided by this invention, Figure 1 The clock energy-saving scheduling method based on multi-source energy consumption data analysis in this embodiment of the invention is illustrated by way of example. Figure 2 An exemplary illustration is provided for the clock energy-saving scheduling system based on multi-source energy consumption data analysis in an embodiment of the present invention.

[0056] The following description of exemplary embodiments is merely illustrative and is in no way intended to limit the invention or its application or use. Techniques and equipment known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques and equipment should be considered part of the specification.

[0057] Reference Figure 1 The diagram shown is a flowchart illustrating a clock power-saving scheduling method based on multi-source energy consumption data analysis according to an embodiment of the present invention. In this embodiment, the clock power-saving scheduling method based on multi-source energy consumption data analysis includes:

[0058] Step 1: By analyzing the changes in power consumption in the clock's energy consumption data, identify the stable power consumption range and fluctuating power consumption range within adjacent time points, as well as the corresponding static background features and dynamic energy consumption pattern labels, in order to construct a time period energy consumption profile of the clock.

[0059] In this embodiment of the invention, the process of identifying stable power consumption intervals and fluctuating power consumption intervals within adjacent time points, as well as the corresponding static background features and dynamic energy consumption pattern labels, by analyzing the power consumption changes in the energy consumption data of the clock, is as follows:

[0060] Based on the multi-source energy consumption data of the clocks aligned by timestamps, the rate of change of power consumption at adjacent time points is calculated, where...

[0061] The formula for calculating the rate of change of power consumption is as follows:

[0062] ;

[0063] In the formula, This represents the rate of change in the rate of electricity consumption at adjacent time points. This indicates the rate of power consumption at the previous time point in the adjacent time points. This indicates the rate of power consumption at the next time point in the adjacent time intervals. This represents the time interval between the adjacent time points;

[0064] When the rate of change of the power consumption rate is lower than the preset rate of change threshold, the corresponding time interval is marked as a stable power consumption interval, and the other time intervals are marked as fluctuating power consumption intervals.

[0065] The average ambient brightness value and average user interaction frequency within the stable power consumption range are used as the static background features of the clock.

[0066] The peak power consumption rate, the gradient of ambient brightness value change, and the abrupt change point of user interaction frequency within the fluctuating power consumption range are correlated and mapped to generate a dynamic energy consumption mode label for the clock.

[0067] The process of constructing the time-period energy consumption profile of the clock is as follows:

[0068] Using the average ambient brightness value in the static background features as the base layer, and the average user interaction frequency as the weight coefficient of the base layer, the steady-state energy consumption base plate of the clock is laid out.

[0069] Each feature value in the dynamic energy consumption mode label is anchored to the corresponding position on the time axis of the steady-state energy consumption base plate to correct the steady-state energy consumption base plate.

[0070] The corrected steady-state energy consumption base plate, together with the embedded dynamic energy consumption mode label, is packaged and encapsulated into the time period energy consumption profile of the clock.

[0071] The process of correcting the steady-state energy consumption base plate is as follows:

[0072] Using the time point corresponding to the peak power consumption rate as the anchor point, the area on the steady-state energy consumption base plate located within a preset range before and after the anchor point is expanded outward along the vertical axis in proportion to the amplitude of the peak power consumption rate.

[0073] Using the direction of the ambient brightness value change gradient as the distortion vector, a shear transformation proportional to the magnitude of the ambient brightness value change gradient is applied to the edge contour of the steady-state energy consumption base plate along the time axis direction of the steady-state energy consumption base plate.

[0074] Using the position of the user interaction frequency mutation point on the time axis as the cutting line, the steady-state energy consumption base plate is divided into multiple independent sub-blocks.

[0075] Multi-source energy consumption data for the clocks is collected, including power consumption, ambient light levels, and user interaction frequency. Each data point is accompanied by a precise timestamp in a unified format of year, month, day, hour, minute, and second. This data is then sorted chronologically by timestamp, from earliest to latest. For the same time point, the corresponding power consumption, ambient light, and user interaction frequency are extracted from different data sources, ensuring that each timestamp contains all three values. If a data point is missing, it is filled using linear interpolation, which calculates the average of the closest known data points before and after that time point. The final result is a complete data table where each row corresponds to a time point and each column corresponds to a type of data, achieving timestamp alignment across multiple data sources.

[0076] The power consumption rate at adjacent time points is calculated by acquiring multi-source energy consumption data from the clock's historical operation. This data includes power consumption values, ambient brightness values, and user interaction frequencies arranged in a strict time series. For the specific indicator of power consumption rate, a sequence of power consumption values ​​sorted by timestamp from smallest to largest is extracted from the dataset. By calculating the difference between the power consumption values ​​at two adjacent time points and dividing it by the fixed time interval between those two points, the instantaneous power consumption rate corresponding to each adjacent time point is obtained, forming a time series of power consumption rates.

[0077] Determine the power consumption rate at the next adjacent time point by directly extracting the value at the end of the power consumption rate time series generated during the above calculation process. This value reflects how quickly the clock consumes power at the next time point. Determine the power consumption rate at the previous adjacent time point by extracting the value at the beginning of the same time series. This value reflects how quickly the clock consumes power at the previous time point. These two rate values ​​constitute the core input for calculating the rate of change.

[0078] The time interval between adjacent time points is determined by directly reading the time difference between the next time point and the previous time point, since the collected multi-source energy consumption data is usually sampled at a fixed frequency.

[0079] The power consumption rate at the next time point is subtracted from the power consumption rate at the previous time point, and the absolute value of the subtraction is taken to obtain the absolute difference in power consumption rates. Simultaneously, the power consumption rates at these two time points are added together to obtain the sum of their rates. The absolute difference is then divided by the product of the sum of rates and the time interval to complete the quantitative calculation of the rate of change of power consumption rate.

[0080] A fixed rate of change threshold is set; this threshold is a predefined value used to determine the stability of power consumption. The rate of change is iterated through all time points, and each rate of change is compared to the threshold. If the rate of change is below the threshold, the time point belongs to a stable power consumption state; if the rate of change is above or equal to the threshold, the time point belongs to a fluctuating power consumption state. All consecutive time points belonging to a stable power consumption state are merged into a single stable power consumption interval, starting from the first stable time point and ending at the last. Similarly, all consecutive time points belonging to a fluctuating power consumption state are merged into a single fluctuating power consumption interval. Finally, a series of time intervals are output, each labeled as either a stable power consumption interval or a fluctuating power consumption interval.

[0081] Within a stable power consumption range, ambient brightness values ​​are extracted for all time points within the range. These values ​​are summed and then divided by the total number of brightness values ​​to obtain the arithmetic mean, which is the average ambient brightness value. Simultaneously, user interaction frequencies are extracted for all time points within the range. These frequency values ​​are summed and then divided by the total number of frequency values ​​to obtain the arithmetic mean, which is the average user interaction frequency. These two averages are combined into a data pair called the clock's static background feature. This feature represents the ambient background and user interaction level of the clock under stable power consumption conditions.

[0082] Within the fluctuating power consumption range, the peak power consumption rate is first identified. The rate values ​​at all time points within the range are iterated through, and points where the rate value is greater than the rate values ​​at the preceding and following time points are identified as local maxima. The magnitude of each peak and its corresponding time point are recorded. Next, the gradient of ambient brightness change is calculated. Starting from the first time point within the range, the change in ambient brightness value between adjacent time points is calculated. The change is the difference between the brightness value at the next time point and the brightness value at the previous time point, then divided by the time interval to obtain the gradient value for each time interval. The gradient direction is indicated by the sign of the change, with positive indicating an increase and negative indicating a decrease. Abrupt points in user interaction frequency are detected by comparing the frequency values ​​at adjacent time points. If the change in frequency value exceeds a fixed threshold, the time point is considered an abrupt change, and its time is recorded. These features are then correlated, for example, by creating a list containing the time and magnitude of each peak, the time and direction of each gradient value, and the time of each abrupt change, forming a dynamic energy consumption pattern label.

[0083] The average ambient brightness value from the static background features is used as the baseline value for the base layer. This base layer is a two-dimensional plane, with the horizontal axis representing the time axis and the vertical axis representing the energy consumption baseline value. The average ambient brightness value is directly assigned to each point in the base layer, giving the entire layer a constant value on the vertical axis. Then, the average user interaction frequency is used as a weighting coefficient. The average user interaction frequency is multiplied by the value of each point in the base layer to obtain a new vertical axis value. This new value is the weighted base layer. This operation lays out the steady-state energy consumption baseline of the clock. The steady-state energy consumption baseline is a continuous plane on the time axis, and its vertical axis value is defined by the weighted average ambient brightness value, reflecting the energy consumption baseline under the static background.

[0084] Feature values ​​are extracted from the dynamic energy consumption mode labels, including the time points corresponding to peak power consumption rates, gradient changes in ambient brightness, and abrupt changes in user interaction frequency. These feature values ​​are then mapped onto the time axis of the steady-state energy consumption platform based on their timestamps. The steady-state energy consumption platform's time axis covers all time points, and the time point of each feature value is aligned with its coordinates on the platform's time axis. The feature values ​​are then appended to their corresponding positions on the platform as data points, achieving anchoring.

[0085] Using the time point corresponding to the peak power consumption rate as the anchor point, a range is defined on the steady-state energy consumption baseline. This range is the area within a fixed time length before and after the anchor point, for example, five seconds before and after the anchor point. Along the vertical axis, the baseline value at each time point within this range is expanded. The expansion ratio is determined by the peak value. Specifically, each value is multiplied by a coefficient, which is the peak value divided by the baseline reference value, which is the weighted average ambient brightness value. A larger peak value results in a larger coefficient and a larger expansion, causing the baseline to bulge higher in that area, simulating the impact of the peak value on energy consumption.

[0086] Using the direction of the gradient of ambient brightness value change as the distortion vector, the gradient direction is indicated by the sign of the change: a positive direction indicates forward along the time axis, and a negative direction indicates backward along the time axis. A shear transformation is applied to the edge contour of the steady-state energy consumption baseline along its time axis. Specifically, starting from one edge of the baseline, the baseline value at each time point is moved along the time axis. The movement distance is proportional to the gradient magnitude; a larger gradient results in a larger movement distance, and a smaller gradient results in a smaller movement distance. For example, if the gradient is positive, the point is moved forward; if the gradient is negative, it is moved backward. After the movement, the shape of the baseline is distorted along the time axis, but the vertical axis value remains unchanged; only the position on the time axis changes, reflecting the impact of ambient brightness changes on the temporal distribution of energy consumption.

[0087] Using the location of abrupt changes in user interaction frequency on the time axis as cutting lines, each abrupt change corresponds to a specific time point. On the steady-state energy consumption baseline, the time coordinates of each abrupt change are located, and then lines perpendicular to the time axis are drawn at these coordinates to segment the baseline. Specifically, the data sequence on the baseline is broken at the abrupt change time points, forming multiple independent sub-blocks. Each sub-block contains all data points from the previous cutting line to the current cutting line. These sub-blocks are unconnected, forming independent data segments for easier subsequent analysis.

[0088] The corrected steady-state energy consumption base plate is treated as a whole data structure, with corrections including previous expansions and shearing transformations. The embedded dynamic energy consumption pattern tags are associated with the base plate, and the tag data is appended as metadata to the base plate data. Then, all data is packaged and encapsulated, creating a data object containing base plate data, tag data, and timeline information; this data object is called the clock's time-period energy consumption profile.

[0089] The time-period energy consumption profile provides a complete description of the watch's energy consumption characteristics over a specific time period, including steady-state benchmarks and dynamic modes, which can be used for further analysis or visualization.

[0090] Step 2: Based on the static background features of the clock at the current moment in the energy consumption profile of the time period, combined with the real-time ambient brightness change trend and user interaction prediction probability within a preset short window after the current moment, generate the three-level power-saving scheduling parameters of the clock.

[0091] In this embodiment of the invention, the real-time ambient brightness change trend is as follows:

[0092] Within a preset short window after the current time of the clock, the ambient light sensor of the clock is woken up sequentially at a fixed sampling interval to read the original ambient brightness reading at each sampling time, so as to form a sequence of discrete brightness points scattered along the time axis of the clock.

[0093] The values ​​in the discrete brightness point sequence are compared sequentially. If the subsequent value is greater than the previous value, it is marked as an increase; if it is less than the previous value, it is marked as a decrease; if it is equal to the previous value, it is marked as a level.

[0094] The number of all markers in the discrete brightness point sequence is counted, and the direction of change corresponding to the marker type with the highest proportion is taken as the real-time ambient brightness change trend of the clock.

[0095] The process for predicting the probability of user interaction is as follows:

[0096] Extract all historical interaction moments that belong to the same time period type as the current moment from the clock's historical operation log;

[0097] The time interval between every two historical interaction moments is mapped onto the time axis to form an interval distribution histogram with the time interval as the horizontal axis and the frequency of occurrence of the time interval as the vertical axis.

[0098] Using the time elapsed since the most recent actual user interaction action from the current time of the clock as the query index, the frequency of occurrence located in the interval distribution histogram is divided by the total number of time intervals to obtain the predicted probability of user interaction of the clock.

[0099] The process of generating the three-level power-saving scheduling parameters for the clock is as follows:

[0100] The average ambient brightness value in the static background features is superimposed on the real-time ambient brightness change trend to obtain the backlight brightness attenuation coefficient of the clock.

[0101] The product of the average user interaction frequency in the static background features and the predicted probability of user interaction is used as the display refresh rate reduction ratio of the clock.

[0102] Using the larger of the backlight brightness attenuation coefficient and the display refresh rate reduction ratio as a benchmark, the benchmark is mapped in stages according to the current remaining power of the clock to obtain the sensor sampling interval extension factor of the clock.

[0103] The backlight brightness attenuation coefficient, the display refresh rate reduction ratio, and the sensor sampling interval extension factor are integrated into the three-level power-saving scheduling parameters of the clock.

[0104] A fixed time window length, such as thirty seconds, is set after the current moment as a preset short-time window. An internal timer is started, beginning from the current moment. When the timer reaches the first fixed sampling interval, for example, once every two seconds, an interrupt signal is triggered. This interrupt signal activates the power management circuit of the ambient light sensor, causing the sensor to switch from sleep mode to working mode. The photosensitive element of the ambient light sensor receives ambient light and generates an analog current signal, which is converted into a digital value, i.e., the raw ambient brightness reading, by an analog-to-digital converter. This digital value is read and stored in a memory buffer, while the corresponding sampling time is recorded. After reading is complete, the power of the ambient light sensor is immediately turned off, returning it to sleep mode to save energy. The timer continues to run, and when it reaches the next fixed sampling interval, the above wake-up, reading, and sleep process is repeated. This cycle is repeated until the accumulated time of the timer reaches the end of the preset short-time window. Finally, the memory buffer stores a series of data pairs arranged in chronological order, each data pair containing a precise sampling time and a corresponding raw ambient brightness reading. These data pairs together constitute a sequence of discrete brightness points.

[0105] Retrieve the discrete brightness point sequence from memory and sort it according to timestamp order to ensure data point order. Initialize three counters to record the number of rising, falling, and leveling marks, respectively. Starting from the first data point in the sequence, process each pair of adjacent data points sequentially. Retrieve the current data point and the next data point, and compare their raw ambient brightness readings. If the value of the next data point is strictly greater than the value of the current data point, increment the rising mark counter and record a rising mark between the two data points. If the value of the next data point is strictly less than the value of the current data point, increment the falling mark counter and record a falling mark. If the two values ​​are equal, increment the leveling mark counter and record a leveling mark. Repeat this process until all adjacent data point pairs in the sequence have been processed. After processing, read the final values ​​of the rising mark counter, falling mark counter, and leveling mark counter, respectively. Compare these three values ​​and find the largest one. If the rising mark counter value is the largest, the real-time ambient brightness change trend is determined to be upward. If the falling mark counter value is the largest, the real-time ambient brightness change trend is determined to be downward. If the value of the leveling counter is at its maximum, the real-time ambient brightness change trend is determined to be in the leveling direction. This trend direction will be used for subsequent power-saving scheduling.

[0106] Access the clock's non-volatile memory, which stores historical operation log files. Parse the log files to extract all recorded user interaction events, each including the interaction type and the precise timestamp of the interaction. Based on the current hour and minute count, and combined with predefined time period classification rules, determine the time period type to which the current time belongs, such as "morning work period" or "night rest period." From the extracted historical interaction events, filter out all events whose interaction timestamps belong to the exact same time period type as the current time period, forming a filtered list of historical interaction moments. Obtain the current system clock value. Find the interaction moment closest to but earlier than the current time from the list of historical interaction moments, calculate the time difference between this interaction moment and the current time, and obtain the inactivity duration in seconds.

[0107] From the filtered list of historical interaction moments, timestamps are sorted in ascending order. The time difference between any two adjacent timestamps is calculated sequentially, resulting in a series of time interval values ​​in seconds. A set of continuous and equally wide time interval intervals, such as 0 to 5 seconds, 5 to 10 seconds, etc., are defined as the horizontal axis groupings of the histogram. All calculated time interval values ​​are iterated over, and each value is categorized into its corresponding interval, with the count value of that interval incremented by one for each value. The set of all interval count values ​​forms an interval distribution histogram, where the center of each interval represents the time interval, and the count value represents the frequency of occurrence of that time interval. The total number of all time interval values ​​is calculated, which is the number of adjacent time pairs in the historical interaction moment list. Using the inactivity duration value as the lookup key, the interval to which the inactivity duration belongs is found in the interval distribution histogram. The count value of that interval is taken as the located frequency of occurrence. This frequency value is divided by the total number of time intervals to obtain the predicted probability of user interaction, which represents the likelihood of user interaction occurring under the current inactivity duration.

[0108] The system extracts the average ambient brightness value from static background features. This is a specific numerical value representing the ambient light level under historical stable conditions. It then acquires the real-time ambient brightness trend, which is a directional indicator (increasing, decreasing, or remaining constant). Based on the trend direction, it selects the corresponding overlay rule from a preset mapping table. If the trend is increasing, the average ambient brightness value is multiplied by a fixed coefficient greater than one to obtain the backlight brightness attenuation coefficient. This coefficient greater than one indicates that the backlight brightness can be increased. If the trend is decreasing, the average ambient brightness value is multiplied by a fixed coefficient less than one to obtain the backlight brightness attenuation coefficient. This coefficient less than one indicates that the backlight brightness needs to be reduced. If the trend is stable, the backlight brightness attenuation coefficient is directly equal to the average ambient brightness value, indicating that the current brightness level is maintained. The backlight brightness attenuation coefficient is a control parameter used to adjust the backlight brightness of the clock.

[0109] The average user interaction frequency is extracted from static background features; this is a numerical value representing the number of interactions per unit time. The predicted probability of user interaction is obtained, a probability value between zero and one. The average user interaction frequency is multiplied by the predicted probability; the result of this multiplication is a new value, the display refresh rate reduction ratio. This ratio indicates the degree to which the clock display module's refresh rate should be reduced; for example, a ratio of 0.5 means the refresh rate should be reduced to half its original value. The display refresh rate reduction ratio directly determines the strength of subsequent display refresh energy-saving strategies.

[0110] Compare the backlight brightness attenuation coefficient and the display refresh rate reduction ratio, and take the larger value as the baseline value. Read the current remaining battery percentage value provided by the watch's battery management chip. Based on the preset range of the remaining battery percentage, look up the corresponding adjustment rule from the graded mapping table. If the remaining battery is above 70%, the baseline value remains unchanged. If the remaining battery is between 30% and 70%, multiply the baseline value by a fixed coefficient between 1 and 1.5. If the remaining battery is below 30%, multiply the baseline value by a fixed coefficient between 1.5 and 3. The adjusted baseline value is the sensor sampling interval extension factor. This factor is used to extend the sampling time interval of the ambient light sensor. For example, a factor of 2 indicates that the sampling interval is doubled, thereby reducing sensor power consumption.

[0111] Create a structured data object containing three fields. The first field stores the backlight brightness attenuation coefficient, which is a floating-point number. The second field stores the display refresh rate reduction ratio, which is also a floating-point number. The third field stores the sensor sampling interval extension factor, which is a floating-point number. Fill these three values ​​into their respective fields. Then, serialize and encode this data object into a byte stream format to ensure data integrity and transmissibility.

[0112] The serialized byte stream is the three-level power-saving scheduling parameter. This parameter set can be directly passed to the underlying hardware controller of the clock to adjust the backlight brightness, display refresh rate and sensor sampling interval, so as to achieve hierarchical energy-saving control.

[0113] Step 3: Apply the three-level power-saving scheduling parameters to the clock. After a preset scheduling cycle, use the newly collected actual energy consumption data of the clock as feedback input to update the energy consumption profile for the specified time period.

[0114] In this embodiment of the invention, the process of applying the three-level power-saving scheduling parameters to the clock is as follows:

[0115] Based on the display refresh rate reduction ratio, remove the corresponding ratio of frame refresh instructions from the display driver instruction stream of the clock.

[0116] Using the backlight brightness attenuation coefficient as the duty cycle adjustment factor for the clock pulse width modulation signal, the output brightness of the clock is reduced.

[0117] According to the sensor sampling interval extension factor, the trigger time for the next data acquisition of the built-in ambient light sensor and accelerometer in the watch is postponed.

[0118] After a preset scheduling period, the newly collected actual energy consumption data of the clocks is used as feedback input to update the energy consumption profile for the specified time period. The process is as follows:

[0119] After a preset scheduling period, the actual energy consumption data within the scheduling period is collected again to construct the actual energy consumption feature vector of the clock.

[0120] The actual energy consumption feature vector is weighted and averaged with the static background features of the corresponding time period in the time period energy consumption profile, and the weighted average result is used to replace the original static background features.

[0121] The fluctuation portion in the actual energy consumption feature vector is compared with the original dynamic energy consumption pattern label. If the deviation exceeds a preset deviation threshold, the original label is replaced with the dynamic energy consumption pattern label extracted from the actual energy consumption feature vector.

[0122] The display refresh rate reduction ratio is extracted from the three-level power-saving scheduling parameters. This value is a decimal between zero and one. In the clock's display driver, frame refresh instructions are generated by a hardware timer at a fixed frequency, with one instruction generated at each trigger. An accumulator variable is maintained, initially set to zero. Each time the hardware timer generates a frame refresh instruction, the display refresh rate reduction ratio is added to the accumulator variable. If the accumulator variable is less than one, the current frame refresh instruction is discarded and not sent to the display controller. If the accumulator variable is greater than or equal to one, the current frame refresh instruction is sent to the display controller, and one is subtracted from the accumulator variable. In this way, frame refresh instructions in the display driver instruction stream are partially eliminated according to the display refresh rate reduction ratio, thus reducing the display refresh rate.

[0123] The backlight brightness attenuation coefficient value is extracted from the level 3 power-saving scheduling parameters; this value is a positive decimal. The clock's backlight control module is accessed. This module controls the backlight brightness via a pulse-width modulation (PWM) signal. The PWM signal's duty cycle register stores the proportion of the current high-level time within the entire cycle. The current duty cycle register value of the PWM signal is read. This value is multiplied by the backlight brightness attenuation coefficient to obtain a new duty cycle value. This new duty cycle value is written into the PWM signal's duty cycle register, thereby changing the proportion of the PWM signal's high-level time and thus reducing the clock's output brightness.

[0124] The sensor sampling interval extension factor value, a real number greater than one, is extracted from the level-three power-saving scheduling parameters. The timing trigger modules of the built-in ambient light sensor and accelerometer are accessed; these modules control the sensor's data acquisition cycle. The current sampling interval time setting values ​​for the ambient light sensor and accelerometer are read; these values ​​represent the fixed time interval between two consecutive data acquisitions. The current sampling interval time setting value is multiplied by the sensor sampling interval extension factor to obtain the new sampling interval time. The timer in the timing trigger module is updated, setting the next data acquisition trigger time to the current system time plus the new sampling interval time, thereby delaying the next data acquisition trigger time for the ambient light sensor and accelerometer and reducing the sensor's data acquisition frequency.

[0125] After a preset scheduling period, such as 30 minutes, the clock's energy consumption monitoring module begins collecting actual energy consumption data within that period. It reads power consumption data, ambient brightness data, and user interaction data from the module, each data point bearing a precise timestamp. This data undergoes timestamp alignment, grouping different data points from the same time point into the same row. The power consumption rate at each time point is calculated, which is the difference between the power consumption value at the next time point and the power consumption value at the previous time point, divided by the time interval. The rate of change of power consumption rate between adjacent time points is also calculated, which is the difference between the rate at the current time point and the rate at the previous time point, divided by the time interval. Based on a preset rate of change threshold, time intervals with a rate of change below the threshold are marked as stable power consumption intervals, while time intervals with a rate of change above or equal to the threshold are marked as fluctuating power consumption intervals. The average ambient brightness value and average user interaction frequency are calculated from the stable power consumption intervals, while the peak power consumption rate, the gradient of ambient brightness value changes, and the abrupt change points in user interaction frequency are extracted from the fluctuating power consumption intervals. These feature values ​​are then sequentially combined into a numerical sequence to form the actual energy consumption feature vector.

[0126] The actual average ambient brightness value and actual average user interaction frequency are extracted from the actual energy consumption feature vector. The original average ambient brightness value and original average user interaction frequency for the corresponding time period are extracted from the time-period energy consumption profile. Using preset original feature weights and actual feature weights, a weighted average calculation is performed on the average ambient brightness value. Specifically, the original average ambient brightness value is multiplied by the original feature weight, plus the actual average ambient brightness value multiplied by the actual feature weight, and then divided by the sum of the original feature weights and the actual feature weights to obtain the weighted average ambient brightness value. The average user interaction frequency is calculated using the same weighted average calculation to obtain the weighted average user interaction frequency. The original static background features in the time-period energy consumption profile are replaced with the weighted average ambient brightness value and average user interaction frequency, completing the update of the static background features.

[0127] Extract the peak value of actual power consumption rate, the gradient of actual ambient brightness value change, and the abrupt change point of actual user interaction frequency from the actual energy consumption feature vector. Extract the original peak value of power consumption rate, the original gradient of ambient brightness value change, and the original abrupt change point of user interaction frequency from the time period energy consumption profile. For the peak value of power consumption rate, calculate the absolute deviation between the actual value and the original value. For the gradient of ambient brightness value change, calculate the absolute deviation between the actual value and the original value. For the abrupt change point of user interaction frequency, calculate the absolute deviation between the actual value and the original value. If the absolute deviation of the peak value of power consumption rate exceeds a preset peak deviation threshold, or the absolute deviation of the gradient of ambient brightness value change exceeds a preset gradient deviation threshold, or the absolute deviation of the abrupt change point of user interaction frequency exceeds a preset abrupt change point deviation threshold, then replace the original dynamic energy consumption pattern label in the time period energy consumption profile with the dynamic energy consumption pattern label extracted from the actual energy consumption feature vector. Otherwise, retain the original dynamic energy consumption pattern label in the time period energy consumption profile.

[0128] As can be seen from the above embodiments, the clock power-saving scheduling method based on multi-source energy consumption data analysis provided by the present invention identifies the stable power consumption range and fluctuating power consumption range in the clock energy consumption data and extracts static background features and dynamic energy consumption pattern labels respectively. It then fuses and constructs a time-period energy consumption profile that reflects the steady-state and dynamic attributes of the time period. This allows the clock to match the static background features in the profile according to the current time. At the same time, it combines the real-time ambient brightness change trend and user interaction prediction probability to generate three-level power-saving scheduling parameters. These three-level power-saving scheduling parameters jointly and adaptively adjust the display refresh rate, backlight brightness, and sensor sampling interval, thereby maintaining smooth response during periods of high user interaction or high ambient brightness, and significantly reducing unnecessary energy consumption during periods of low demand. In addition, the actual energy consumption feedback after the preset scheduling cycle updates the profile to form a closed-loop optimization, which significantly improves the accuracy and timeliness of the power-saving strategy. Under the same usage scenario, it extends the battery life of the clock while maintaining a good user experience.

[0129] like Figure 2 The diagram shown is a functional block diagram of a clock power-saving scheduling system 100 based on multi-source energy consumption data analysis provided in an embodiment of the present invention, including a profile construction module 101, a parameter generation module 102, and a feedback update module 103.

[0130] In this embodiment, the functions of each module are as follows:

[0131] The profile building module 101 is used to identify stable power consumption intervals and fluctuating power consumption intervals within adjacent time points, as well as corresponding static background features and dynamic energy consumption pattern labels, by analyzing the power consumption changes in the energy consumption data of the clock, so as to build a time period energy consumption profile of the clock.

[0132] The parameter generation module 102 is used to generate three-level power-saving scheduling parameters for the clock based on the static background features of the clock at the current moment in the energy consumption profile of the time period, combined with the real-time ambient brightness change trend and user interaction prediction probability of the clock within a preset short window after the current moment.

[0133] The feedback update module 103 is used to apply the three-level power-saving scheduling parameters to the clock. After a preset scheduling cycle, the newly collected actual energy consumption data of the clock is used as feedback input to update the energy consumption profile of the time period.

[0134] As can be seen from the above embodiments, the clock power-saving scheduling system based on multi-source energy consumption data analysis provided by the present invention identifies the stable power consumption range and fluctuating power consumption range in the clock energy consumption data and extracts static background features and dynamic energy consumption pattern labels respectively. It then fuses and constructs a time-period energy consumption profile that reflects the steady-state and dynamic attributes of the time period. This allows the clock to match the static background features in the profile according to the current time. At the same time, it combines the real-time ambient brightness change trend and user interaction prediction probability to generate three-level power-saving scheduling parameters. These three-level power-saving scheduling parameters jointly and adaptively adjust the display refresh rate, backlight brightness, and sensor sampling interval, thereby maintaining smooth response during periods of high user interaction or high ambient brightness, and significantly reducing unnecessary energy consumption during periods of low demand. In addition, the actual energy consumption feedback after the preset scheduling cycle updates the profile to form a closed-loop optimization, which significantly improves the accuracy and timeliness of the power-saving strategy. Under the same usage scenario, it extends the battery life of the clock while maintaining a good user experience.

[0135] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0136] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A clock energy-saving scheduling method based on multi-source energy consumption data analysis, characterized in that, The method includes: Step 1: By analyzing the changes in power consumption in the clock's energy consumption data, identify the stable power consumption range and fluctuating power consumption range within adjacent time points, as well as the corresponding static background features and dynamic energy consumption pattern labels, in order to construct a time period energy consumption profile of the clock. Step 2: Based on the static background features of the clock at the current moment in the energy consumption profile of the time period, combined with the real-time ambient brightness change trend and user interaction prediction probability within a preset short window after the current moment, generate the three-level power-saving scheduling parameters of the clock. Step 3: Apply the three-level power-saving scheduling parameters to the clock. After a preset scheduling cycle, use the newly collected actual energy consumption data of the clock as feedback input to update the energy consumption profile for the specified time period.

2. The clock energy-saving scheduling method based on multi-source energy consumption data analysis as described in claim 1, characterized in that, The process of identifying stable power consumption ranges and fluctuating power consumption ranges within adjacent time points, along with corresponding static background features and dynamic energy consumption pattern labels, by analyzing power consumption changes in clock energy consumption data, is as follows: Based on the multi-source energy consumption data of the clocks aligned by timestamps, the rate of change of power consumption rate at adjacent time points is calculated. The formula for calculating the rate of change of power consumption rate is as follows: ； In the formula, This represents the rate of change in the rate of electricity consumption at adjacent time points. This indicates the rate of power consumption at the previous time point in the adjacent time points. This indicates the rate of power consumption at the next time point in the adjacent time intervals. This represents the time interval between the adjacent time points; When the rate of change of the power consumption rate is lower than the preset rate of change threshold, the corresponding time interval is marked as a stable power consumption interval, and the other time intervals are marked as fluctuating power consumption intervals. The average ambient brightness value and average user interaction frequency within the stable power consumption range are used as the static background features of the clock. The peak power consumption rate, the gradient of ambient brightness value change, and the abrupt change point of user interaction frequency within the fluctuating power consumption range are correlated and mapped to generate a dynamic energy consumption mode label for the clock.

3. The clock energy-saving scheduling method based on multi-source energy consumption data analysis as described in claim 2, characterized in that, The process of constructing the time-period energy consumption profile of the clock is as follows: Using the average ambient brightness value in the static background features as the base layer, and the average user interaction frequency as the weight coefficient of the base layer, the steady-state energy consumption base plate of the clock is laid out. Each feature value in the dynamic energy consumption mode label is anchored to the corresponding position on the time axis of the steady-state energy consumption base plate to correct the steady-state energy consumption base plate. The corrected steady-state energy consumption base plate, together with the embedded dynamic energy consumption mode label, is packaged and encapsulated into the time period energy consumption profile of the clock.

4. The clock energy-saving scheduling method based on multi-source energy consumption data analysis as described in claim 3, characterized in that, The process of correcting the steady-state energy consumption base plate is as follows: Using the time point corresponding to the peak power consumption rate as the anchor point, the area on the steady-state energy consumption base plate located within a preset range before and after the anchor point is expanded outward along the vertical axis in proportion to the amplitude of the peak power consumption rate. Using the direction of the ambient brightness value change gradient as the distortion vector, a shear transformation proportional to the magnitude of the ambient brightness value change gradient is applied to the edge contour of the steady-state energy consumption base plate along the time axis direction of the steady-state energy consumption base plate. Using the position of the user interaction frequency mutation point on the time axis as the cutting line, the steady-state energy consumption base plate is divided into multiple independent sub-blocks.

5. The clock energy-saving scheduling method based on multi-source energy consumption data analysis as described in claim 1, characterized in that, The real-time ambient brightness change trend is described as follows: Within a preset short window after the current time of the clock, the ambient light sensor of the clock is woken up sequentially at a fixed sampling interval to read the original ambient brightness reading at each sampling time, so as to form a sequence of discrete brightness points scattered along the time axis of the clock. The values ​​in the discrete brightness point sequence are compared sequentially. If the subsequent value is greater than the previous value, it is marked as an increase; if it is less than the previous value, it is marked as a decrease; if it is equal to the previous value, it is marked as a level. The number of all markers in the discrete brightness point sequence is counted, and the direction of change corresponding to the marker type with the highest proportion is taken as the real-time ambient brightness change trend of the clock.

6. The clock energy-saving scheduling method based on multi-source energy consumption data analysis as described in claim 5, characterized in that, The process for predicting the probability of user interaction is as follows: Extract all historical interaction moments that belong to the same time period type as the current moment from the clock's historical operation log; The time interval between every two historical interaction moments is mapped onto the time axis to form an interval distribution histogram with the time interval as the horizontal axis and the frequency of occurrence of the time interval as the vertical axis. Using the time elapsed since the most recent actual user interaction action from the current time of the clock as the query index, the frequency of occurrence located in the interval distribution histogram is divided by the total number of time intervals to obtain the predicted probability of user interaction of the clock.

7. The clock energy-saving scheduling method based on multi-source energy consumption data analysis as described in claim 6, characterized in that, The process of generating the three-level power-saving scheduling parameters for the clock is as follows: The average ambient brightness value in the static background features is superimposed on the real-time ambient brightness change trend to obtain the backlight brightness attenuation coefficient of the clock. The product of the average user interaction frequency in the static background features and the predicted probability of user interaction is used as the display refresh rate reduction ratio of the clock. Using the larger of the backlight brightness attenuation coefficient and the display refresh rate reduction ratio as a benchmark, the benchmark is mapped in stages according to the current remaining power of the clock to obtain the sensor sampling interval extension factor of the clock. The backlight brightness attenuation coefficient, the display refresh rate reduction ratio, and the sensor sampling interval extension factor are integrated into the three-level power-saving scheduling parameters of the clock.

8. The clock energy-saving scheduling method based on multi-source energy consumption data analysis as described in claim 7, characterized in that, The process of applying the three-level power-saving scheduling parameters to the clock is as follows: Based on the display refresh rate reduction ratio, remove the corresponding ratio of frame refresh instructions from the display driver instruction stream of the clock. Using the backlight brightness attenuation coefficient as the duty cycle adjustment factor for the clock pulse width modulation signal, the output brightness of the clock is reduced. According to the sensor sampling interval extension factor, the trigger time for the next data acquisition of the built-in ambient light sensor and accelerometer in the watch is postponed.

9. The clock energy-saving scheduling method based on multi-source energy consumption data analysis as described in claim 8, characterized in that, After a preset scheduling period, the newly collected actual energy consumption data of the clocks is used as feedback input to update the energy consumption profile for the specified time period. The process is as follows: After a preset scheduling period, the actual energy consumption data within the scheduling period is collected again to construct the actual energy consumption feature vector of the clock. The actual energy consumption feature vector is weighted and averaged with the static background features of the corresponding time period in the time period energy consumption profile, and the weighted average result is used to replace the original static background features. The fluctuation portion in the actual energy consumption feature vector is compared with the original dynamic energy consumption pattern label. If the deviation exceeds a preset deviation threshold, the original label is replaced with the dynamic energy consumption pattern label extracted from the actual energy consumption feature vector.

10. A clock energy-saving scheduling system based on multi-source energy consumption data analysis, characterized in that, The system is used to implement the clock energy-saving scheduling method based on multi-source energy consumption data analysis as described in any one of claims 1-9, the system comprising: The profile building module is used to identify stable power consumption ranges and fluctuating power consumption ranges within adjacent time points, as well as corresponding static background features and dynamic energy consumption pattern labels, by analyzing the power consumption changes in the energy consumption data of the clock, so as to construct the time period energy consumption profile of the clock. The parameter generation module is used to generate three-level power-saving scheduling parameters for the clock based on the static background features of the clock at the current moment in the energy consumption profile of the time period, combined with the real-time ambient brightness change trend and user interaction prediction probability of the clock within a preset short window after the current moment. The feedback update module is used to apply the three-level power-saving scheduling parameters to the clock. After a preset scheduling cycle, the newly collected actual energy consumption data of the clock is used as feedback input to update the energy consumption profile of the time period.

Images