WO2020252822A1 - Time series data processing method and apparatus - Google Patents

Abstract

A time series data processing method and apparatus, the method comprising: acquiring time series data, each item of data in the time series being a value at a time point (101); sequentially calculating the fluctuating energy at each time point E(n)=V(n)‑V(n‑1), wherein E(n) is the fluctuating energy at a current time point, V(n) is a value at the current time point, and V(n‑1) is a value at a previous time point (102); cyclically traversing each time point in the time series, and determining the state at the time points according to the fluctuating energy at each time point, the states comprising: a temporary state and a stable state (103); and using a stable state at the time points as the state mode of the time points, the state mode comprising: a steady mode, an ascending mode, and a descending mode (104). By using the present method, the segmentation of time series data and the identification of state modes may be simply and accurately achieved.

Description

Time series data processing method and device Technical field

The invention relates to the field of data processing, in particular to a method and device for processing time series data.

Background technique

Time series data refers to data collected in chronological order. This type of data reflects the change state or degree of a certain thing, phenomenon, etc. over time. It can be intuitively represented in the plane as a time series waveform, such as electrocardiogram (ECG) ), electroencephalogram (EEG), current and voltage signals in manufacturing, K lines for stock trading, time-domain waveforms of voice signals, etc. In the analysis and application of this kind of data, it is usually necessary to segment the waveforms and classify the segmented waveforms to provide corresponding information for various applications.

In the prior art, there are many ways to segment time series data, but most of the prior art has various limitations or defects, such as high computational complexity, inaccurate segmentation in complex scenes, and inapplicability to Real-time calculation, large influence by noisy data, strong assumptions on the distribution or source of data, supervised learning requires training data and labels, or only suitable for a certain signal with a specific graphic mode.

Summary of the invention

The embodiments of the present invention provide a time series data processing method and device to solve one or more of the above-mentioned problems in the prior art.

To this end, the present invention provides the following technical solutions:

A time series data processing method, the method includes:

Acquiring time series data, where each data in the time series is a value at a time point;

Calculate the fluctuating energy E(n)=V(n)-V(n-1) at each time point in turn, where E(n) is the fluctuating energy at the current time point, and V(n) is the value at the current time point, V(n-1) is the value at the previous time point;

Loop through each time point in the time sequence, and determine the state of the time point according to the fluctuating energy at each time point, the state includes: a temporary state and a stable state;

The stable state at the time point is taken as the state mode at the time point; the state mode includes: a steady mode, a rising mode, and a falling mode.

Optionally, the loop traversing each time point in the time sequence, and determining the state of the time point according to the fluctuating energy at each time point includes:

Loop through each time point in the time sequence, and determine the temporary state of the time point according to the fluctuating energy at the time point;

The stable state at the time point is determined according to the absolute value of the fluctuating energy at the time point and the temporary state.

Optionally, the determining the temporary state at the time point according to the fluctuating energy at the time point includes:

If the fluctuating energy E(n)>0 at the current time point, it is determined that the temporary state St(n) at the current time point is the rising mode;

If the fluctuating energy E(n) at the current time point is less than 0, it is determined that the temporary state St(n) at the current time point is in a descending mode;

If the fluctuating energy E(n) at the current time point is equal to 0, it is determined that the temporary state St(n) at the current time point is a steady mode.

Optionally, the determining the stable state at the time point according to the absolute value of the fluctuating energy at the time point and the temporary state includes:

If the absolute value of fluctuating energy abs(E(n)) at the current time point is greater than or equal to the set fluctuating energy threshold, then the temporary state St(n) at the current time point is taken as the stable state at the current time point;

If the absolute value abs(E(n)) of fluctuating energy at the current time point is less than the fluctuating energy threshold, and the temporary state St(n) at the current time point is the same as the temporary state St(n-1) at the previous time point , The last stable state is regarded as the stable state at the current time point;

If the absolute value abs(E(n)) of fluctuating energy at the current time point is less than the fluctuating energy threshold, and the temporary state St(n) at the current time point is different from the temporary state St(n-1) at the previous time point , The dynamic observation window is used to track the unstable state, and the stable state at each time point in the observation window is determined.

Optionally, the use of the dynamic observation window to track the unstable state, and determining the stable state at the current time point includes:

Open the dynamic observation window;

Record the length of the observation window and calculate the cumulative energy sum(E) and the absolute value of the cumulative energy abs(sum(E)) at all time points in the observation window;

Determine whether the observation window meets the termination condition;

If so, terminate the observation window, and determine the stable state at each time point in the observation window according to the termination condition.

Optionally, the termination condition includes any one of the following:

A specific time point appears in the observation window, and the specific time point refers to a time point at which the absolute value of fluctuating energy abs(E(i)) is greater than or equal to the fluctuating energy threshold;

The observation window reaches the set length threshold w;

The absolute value abs(sum(E)) of the accumulated energy at all time points in the observation window is greater than or equal to the fluctuating energy threshold;

The determining a stable state at each time point in the observation window according to the termination condition includes:

In the case where the observation window terminated at the specific time point appears in the observation window: move the termination point of the observation window to a time point before the specific time point to obtain an updated observation window; if updated The length of the observation window is greater than half of the set length threshold w, then the stable state at each time point in the updated observation window is set to the last stable state, otherwise the stable state at each time point in the updated observation window Set to steady mode;

When the observation window reaches the set length threshold w to terminate the observation window: if the absolute value of the cumulative energy abs(sum(E)) is greater than the fluctuating energy threshold, and the cumulative energy sum(E) >0, the stable state at each time point in the observation window is set to the rising mode; if the absolute value of the cumulative energy abs(sum(E)) is greater than the fluctuating energy threshold, and the cumulative energy sum( E)<0, the stable state at each time point in the observation window is set to the falling mode; if the absolute value of the cumulative energy abs(sum(E)) is less than or equal to the fluctuating energy threshold, the The stable state at each time point in the observation window is set to stable mode;

When the absolute value of the cumulative energy abs(sum(E)) at all time points in the observation window is greater than or equal to the fluctuating energy threshold to terminate the observation window: if the absolute value of the cumulative energy abs(sum(E) ) Is greater than the fluctuating energy threshold, and the cumulative energy sum(E)>0, then the stable state at each time point in the observation window is set to the rising mode; if the absolute value of the cumulative energy abs(sum (E)) is greater than the fluctuating energy threshold, and the accumulated energy sum(E)<0, then the stable state at each time point in the observation window is set to the falling mode.

Optionally, the acquiring time series data includes:

Sampling the timing diagram to obtain time series data.

Optionally, the sampling of the timing diagram includes:

Perform periodic sampling on the timing diagram, or perform aperiodic sampling on the timing diagram.

Optionally, the method further includes:

Show the state mode at each time point in the time series data.

A time series data processing device, the device comprising:

A data acquisition module for acquiring time series data, where each data in the time series is a value at a time point;

The fluctuating energy calculation module is used to sequentially calculate the fluctuating energy at each time point E(n)=V(n)-V(n-1), where E(n) is the fluctuating energy at the current time point, V(n) Is the value at the current time point, and V(n-1) is the value at the previous time point;

The traversal module is configured to traverse each time point in the time sequence in a loop, and determine the state of the time point according to the fluctuating energy at each time point, and the state includes: a temporary state and a stable state;

The output module is configured to use the stable state at the time point as the state mode at the time point; the state mode includes: a steady mode, an ascending mode, and a descending mode.

Optionally, the traversal module includes:

A temporary state determining module, configured to loop through each time point in the time sequence, and determine the temporary state at the time point according to the fluctuating energy at the time point;

The stable state determining module is used to determine the stable state at the time point according to the absolute value of the fluctuating energy at the time point and the temporary state.

Optionally, the temporary state determination module is specifically configured to determine that the temporary state St(n) at the current time point is in the rising mode when the fluctuating energy E(n) at the current time point is in the rising mode; the fluctuation at the current time point When the energy E(n)<0, the temporary state St(n) at the current time point is determined to be a descending mode; when the fluctuating energy E(n) at the current time point is 0, the temporary state St(n) at the current time point is determined It is a steady mode.

Optionally, the stable state determination module includes:

The judging unit is used to determine whether the absolute value of fluctuating energy abs(E(n)) at the current time point is greater than or equal to the set fluctuating energy threshold, and the temporary state St(n) at the current time point and the temporary state at the previous time point Whether the state St(n-1) is the same;

The first steady-state determination unit is used to use the temporary state St(n) at the current time point as the current time when the absolute value of the fluctuating energy abs(E(n)) at the current time point is greater than or equal to the set fluctuating energy threshold Point of steady state;

The second steady-state determination unit is used for the absolute value abs(E(n)) of the fluctuating energy at the current time point is less than the fluctuating energy threshold, and the temporary state St(n) at the current time point and the previous time point When the temporary state St(n-1) is the same, the last stable state is taken as the stable state at the current time point;

The third steady state determination unit is used for the absolute value abs(E(n)) of the fluctuating energy at the current time point is less than the fluctuating energy threshold, and the temporary state St(n) at the current time point and the previous time point When the temporary state St(n-1) is different, the dynamic observation window is used to track the unstable state to determine the stable state at each time point in the observation window.

Optionally, the third steady-state determination unit is specifically configured to open a dynamic observation window, record the length of the observation window, and calculate the cumulative energy sum(E) and the cumulative energy at all time points in the observation window The absolute value of abs(sum(E)), terminates the observation window after the observation window meets the termination condition, and determines the stable state at each time point in the observation window according to the termination condition.

Optionally, the termination condition includes any one of the following:

A specific time point appears in the observation window, and the specific time point refers to a time point at which the absolute value of fluctuating energy abs(E(i)) is greater than or equal to the fluctuating energy threshold;

The observation window reaches the set length threshold w;

The absolute value abs(sum(E)) of the accumulated energy at all time points in the observation window is greater than or equal to the fluctuating energy threshold;

The third steady state determination unit determines the steady state at each time point in the observation window in the following manner:

In the case where the observation window terminated at the specific time point appears in the observation window: move the termination point of the observation window to a time point before the specific time point to obtain an updated observation window; if updated The length of the observation window is greater than half of the set length threshold w, then the stable state at each time point in the updated observation window is set to the last stable state, otherwise the stable state at each time point in the updated observation window Set to steady mode;

When the observation window reaches the set length threshold w to terminate the observation window: if the absolute value of the cumulative energy abs(sum(E)) is greater than the fluctuating energy threshold, and the cumulative energy sum(E) >0, the stable state at each time point in the observation window is set to the rising mode; if the absolute value of the cumulative energy abs(sum(E)) is greater than the fluctuating energy threshold, and the cumulative energy sum( E)<0, the stable state at each time point in the observation window is set to the falling mode; if the absolute value of the cumulative energy abs(sum(E)) is less than or equal to the fluctuating energy threshold, the The stable state at each time point in the observation window is set to stable mode;

When the absolute value of the cumulative energy abs(sum(E)) at all time points in the observation window is greater than or equal to the fluctuating energy threshold to terminate the observation window: if the absolute value of the cumulative energy abs(sum(E) ) Is greater than the fluctuating energy threshold, and the cumulative energy sum(E)>0, then the stable state at each time point in the observation window is set to the rising mode; if the absolute value of the cumulative energy abs(sum (E)) is greater than the fluctuating energy threshold, and the accumulated energy sum(E)<0, then the stable state at each time point in the observation window is set to the falling mode.

Optionally, the data acquisition module is specifically configured to sample the time sequence diagram to obtain time series data.

Optionally, the data acquisition module sampling the timing diagram includes: periodically sampling the timing diagram, or non-periodically sampling the timing diagram.

Optionally, the device further includes:

The display module is used to display the state mode of each time point in the time series data.

An electronic device, including: one or more processors and memories;

The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the aforementioned method.

A readable storage medium having instructions stored thereon, and the instructions are executed to implement the aforementioned method.

The time series data processing method and device provided by the embodiments of the present invention calculate and calculate the fluctuation energy at each time point for the time series data, and use the fluctuation energy at each time point to loop through the time series at each time point. The method determines the state mode at each time point.

Further, the division of steady state and non-steady state is proposed. In the traversal process, first determine the temporary state of the time point according to the fluctuating energy at each time point; then according to the absolute value of the fluctuating energy and the temporary state of the time point By determining the stable state at the time point, the state mode at each time point can be obtained.

Further, when the fluctuating energy at the current time point is small, if the temporary state at the current time point is different from the temporary state at the previous time point, the dynamic observation window is used to track the unstable state to determine the stability of each time point in the observation window State, that is to say, when it is impossible to determine the state pattern at the current time point only based on the fluctuating energy at the current time point and the temporary state at the previous time point, consider the temporary status at one or more time points after the current time point. The state condition can finally determine the state mode at the current time point, so that the determination result of the state mode at each time point can be more accurate, and the continuation of the temporary state can be accurately divided effectively, reducing the impact of the short-term change of the energy direction. Improper segmentation of time-continuous state patterns. In addition, the non-steady state tracking can accurately determine the impact of a certain length or time of energy accumulation changes on the current point in time, which improves the accuracy of state mode judgment.

Description of the drawings

In order to more clearly describe the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed in the embodiments. Obviously, the drawings in the following description are only those described in the present invention. For some of the embodiments, for those of ordinary skill in the art, other drawings may be obtained based on these drawings.

Fig. 1 is a flowchart of a time series data processing method according to an embodiment of the present invention;

FIG. 2 is a flow chart of using a dynamic observation window to track unstable states in an embodiment of the present invention;

3 is a structural block diagram of a time series data processing device according to an embodiment of the present invention;

Fig. 4 is another structural block diagram of a time series data processing device according to an embodiment of the present invention.

Detailed ways

In order to enable those skilled in the art to better understand the solutions of the embodiments of the present invention, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings and implementation manners.

The embodiment of the present invention provides a time series data processing method and device. By calculating the fluctuating energy at each time point in the time series data, using the fluctuating energy at each time point to cycle through each time point in the time series, according to The fluctuating energy at each time point determines the state at the time point, and the state includes: a temporary state and a stable state; and the stable state at the time point is taken as the state mode at the time point.

As shown in FIG. 1, it is a flowchart of a time series data processing method according to an embodiment of the present invention, which includes the following steps:

Step 101: Obtain time series data, where each data in the time series is a value at a time point.

In the embodiment of the present invention, the time series data may be sampling data for an arbitrary waveform, that is, a timing diagram, and may be periodic sampling or aperiodic sampling, which is not limited in the embodiment of the present invention.

The data in the time series data is sorted in ascending order of time or index.

Step 102: Calculate the fluctuation energy E(n)=V(n)-V(n-1) at each time point in sequence.

Among them, E(n) is the fluctuating energy at the current time point, V(n) is the value at the current time point, and V(n-1) is the value at the previous time point.

For the first time point in the time series data, the fluctuation energy can be set to 0, that is, E(1)=0.

Step 103: Loop through each time point in the time sequence, and determine the state of the time point according to the fluctuating energy at each time point, and the state includes: a temporary state and a stable state.

Since the fluctuating energy only reflects the change between the current time point and the previous time point, it is only a relative state, not a final state. The final state of each time point is not only related to the state of the previous time point, but also It may also be related to the state at one or more subsequent points in time. Therefore, in the embodiment of the present invention, the temporary state at that time can be determined according to the fluctuating energy at each time point, and then the absolute value and the temporary state of the fluctuating energy at the time point can be determined by loop traversal. The steady state at that point in time. For the convenience of description, the absolute value of fluctuating energy will be recorded as abs(E) later.

The temporary state indicates that the state mode at the current time point has not been finalized, and will be affected by the state mode at one or more subsequent time points; the stable state indicates that the state mode at the time point has stabilized and is not affected by the subsequent time. The impact of changes in the state of the point. In the embodiment of the present invention, there can be three state modes at each time point, namely: a steady mode, an ascending mode, and a descending mode. In addition, in practical applications, some applications only need to distinguish whether the trend of the data in the sequence is stable. Therefore, according to different application requirements, the state mode can also be divided into a stable mode and a fluctuating mode. Of course, there may be more divisions with different granularities, which is not limited in the embodiment of the present invention.

The following takes the status mode including steady mode, rising mode, and falling mode as examples for description. For the convenience of description, use 0, 1, and 2 to represent the above three state modes.

Correspondingly, the temporary state and the stable state also respectively include the above three state modes. For the convenience of description, the temporary state is denoted as St and the stable state is denoted as S.

When determining the temporary state at each point in time, the following principles can be used:

If the fluctuating energy E(n)>0 at the current time point, the temporary state St(n) at the current time point is determined to be in the rising mode, denoted as St(n)=1;

If the fluctuating energy E(n) at the current time point is less than 0, it is determined that the temporary state St(n) at the current time point is in a descending mode, which is recorded as St(n)=2;

If the fluctuating energy E(n)=0 at the current time point, then the temporary state St(n) at the current time point is determined to be a steady mode, which is recorded as St(n)=0.

It should be noted that, for the first time point in the time series data, its temporary state can be set to a stationary mode, that is, St(1)=0.

When determining the stable state at each time point, the following principles can be used:

If the absolute value abs(E(n)) of the fluctuating energy at the current time point is greater than or equal to the set fluctuating energy threshold e, that is, abs(E(n))>=e, then the temporary state St(n) at the current time point ) As the stable state at the current time point, that is, the temporary state at the current time point becomes a stable state;

If the absolute value abs(E(n)) of the fluctuating energy at the current time point is less than the fluctuating energy threshold e, and the temporary state St(n) at the current time point and the temporary state St(n-1) at the previous time point Same, that is, abs(E(n))<e and St(n)=St(n-1), then the last stable state (denoted as Sp) is taken as the stable state at the current time point, that is, the current time The stable state of the point inherits the last stable state Sp; it should be noted that the last stable state refers to the last stable state before the current point in time, not the stable state at the previous point in time;

If the absolute value abs(E(n)) of the fluctuating energy at the current time point is less than the fluctuating energy threshold e, and the temporary state St(n) at the current time point and the temporary state St(n-1) at the previous time point Different, namely abs(E(n))<e and St(n)! = St(n-1), then use the dynamic observation window to track the unstable state, and determine the stable state at each time point in the observation window.

The process of using the dynamic observation window to track the unstable state will be described in detail later.

Step 104: Use the stable state at the time point as the state mode at the time point; the state mode includes: a steady mode, an ascending mode, and a descending mode.

As mentioned earlier, the stable state means that the state mode at that time point has been stabilized and is not affected by the state changes at subsequent time points. Therefore, after the stable state at each time point is determined, each time point can be changed The steady state output. Of course, it is also possible to determine the stable state at a time point every time, that is, to output the state mode at that time point, which is not limited in the embodiment of the present invention.

Different status modes can be represented by different marks. For example, the numbers 0, 1, and 2 mentioned above represent steady mode, rising mode, and falling mode respectively. Correspondingly, when outputting, the state mode corresponding to each time point can be output in the form of a state mode sequence or a table. Further, in order to make the changes of the data look more intuitive, the state mode of each time point in the time series data can also be displayed in the form of a waveform diagram, wherein different state modes can be displayed in different forms, such as using Different colors or shapes represent different state modes of data points.

As shown in Fig. 2, it is a flow chart of using a dynamic observation window to track an unstable state in an embodiment of the present invention, which includes the following steps:

Step 201: Open the dynamic observation window.

The dynamic observation window refers to that the observation window is dynamically changing, that is, it extends backward in sequence from the current time point when the dynamic observation window is opened, adding a subsequent time point each time. Moreover, each time a subsequent time point is added in the observation window, the following step 202 needs to be executed again.

Step 202: Record the length of the observation window and calculate the cumulative energy sum(E) and the absolute value of the cumulative energy abs(sum(E)) at all time points in the observation window.

Step 203: Determine whether the observation window meets the termination condition; if so, perform step 204; otherwise, return to step 202.

Due to the uncertainty of time series data changes, different termination conditions can be set for different changes characteristics. In the embodiment of the present invention, the termination condition may include any one of the following:

(1) A specific time point appears in the observation window, and the specific time point refers to a time point at which the absolute value of fluctuating energy abs(E(i)) is greater than or equal to the fluctuating energy threshold;

(2) The observation window reaches the set length threshold w;

(3) The absolute value abs(sum(E)) of the accumulated energy at all time points in the observation window is greater than or equal to the fluctuating energy threshold.

That is, as long as any one of the above conditions is met, the observation window is immediately terminated.

Step 204: Terminate the observation window, and determine a stable state at each time point in the observation window according to the termination condition.

Since the time points in the observation window will have different characteristics under different termination conditions, it is necessary to determine the stable state of the time points in the observation window according to its characteristics to ensure the final determination of each time The accuracy of the point's state pattern.

Corresponding to the above termination conditions, there can be the following specific situations:

1) In the case where the observation window is terminated at the specified time point in the observation window: move the termination point of the observation window to a time point before the specified time point to obtain the updated observation window; if The length of the updated observation window is greater than half of the set length threshold w, then the stable state of each time point in the updated observation window is set to the last stable state Sp, otherwise the time point in the updated observation window is set The steady state is set to steady mode;

2) When the observation window reaches the set length threshold w to terminate the observation window: if the absolute value of the cumulative energy abs(sum(E)) is greater than the fluctuating energy threshold, that is, abs(sum(E) )>e, and the accumulated energy sum(E)>0, set the stable state at each time point in the observation window to the rising mode; if the absolute value of the accumulated energy abs(sum(E)) is greater than If the fluctuating energy threshold is abs(sum(E))>e, and the accumulated energy sum(E)<0, then the stable state at each time point in the observation window is set to the falling mode; if the The absolute value of the accumulated energy abs(sum(E)) is less than or equal to the fluctuating energy threshold, that is, abs(sum(E))<=e, then the stable state at each time point in the observation window is set to the stable mode;

3) In the case where the absolute value of the cumulative energy abs(sum(E)) at all time points in the observation window is greater than or equal to the fluctuating energy threshold to terminate the observation window: if the absolute value of the cumulative energy abs(sum( E)) is greater than the fluctuating energy threshold, that is, abs(sum(E))>e, and the accumulated energy sum(E)>0, then the stable state at each time point in the observation window is set to rise Mode; if the absolute value of the cumulative energy abs(sum(E)) is greater than the fluctuating energy threshold, and the cumulative energy sum(E)<0, set the stable state at each time point in the observation window It is down mode.

The time series data processing method provided by the embodiment of the present invention calculates the fluctuation energy at each time point for time series data, and uses the fluctuation energy at each time point to determine by looping through each time point in the time series State mode at each point in time. Further, the division of steady state and non-steady state is proposed. In the traversal process, first determine the temporary state of the time point according to the fluctuating energy at each time point; then according to the absolute value of the fluctuating energy and the temporary state of the time point By determining the stable state at the time point, the state mode at each time point can be obtained.

Further, when the fluctuating energy at the current time point is small, if the temporary state at the current time point is different from the temporary state at the previous time point, the dynamic observation window is used to track the unstable state to determine the stability of each time point in the observation window State, that is to say, when it is impossible to determine the state pattern at the current time point only based on the fluctuating energy at the current time point and the temporary state at the previous time point, consider the temporary status at one or more time points after the current time point. The state condition can finally determine the state mode at the current time point, so that the determination result of the state mode at each time point can be more accurate, and the continuation of the temporary state can be accurately divided effectively, reducing the impact of the short-term change of the energy direction. Improper segmentation of time-continuous state patterns. In addition, the non-steady state tracking can accurately determine the impact of a certain length or time of energy accumulation changes on the current point in time, which improves the accuracy of state mode judgment.

Correspondingly, the embodiment of the present invention also provides a time series data processing device, as shown in FIG. 3, which is a structural block diagram of the device.

In this embodiment, the device includes the following modules:

The data acquisition module 301 is used to acquire time series data. Each data in the time series is a value at a time point; for example, the time series diagram is sampled periodically or non-periodically to obtain the time series data, or the user Sampling data provided by time, etc.;

The fluctuating energy calculation module 302 is used to sequentially calculate the fluctuating energy E(n)=V(n)-V(n-1) at each time point, where E(n) is the fluctuating energy at the current time point, and V(n) ) Is the value at the current time point, and V(n-1) is the value at the previous time point;

The traversal module 303 is configured to loop through each time point in the time sequence, and determine the state of the time point according to the fluctuating energy at each time point, and the state includes: a temporary state and a stable state;

The output module 304 is configured to use the stable state at the time point as the state mode at the time point; the state mode includes: a steady mode, an ascending mode, and a descending mode.

Since the fluctuating energy only reflects the change between the current time point and the previous time point, it is only a relative state, not a final state. The final state of each time point is not only related to the state of the previous time point, but also It may also be related to the state at one or more subsequent points in time. Therefore, in the embodiment of the present invention, the temporary state at that time can be determined according to the fluctuating energy at each time point, and then the absolute value and the temporary state of the fluctuating energy at the time point can be determined by loop traversal. The steady state at that point in time. Correspondingly, the aforementioned traversal module may include: a temporary state determination module and a stable state determination module; wherein:

The temporary state determining module is configured to loop through each time point in the time sequence, and determine the temporary state at the time point according to the fluctuating energy at the time point;

The stable state determining module is configured to determine the stable state at the time point according to the absolute value of the fluctuating energy at the time point and the temporary state.

The above-mentioned temporary state determination module can specifically determine the temporary state at each time point according to the following principles: when the fluctuating energy at the current time point E(n)>0, determine that the temporary state St(n) at the current time point is in the rising mode; When the fluctuating energy E(n)<0 at the time point, the temporary state St(n) at the current time point is determined to be a descending mode; when the fluctuating energy E(n) at the current time point is 0, the temporary state at the current time point is determined St(n) is the stationary mode. In addition, for the first time point in the time series data, its temporary state can be set to a stationary mode, that is, St(1)=0.

The aforementioned stable state determination module may specifically include the following units:

The judging unit is used to determine whether the absolute value of fluctuating energy abs(E(n)) at the current time point is greater than or equal to the set fluctuating energy threshold, and the temporary state St(n) at the current time point and the temporary state at the previous time point Whether the state St(n-1) is the same;

The first steady-state determination unit is used to use the temporary state St(n) at the current time point as the current time when the absolute value of the fluctuating energy abs(E(n)) at the current time point is greater than or equal to the set fluctuating energy threshold Point of steady state;

The second steady-state determination unit is used for the absolute value abs(E(n)) of the fluctuating energy at the current time point is less than the fluctuating energy threshold, and the temporary state St(n) at the current time point and the previous time point When the temporary state St(n-1) is the same, the last stable state is taken as the stable state at the current time point;

The third steady state determination unit is used for the absolute value abs(E(n)) of the fluctuating energy at the current time point is less than the fluctuating energy threshold, and the temporary state St(n) at the current time point and the previous time point When the temporary state St(n-1) is different, the dynamic observation window is used to track the unstable state to determine the stable state at each time point in the observation window.

The third steady-state determination unit is specifically configured to open a dynamic observation window, record the length of the observation window, and calculate the cumulative energy sum(E) at all time points in the observation window and the absolute value of the cumulative energy abs( sum(E)), terminate the observation window after the observation window meets a termination condition, and determine the stable state at each time point in the observation window according to the termination condition.

For the termination condition and the manner of determining the stable state at each time point in the observation window according to the termination condition, please refer to the description in the foregoing method embodiment of the present invention, which will not be repeated here.

The time series data processing device provided by the embodiment of the present invention calculates and calculates the fluctuating energy at each time point for time series data, and uses the fluctuating energy at each time point to determine by looping through each time point in the time series State mode at each point in time. Further, the division of steady state and non-steady state is proposed. In the traversal process, first determine the temporary state of the time point according to the fluctuating energy at each time point; then according to the absolute value of the fluctuating energy and the temporary state of the time point By determining the stable state at the time point, the state mode at each time point can be obtained.

Further, when the fluctuating energy at the current time point is small, if the temporary state at the current time point is different from the temporary state at the previous time point, the dynamic observation window is used to track the unstable state to determine the stability of each time point in the observation window State, that is to say, when it is impossible to determine the state pattern at the current time point only based on the fluctuating energy at the current time point and the temporary state at the previous time point, consider the temporary status at one or more time points after the current time point. The state condition can finally determine the state mode at the current time point, so that the determination result of the state mode at each time point can be more accurate, and the continuation of the temporary state can be accurately divided effectively, reducing the impact of the short-term change of the energy direction. Improper segmentation of time-continuous state patterns. In addition, the non-steady state tracking can accurately determine the impact of a certain length or time of energy accumulation changes on the current point in time, which improves the accuracy of state mode judgment.

As shown in FIG. 4, it is another structural block diagram of the time series data processing device of the embodiment of the present invention.

Different from the embodiment shown in FIG. 3, in this embodiment, the apparatus further includes:

The display module 401 is used to display the state mode of each time point in the time series data.

Different status modes can be represented by different marks. For example, the numbers 0, 1, and 2 mentioned above represent steady mode, rising mode, and falling mode respectively. Correspondingly, when the output module 304 outputs the state mode at each time point, the state mode corresponding to each time point can be output in the form of a state mode sequence or a table.

Further, the display module 401 can display the state mode of each time point in the time series data in the form of a waveform graph, wherein different state modes can be displayed in different forms, such as using different colors or shapes to represent the data. Different state modes of points.

It should be noted that, for each embodiment of the above-mentioned time-series data processing device, since the function implementation of each module and unit is similar to that of the corresponding method, the description of each embodiment of the dialog generation device is relatively simple, and related For details, refer to the description of the corresponding part of the method embodiment.

The time series data processing method and device provided by the embodiments of the present invention can realize the segmentation of arbitrary waveform time series data and state mode judgment, such as electrocardiogram, electroencephalogram, current and voltage signals in manufacturing, and stock trading K-line, the time-domain waveform of the voice signal, etc., and is not affected by noise data, does not require training data, and has high accuracy and universality. According to the state mode at each time point in the time series data, the change point of the time series can be obtained, or different state modes can be combined into a compound mode, so as to provide effective information for industry analysis and application.

Compared with the prior art, the solution of the present invention can perform accurate state pattern recognition on arbitrarily sampled, without data distribution assumptions, with or without time series period, steady-state or non-steady-state time series data, which can not only realize Offline recognition, and online recognition can be realized, with strong versatility, and not restricted by the application environment.

It should be noted that the terms "first" and "second" in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and not necessarily used to describe a specific sequence or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations of them are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to the clearly listed Those steps or units may include other steps or units that are not clearly listed or are inherent to these processes, methods, products, or equipment.

The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. Moreover, the system embodiments described above are only illustrative, and the modules and units described as separate components may or may not be physically separated, that is, they may be located on one network unit, or may be distributed to multiple On the network unit. Some or all of the modules may be selected according to actual needs to achieve the objectives of the solutions of the embodiments. Those of ordinary skill in the art can understand and implement it without creative work.

A person of ordinary skill in the art can understand that all or part of the steps in the above-mentioned method embodiments can be implemented by a program instructing relevant hardware. The program can be stored in a computer-readable storage medium, which is referred to as storage herein. Medium, such as: ROM/RAM, magnetic disk, optical disk, etc.

Correspondingly, an embodiment of the present invention also provides a device for a time series data processing method. The device is an electronic device, such as a mobile terminal, a computer, a tablet device, a medical device, a fitness device, or a personal digital assistant. Wait. The electronic device may include one or more processors and memories; wherein the memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions to implement the foregoing method.

The embodiments of the present invention are described in detail above, and specific implementations are used to illustrate the present invention. The descriptions of the above embodiments are only used to help understand the methods and devices of the present invention, and they are only part of the embodiments of the present invention. Not all examples. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative work shall fall within the protection scope of the present invention, and the contents of this specification should not be construed as limiting the present invention. Therefore, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (20)

1. A time series data processing method, characterized in that the method includes:

   Acquiring time series data, where each data in the time series is a value at a time point;

   Calculate the fluctuating energy E(n)=V(n)-V(n-1) at each time point in turn, where E(n) is the fluctuating energy at the current time point, and V(n) is the value at the current time point, V(n-1) is the value at the previous time point;

   Loop through each time point in the time sequence, and determine the state of the time point according to the fluctuating energy at each time point, the state includes: a temporary state and a stable state;

   The stable state at the time point is taken as the state mode at the time point; the state mode includes: a steady mode, a rising mode, and a falling mode.
2. The method according to claim 1, wherein the looping through each time point in the time sequence, and determining the state of the time point according to fluctuating energy at each time point comprises:

   Loop through each time point in the time sequence, and determine the temporary state of the time point according to the fluctuating energy at the time point;

   The stable state at the time point is determined according to the absolute value of the fluctuating energy at the time point and the temporary state.
3. The method according to claim 2, wherein the determining the temporary state at the time point according to the fluctuating energy at the time point comprises:

   If the fluctuating energy E(n)>0 at the current time point, it is determined that the temporary state St(n) at the current time point is the rising mode;

   If the fluctuating energy E(n) at the current time point is less than 0, it is determined that the temporary state St(n) at the current time point is in a descending mode;

   If the fluctuating energy E(n) at the current time point is equal to 0, it is determined that the temporary state St(n) at the current time point is a steady mode.
4. The method according to claim 2, wherein the determining the stable state at the time point according to the absolute value of the fluctuating energy at the time point and the temporary state comprises:

   If the absolute value of fluctuating energy abs(E(n)) at the current time point is greater than or equal to the set fluctuating energy threshold, then the temporary state St(n) at the current time point is taken as the stable state at the current time point;

   If the absolute value abs(E(n)) of fluctuating energy at the current time point is less than the fluctuating energy threshold, and the temporary state St(n) at the current time point is the same as the temporary state St(n-1) at the previous time point , The last stable state is regarded as the stable state at the current time point;

   If the absolute value abs(E(n)) of fluctuating energy at the current time point is less than the fluctuating energy threshold, and the temporary state St(n) at the current time point is different from the temporary state St(n-1) at the previous time point , The dynamic observation window is used to track the unstable state, and the stable state at each time point in the observation window is determined.
5. The method according to claim 4, characterized in that said using the dynamic observation window to track the unstable state and determining the stable state at the current time point comprises:

   Open the dynamic observation window;

   Record the length of the observation window and calculate the cumulative energy sum(E) and the absolute value of the cumulative energy abs(sum(E)) at all time points in the observation window;

   Determine whether the observation window meets the termination condition;

   If so, terminate the observation window, and determine the stable state at each time point in the observation window according to the termination condition.
6. The method according to claim 5, wherein the termination condition includes any one of the following:

   A specific time point appears in the observation window, and the specific time point refers to a time point at which the absolute value of fluctuating energy abs(E(i)) is greater than or equal to the fluctuating energy threshold;

   The observation window reaches the set length threshold w;

   The absolute value abs(sum(E)) of the accumulated energy at all time points in the observation window is greater than or equal to the fluctuating energy threshold;

   The determining a stable state at each time point in the observation window according to the termination condition includes:

   In the case where the observation window terminated at the specific time point appears in the observation window: move the termination point of the observation window to a time point before the specific time point to obtain an updated observation window; if updated The length of the observation window is greater than half of the set length threshold w, then the stable state at each time point in the updated observation window is set to the last stable state, otherwise the stable state at each time point in the updated observation window Set to steady mode;

   When the observation window reaches the set length threshold w to terminate the observation window: if the absolute value of the cumulative energy abs(sum(E)) is greater than the fluctuating energy threshold, and the cumulative energy sum(E) >0, the stable state at each time point in the observation window is set to the rising mode; if the absolute value of the cumulative energy abs(sum(E)) is greater than the fluctuating energy threshold, and the cumulative energy sum( E)<0, the stable state at each time point in the observation window is set to the falling mode; if the absolute value of the cumulative energy abs(sum(E)) is less than or equal to the fluctuating energy threshold, the The stable state at each time point in the observation window is set to stable mode;

   When the absolute value of the cumulative energy abs(sum(E)) at all time points in the observation window is greater than or equal to the fluctuating energy threshold to terminate the observation window: if the absolute value of the cumulative energy abs(sum(E) ) Is greater than the fluctuating energy threshold, and the cumulative energy sum(E)>0, then the stable state at each time point in the observation window is set to the rising mode; if the absolute value of the cumulative energy abs(sum (E)) is greater than the fluctuating energy threshold, and the accumulated energy sum(E)<0, then the stable state at each time point in the observation window is set to the falling mode.
7. The method according to claim 1, wherein said acquiring time series data comprises:

   Sampling the timing diagram to obtain time series data.
8. The method according to claim 1, wherein the sampling a timing diagram comprises:

   Perform periodic sampling on the timing diagram, or perform aperiodic sampling on the timing diagram.
9. The method according to any one of claims 1 to 8, wherein the method further comprises:

   Show the state mode at each time point in the time series data.
10. A time series data processing device, characterized in that the device includes:

    A data acquisition module for acquiring time series data, where each data in the time series is a value at a time point;

    The fluctuating energy calculation module is used to sequentially calculate the fluctuating energy at each time point E(n)=V(n)-V(n-1), where E(n) is the fluctuating energy at the current time point, V(n) Is the value at the current time point, and V(n-1) is the value at the previous time point;

    The traversal module is configured to traverse each time point in the time sequence in a loop, and determine the state of the time point according to the fluctuating energy at each time point, and the state includes: a temporary state and a stable state;

    The output module is configured to use the stable state at the time point as the state mode at the time point; the state mode includes: a steady mode, an ascending mode, and a descending mode.
11. The device according to claim 10, wherein the traversal module comprises:

    A temporary state determining module, configured to loop through each time point in the time sequence, and determine the temporary state at the time point according to the fluctuating energy at the time point;

    The stable state determining module is used to determine the stable state at the time point according to the absolute value of the fluctuating energy at the time point and the temporary state.
12. The device according to claim 11, wherein:

    The temporary state determining module is specifically used to determine that the temporary state St(n) at the current time point is in the rising mode when the fluctuating energy E(n) at the current time point is in the rising mode; the fluctuating energy E(n) at the current time point When )<0, it is determined that the temporary state St(n) at the current time point is a descending mode; when the fluctuating energy E(n) at the current time point is 0, it is determined that the temporary state St(n) at the current time point is a steady mode.
13. The device according to claim 11, wherein the stable state determining module comprises:

    The judging unit is used to determine whether the absolute value of fluctuating energy abs(E(n)) at the current time point is greater than or equal to the set fluctuating energy threshold, and the temporary state St(n) at the current time point and the temporary state at the previous time point Whether the state St(n-1) is the same;

    The first steady-state determination unit is used to use the temporary state St(n) at the current time point as the current time when the absolute value of the fluctuating energy abs(E(n)) at the current time point is greater than or equal to the set fluctuating energy threshold Point of steady state;

    The second steady-state determination unit is used for the absolute value abs(E(n)) of the fluctuating energy at the current time point is less than the fluctuating energy threshold, and the temporary state St(n) at the current time point and the previous time point When the temporary state St(n-1) is the same, the last stable state is taken as the stable state at the current time point;

    The third steady-state determination unit is used for the absolute value abs(E(n)) of the fluctuating energy at the current time point is less than the fluctuating energy threshold, and the temporary state St(n) at the current time point and the previous time point When the temporary state St(n-1) is different, the dynamic observation window is used to track the unstable state to determine the stable state at each time point in the observation window.
14. The device according to claim 13, wherein:

    The third steady-state determination unit is specifically configured to open a dynamic observation window, record the length of the observation window, and calculate the cumulative energy sum(E) at all time points in the observation window and the absolute value abs of the cumulative energy (sum(E)), terminate the observation window after the observation window meets a termination condition, and determine the stable state at each time point in the observation window according to the termination condition.
15. The device according to claim 14, wherein the termination condition comprises any one of the following:

    A specific time point appears in the observation window, and the specific time point refers to a time point at which the absolute value of fluctuating energy abs(E(i)) is greater than or equal to the fluctuating energy threshold;

    The observation window reaches the set length threshold w;

    The absolute value abs(sum(E)) of the accumulated energy at all time points in the observation window is greater than or equal to the fluctuating energy threshold;

    The third steady state determination unit determines the steady state at each time point in the observation window in the following manner:

    In the case where the observation window terminated at the specific time point appears in the observation window: move the termination point of the observation window to a time point before the specific time point to obtain an updated observation window; if updated The length of the observation window is greater than half of the set length threshold w, then the stable state at each time point in the updated observation window is set to the last stable state, otherwise the stable state at each time point in the updated observation window Set to steady mode;

    When the observation window reaches the set length threshold w to terminate the observation window: if the absolute value of the cumulative energy abs(sum(E)) is greater than the fluctuating energy threshold, and the cumulative energy sum(E) >0, the stable state at each time point in the observation window is set to the rising mode; if the absolute value of the cumulative energy abs(sum(E)) is greater than the fluctuating energy threshold, and the cumulative energy sum( E)<0, the stable state at each time point in the observation window is set to the falling mode; if the absolute value of the accumulated energy abs(sum(E)) is less than or equal to the fluctuating energy threshold, then the The stable state at each time point in the observation window is set to stable mode;

    When the absolute value of the cumulative energy abs(sum(E)) at all time points in the observation window is greater than or equal to the fluctuating energy threshold to terminate the observation window: if the absolute value of the cumulative energy abs(sum(E) ) Is greater than the fluctuating energy threshold, and the cumulative energy sum(E)>0, then the stable state at each time point in the observation window is set to the rising mode; if the absolute value of the cumulative energy abs(sum (E)) is greater than the fluctuating energy threshold, and the accumulated energy sum(E)<0, then the stable state at each time point in the observation window is set to the falling mode.
16. The device according to claim 10, wherein:

    The data acquisition module is specifically used to sample the time sequence diagram to obtain time series data.
17. The device according to claim 10, wherein the sampling of the timing diagram by the data acquisition module comprises: sampling the timing diagram periodically, or sampling the timing diagram aperiodically.
18. The device according to any one of claims 10 to 17, wherein the device further comprises:

    The display module is used to display the state mode of each time point in the time series data.
19. An electronic device, characterized in that it includes: one or more processors and memories;

    The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the method according to any one of claims 1 to 9.
20. A readable storage medium having instructions stored thereon, and the instructions are executed to implement the method according to any one of claims 1 to 9.

Images