Service scene time sequence data determination method and system based on multi-time scale fusion

文档序号：1952688 发布日期：2021-12-10 浏览：20次中文

阅读说明：本技术 基于多时间尺度融合的业务场景时序数据确定方法及系统 (Service scene time sequence data determination method and system based on multi-time scale fusion ) 是由闫军纪双西于 2021-09-22 设计创作，主要内容包括：本发明公开一种基于多时间尺度融合的业务场景时序数据确定方法及系统,涉及智能停车管理领域,包括：利用历史业务时序数据以及相关数据,通过进行分钟级和天级两种时间尺度的数据分解,并根据分解后的时序数据特征分别采用不同的方法进行时序预测,并最终通过两种时间尺度时序数据的融合,从而实现了更为稳定的以天为时间单位的长周期趋势性波动预测,还可以利用提取的分钟级24小时时序模式实现未来时间段内的细粒度业务量预测,并且由于利用了信号分解技术,使得整体计算流程更为简洁稳定,无需复杂的调参过程以及包含多个长业务周期的大量时序数据的计算,从而提升了业务场景时序数据的确定效率。(The invention discloses a method and a system for determining service scene time sequence data based on multi-time scale fusion, which relate to the field of intelligent parking management and comprise the following steps: by utilizing historical service time sequence data and related data, time sequence prediction is carried out by adopting different methods according to the time sequence data characteristics after decomposition through data decomposition of a minute-scale time scale and a day-scale time scale, and finally through fusion of the time sequence data of the two time scales, more stable long-period trend fluctuation prediction with a day as a time unit is realized, fine-grained service volume prediction in a future time period can be realized by utilizing an extracted minute-scale 24-hour time sequence mode, and the whole calculation process is more concise and stable due to the utilization of a signal decomposition technology, a complex parameter adjusting process and calculation of a large amount of time sequence data containing a plurality of long service periods are not needed, so that the determination efficiency of the service scene time sequence data is improved.)

1. A method for determining service scene time sequence data based on multi-time scale fusion is characterized by comprising the following steps:

performing trend and local periodic two-component mode decomposition on the abnormal minute-level time sequence data to obtain a trend partial sequence and a local periodic mode partial sequence;

performing data integration on the trend partial sequence by taking days as sampling units, and performing secondary decomposition on the integrated day-level time sequence data through a preset decomposition algorithm to obtain a smoothness trend component and a smooth fluctuation component corresponding to the day-level time sequence data;

respectively carrying out time sequence prediction on the smoothness trend component and the smooth fluctuation component to obtain a smoothness trend component estimation sequence, a smooth fluctuation component estimation sequence and a residual error statistic estimation value of the smooth fluctuation component;

acquiring trend prediction data corresponding to day-level service time sequence data according to the smoothness trend component estimation sequence, the stationary wave component estimation sequence and the residual error statistic estimation value of the stationary wave component;

and acquiring an average value of data sampling frequency in the day according to the day-level trend prediction data, and merging the average value with the partial periodic mode sequence of the minute-level time sequence data one by one to obtain a minute-level long-time traffic estimation value.

2. The method for determining service scene time series data based on multi-time scale fusion as claimed in claim 1, wherein before the step of performing trend and local periodic two-component mode decomposition on the abnormal minute-level time series data, the method further comprises:

acquiring time sequence data of minute-level sampling frequency;

and when the minute-level time sequence data has partial sampling point data loss or obvious abnormal points, correcting the abnormal data by using the average value of the minute-level time sequence data.

3. The method for determining the service scene time series data based on the multi-time scale fusion as claimed in claim 1, wherein the step of performing the secondary decomposition on the integrated day-level time series data by using a preset decomposition algorithm comprises:

and carrying out secondary decomposition on the integrated day-level time sequence data through an Empirical Mode Decomposition (EMD) algorithm.

4. The method for determining the service scene time series data based on the multi-time scale fusion as claimed in claim 1, wherein the step of respectively performing the time series prediction on the smoothness trend component and the stationary volatility component to obtain the smoothness trend component estimation sequence, the stationary volatility component estimation sequence, and the stationary component residual statistic estimation sequence comprises:

predicting the length np of the smoothness trend component T _ smooth by an ARIMA algorithm to generate a smoothness trend component estimation sequence Tpred _ smooth, wherein np is a prediction time step in days;

predicting the length np of the stationary volatility component T _ station by using a Singular Spectrum Analysis (SSA) algorithm to generate a stationary volatility component estimation sequence Tpred _ station and a data fitting part Tsit _ station;

and carrying out statistical calculation aiming at an estimation error to obtain a residual statistic estimation value of the stationary fluctuation component, wherein the estimation error is calculated according to a formula Terr _ station-T _ station-Tf _ station.

5. The method for determining service scene time series data based on multi-time scale fusion as claimed in claim 4, wherein the step of obtaining trend prediction data corresponding to day-level service time series data according to the smoothness trend component estimation sequence, stationary volatility component estimation sequence, and residual statistic estimation value of stationary volatility component comprises:

acquiring trend prediction data corresponding to the day-level service time sequence data according to a formula Tpred _ trend ═ Tpred _ smooth + Tpred _ station + u, wherein u is an average value in statistic estimation aiming at a residual sequence of a stationary fluctuation component;

and calculating upper and lower bound error limits corresponding to the antenna-level service time sequence data according to formulas Tpred _ trend _ upper and Tpred _ trend + std and Tpred _ trend _ low as Tpred _ trend-std, wherein std is a standard deviation in statistic estimation aiming at a residual sequence of the stationary volatility component.

6. The method for determining the service scene time series data based on the multi-time scale fusion as claimed in claim 5, wherein the step of obtaining the average value of the sampling frequency of the data within the day according to the day-level trend prediction data, and merging the average value with the partial sequence of the local periodic pattern of the minute-level time series data for the period-by-period time series data to obtain the estimated value of the service volume of the minute-level long time comprises:

and calculating according to a formula tpred _ i + t _ search, wherein tpred _ i is a traffic volume estimation value in minutes, trend _ i is a minute-level average value of day-level trend prediction, and t _ search is day-mode traffic data.

7. A service scene time sequence data determination system based on multi-time scale fusion is characterized by comprising:

the decomposition module is used for performing trend and local periodic two-component mode decomposition on the abnormal minute-level time sequence data to obtain a trend partial sequence and a local periodic mode partial sequence;

the decomposition module is further used for performing data integration on the trend partial sequence by taking days as sampling units, and performing secondary decomposition on the integrated day-level time sequence data through a preset decomposition algorithm to obtain a smoothness trend component and a smooth volatility component corresponding to the day-level time sequence data;

the prediction module is used for respectively carrying out time sequence prediction on the smoothness trend component and the stationary volatility component to obtain a smoothness trend component estimation sequence, a stationary volatility component estimation sequence and a residual error statistic estimation value of the stationary volatility component;

the obtaining module is used for obtaining trend prediction data corresponding to the day-level service time sequence data according to the smoothness trend component estimation sequence, the stationary volatility component estimation sequence and the residual statistic estimation value of the stationary volatility component;

the acquisition module is further used for acquiring an average value of data sampling frequency in a day according to the day-level trend prediction data, and merging the average value with the minute-level time sequence data local period mode partial sequence one by one to obtain a minute-level long-time traffic estimation value.

8. The system according to claim 7, further comprising: a correction module;

the correction module is used for acquiring time sequence data of minute-level sampling frequency; and when the minute-level time sequence data has partial sampling point data loss or obvious abnormal points, correcting the abnormal data by using the average value of the minute-level time sequence data.

9. The system for determining timing data of service scenario based on multi-time scale fusion as claimed in claim 7,

and the decomposition module is specifically used for carrying out secondary decomposition on the integrated day-level time sequence data through an Empirical Mode Decomposition (EMD) algorithm.

10. The system for determining timing data of service scenario based on multi-time scale fusion as claimed in claim 7,

the prediction module is specifically configured to perform prediction with a length of np on the smoothness trend component T _ smooth through an ARIMA algorithm, and generate a smoothness trend component estimation sequence Tpred _ smooth, where np is a prediction time step in days; predicting the length np of the stationary volatility component T _ station by using a Singular Spectrum Analysis (SSA) algorithm to generate a stationary volatility component estimation sequence Tpred _ station and a data fitting part Tsit _ station; and carrying out statistical calculation aiming at an estimation error to obtain a residual statistic estimation value of the stationary fluctuation component, wherein the estimation error is calculated according to a formula Terr _ station-T _ station-Tf _ station.

11. The system according to claim 10, wherein the obtaining module is further configured to obtain trend prediction data corresponding to day-level traffic timing data according to a formula Tpred _ tend + Tpred _ station + u, where u is a mean value in statistic estimation for a residual sequence of stationary volatility components;

12. The system for determining timing data of service scenario based on multi-time scale fusion as claimed in claim 11,

the obtaining module is specifically configured to perform calculation according to a formula tpred _ i ═ trand _ i + t _ search, where tpred _ i is a traffic volume estimation value in minutes, trend _ i is a minute-level average value of day-level trend prediction, and t _ search is day-mode traffic data.

Technical Field

The invention relates to the field of intelligent parking management, in particular to a method and a system for determining service scene time sequence data based on multi-time scale fusion.

Background

In a static traffic scene, with the wide application of the AI technology, the automatic acquisition of parking state of a berthing vehicle, license plate and other vehicle information greatly improves the real-time efficiency of service processing, but in view of the technical bottleneck that the AI technology inevitably has and cannot achieve 100% recognition accuracy, efficient manual auxiliary processing plays a crucial role in improving the overall service level under the current technical conditions and service modes. In general, in an automatic information processing flow constructed by a series of algorithms, a system captures recognition results and related information which cannot be confirmed by the algorithms, wherein the information may include correct recognition results with low confidence coefficient or wrong recognition results, and the system generates system tasks from the information to be checked, returns the system tasks to background customer service staff for manual checking, and returns the corrected or confirmed results to a system database to complete the closed-loop operation of the whole processing flow. Therefore, today with the increasing labor cost, how to perform reasonable manual configuration according to the system traffic and perform flexible scheduling plays a crucial role in cost reduction and efficiency improvement of enterprises, and one of the most critical steps is how to perform reasonable and accurate prediction on the traffic according to historical traffic data information, which provides an important and necessary information basis for optimization of human configuration.

The currently common timing prediction methods are mainly divided into two modes according to the predictable duration: the method is characterized in that firstly, long-term trend prediction is carried out, a basic trend fluctuation mode of time sequence data under a long period time, such as a condition from months to a year, the algorithm carries out data fitting based on a set mathematical model, but a long-term sequence comprising a plurality of time sequence periods is required to carry out model fitting in calculation, when the time sequence data reflecting the long period rule is insufficient or the predicted time sequence is too long, a more accurate time sequence fluctuation evolution result cannot be obtained, in addition, more prior parameters are required to be preset according to sequence characteristics, and strong correlation exists between result accuracy and prior parameter setting, so that the method has strong limitation. The second method is short-term volatility prediction and is mainly used for evaluating the fluctuation mode of time sequence data in a short time, but because the method generally has no preset prior model for solving space control, large-scale data is required to be subjected to model training for extracting a complex and stable time sequence mode, and the method is sensitive to training data; meanwhile, a large amount of business data with novel business modes are difficult to obtain, and high-performance equipment is required to perform rapid training, evaluation, tuning and iteration models in the method, so that the method is difficult to implement and high in implementation cost.

Disclosure of Invention

In order to solve the technical problems, the invention provides a service scene time sequence data determination method and a service scene time sequence data determination system based on multi-time scale fusion, which can solve the problems that the conventional service scene time sequence data determination method is high in implementation difficulty and implementation cost, and the accuracy of service scene time sequence data prediction is difficult to guarantee.

In order to achieve the above object, in one aspect, the present invention provides a method for determining service scene time series data based on multi-time scale fusion, where the method includes: performing trend and local periodic two-component mode decomposition on the abnormal minute-level time sequence data to obtain a trend partial sequence and a local periodic mode partial sequence;

Further, before the step of performing trend and local periodic two-component pattern decomposition on the abnormal processed minute-level time series data, the method further includes:

acquiring time sequence data of minute-level sampling frequency;

Further, the step of performing secondary decomposition on the integrated day-level time sequence data through a preset decomposition algorithm comprises:

and carrying out secondary decomposition on the integrated day-level time sequence data through an Empirical Mode Decomposition (EMD) algorithm.

Further, the step of respectively performing time sequence prediction on the smoothness trend component and the stationary volatility component to obtain a smoothness trend component estimation sequence, a stationary volatility component estimation sequence, and a residual statistic estimation sequence of the stationary component includes:

Further, the step of obtaining trend prediction data corresponding to the day-level service time sequence data according to the smoothness trend component estimation sequence, the stationary volatility component estimation sequence and the residual statistic estimation value of the stationary volatility component includes:

Further, the step of obtaining an average value of data sampling frequency within a day according to the day-level trend prediction data, and merging the average value with the minute-level time series data local period mode partial sequence by the time series data of one period to obtain a minute-level long-time traffic estimation value includes:

In another aspect, the present invention provides a system for determining time series data of a service scene based on multi-time scale fusion, where the system includes: the decomposition module is used for performing trend and local periodic two-component mode decomposition on the abnormal minute-level time sequence data to obtain a trend partial sequence and a local periodic mode partial sequence;

Further, the system further comprises: a correction module;

Further, the decomposition module is specifically configured to perform secondary decomposition on the integrated day-level time sequence data through an empirical mode decomposition algorithm EMD. .

Further, the prediction module is specifically configured to perform prediction with a length np on the smoothness trend component T _ smooth through an ARIMA algorithm, and generate a smoothness trend component estimation sequence Tpred _ smooth, where np is a prediction time step in days;

Further, the obtaining module is specifically configured to obtain trend prediction data corresponding to the day-level service time sequence data according to a formula Tpred _ trend ═ Tpred _ smooth + Tpred _ station + u, where u is a mean value in statistic estimation performed on a residual sequence of a stationary volatility component;

Further, the obtaining module is specifically configured to perform calculation according to a formula tpred _ i ═ trand _ i + t _ session, where tpred _ i is a traffic volume estimation value in minutes, trand _ i is a minute-level average value of day-level trend prediction, and t _ session is day-mode traffic data.

The invention provides a method and a system for determining service scene time sequence data based on multi-time scale fusion, which utilize historical service time sequence data and related data, decompose data of two different time scales of a minute scale and a day scale by utilizing the historical service time sequence data and the related data, respectively adopt different methods to predict time sequences according to the characteristics of the decomposed time sequence data, and finally realize a fine service volume prediction task aiming at a long period of the service time sequence data and a short time scale of the minute scale by fusing the time sequence data of the two time scales, thereby realizing more stable long period trend fluctuation prediction by taking the day as a time unit, and also realizing fine-grained service volume prediction in a future time period by utilizing an extracted 24-hour time sequence mode of the minute scale, and because the method adopts a signal decomposition step, the whole calculation process is simpler and more stable, a complex parameter adjusting process and the calculation of a large amount of time sequence data containing a plurality of long service periods are not needed, therefore, the determining efficiency of the service scene time sequence data is improved.

Drawings

Fig. 1 is a flowchart of a method for determining service scene time series data based on multi-time scale fusion according to the present invention;

fig. 2 is a schematic structural diagram of a service scene time series data determination system based on multi-time scale fusion, provided by the invention.

Detailed Description

The technical solution of the present invention is further described in detail by the accompanying drawings and embodiments.

As shown in fig. 1, a method for determining service scene time series data based on multi-time scale fusion provided by the embodiment of the present invention includes the following steps:

101. and performing trend and local periodic two-component mode decomposition on the minute-level time sequence data after exception processing to obtain a trend partial sequence and a local periodic mode partial sequence.

For the embodiment of the present invention, step 101 may specifically include: for the time sequence T, a high-frequency oscillation part in the data can be filtered by using methods such as high-frequency filtering, moving average, gaussian smoothing and the like, a trend decomposition part T _ trend0 is obtained, and the remaining part T1 ═ T-T _ trend0 is obtained; for sequence T1, the truncated timestamp interval is 0: n complete subsequences of time period 00 to 23:59, i 1 — i 1.. n, length f, the basic intra-day period pattern is obtained by numerical averaging of the n subsequences: t _ session _ j ═ sum _ i (T1_ ij)/n, j ═ 1,. f, and is expanded into a sequence of equal length T _ session 0 ═ expanded (T _ session) according to the corresponding intra-day timestamp, and the remainder T2 is acquired as T1-T _ session 0; setting a large-scale Day data cycle frequency f2 for a sequence T2, for example, f2 equals 7Day, correspondingly calculating a complete sampling cycle of T2 to m, obtaining m local average Day cycle patterns by calculating a sequence average value T _ locseason _ i (i equals 1.. m, length is f) in the cycle, filling the remaining incomplete cycle values with a local cycle value T _ locseason _ i of a timestamp corresponding to the latest time interval, obtaining local cycle decomposition parts T _ locseason with equal lengths, and updating and calculating T _ season equal to T _ season0+ T _ locseason, and T _ term equals T-T _ season; a preliminary two-mode decomposition of timing into T ═ T _ trend + T _ search is achieved.

For the embodiment of the present invention, the exception handling process may include: acquiring time sequence data of minute-level sampling frequency; and when the minute-level time sequence data has partial sampling point data loss or obvious abnormal points, correcting the abnormal data by using the average value of the minute-level time sequence data. For the embodiment of the invention, before data decomposition, the accuracy of data prediction can be further improved by correcting the abnormal data.

The process of correcting the abnormal data may specifically include: calculating a mean value u and a standard deviation std of the sequence according to the read time sequence data T; calculate normalized value of sequence: t ' ═ abs (T-u)/std, and the maximum value and mean maxT ' ═ max (T '), meanT ' ═ mean (T '); if maxT ' > means ' 10, idx ═ (T ' > maxT/10), T [ idx ] ═ mean (T), and the removed abnormal point timestamp information time1[ idx ] is recorded.

102. And performing data integration taking days as sampling units on the trend partial sequence, and performing secondary decomposition on the integrated day-level time sequence data through a preset decomposition algorithm to obtain a smoothness trend component and a smooth fluctuation component corresponding to the day-level time sequence data.

For the embodiment of the invention, the step of carrying out secondary decomposition on the integrated day-level time sequence data by a preset decomposition algorithm comprises the following steps: and carrying out secondary decomposition on the integrated day-level time sequence data through an Empirical Mode Decomposition (EMD) algorithm. Specifically, data reintegration is performed on T _ trend in a day sampling period, time sequence data in one day are summed and combined, time sequence data T _ trend 'with a new frequency of day is obtained, and initial T _ state is set to be T _ trend', and T _ smooth is set to be 0; the following iterative calculations are performed: performing ADF verification on whether the time sequence is a steady time sequence or not aiming at the T _ station, returning to the T _ smooth and the T _ station if the time sequence is the steady time sequence, and stopping iteration, otherwise, performing the next step; for T _ station, EMD is used to perform pattern decomposition T _ sum _ k (tk), and the lowest frequency component update T _ smooth + T0 and T _ station-T0 are obtained, and the process returns to the previous step.

103. And respectively carrying out time sequence prediction on the smoothness trend component and the smooth fluctuation component to obtain a smoothness trend component estimation sequence, a smooth fluctuation component estimation sequence and a residual statistic estimation value of the smooth fluctuation component.

For the embodiment of the present invention, step 103 may specifically include: predicting the length np of the smoothness trend component T _ smooth by an ARIMA algorithm to generate a smoothness trend component estimation sequence Tpred _ smooth, wherein np is a prediction time step in days; predicting the length np of the stationary volatility component T _ station by using a Singular Spectrum Analysis (SSA) algorithm to generate a stationary volatility component estimation sequence Tpred _ station and a data fitting part Tsit _ station; and carrying out statistical calculation aiming at an estimation error to obtain a residual statistic estimation value of the stationary fluctuation component, wherein the estimation error is calculated according to a formula Terr _ station-T _ station-Tf _ station.

Further, the above component prediction results are merged into an overall trend prediction sequence Tpred _ trend ═ Tpred _ smooth + Tpred _ station + u, and corresponding upper and lower bound error limits are given: tpred _ trend _ upper ═ Tpred _ trend + std, and Tpred _ trend _ low ═ Tpred _ trend-std.

104. And acquiring trend prediction data corresponding to the day-level service time sequence data according to the smoothness trend component estimation sequence, the stationary wave component estimation sequence and the residual error statistic estimation value of the stationary wave component.

For the embodiment of the present invention, step 104 may specifically include: acquiring trend prediction data corresponding to the day-level service time sequence data according to a formula Tpred _ trend ═ Tpred _ smooth + Tpred _ station + u, wherein u is an average value in statistic estimation aiming at a residual sequence of a stationary fluctuation component; and calculating upper and lower bound error limits corresponding to the antenna-level service time sequence data according to formulas Tpred _ trend _ upper and Tpred _ trend + std and Tpred _ trend _ low as Tpred _ trend-std, wherein std is a standard deviation in statistic estimation aiming at a residual sequence of the stationary volatility component.

105. And acquiring an average value of data sampling frequency in the day according to the day-level trend prediction data, and merging the average value with the partial periodic mode sequence of the minute-level time sequence data one by one to obtain a minute-level long-time traffic estimation value.

For the embodiment of the present invention, step 105 may specifically include: and calculating according to a formula tpred _ i + t _ search, wherein tpred _ i is a traffic volume estimation value in minutes, trend _ i is a minute-level average value of day-level trend prediction, and t _ search is day-mode traffic data.

Further, the prediction result Tpred { Tpred _ i, Tpred _ upper _ i, Tpred _ low _ i, i ═ 1.. np } and the corresponding abnormal point sequence time1, time2 may be displayed by driving of an intermediate visualization result map, such as time series data decomposition, estimation error normality estimation qq map, and the like, for various subsequent analysis operations.

The method for determining the service scene time sequence data based on the multi-time scale fusion, provided by the embodiment of the invention, utilizes historical service time sequence data and related data, decomposes the data of a minute scale and a day scale in two different time scales, respectively adopts different methods to predict the time sequence according to the characteristics of the decomposed time sequence data, and finally realizes a fine service volume prediction task aiming at a long period and a minute scale short time scale of the service time sequence data by fusing the time sequence data of the two time scales, thereby realizing more stable long period trend fluctuation prediction by taking the day as a time unit, and also realizing fine-grained service volume prediction in a future time period by utilizing an extracted minute scale 24 hour time sequence mode, and as the method adopts a signal decomposition step, the whole calculation process is more concise and stable, a complex parameter adjusting process and calculation of a large amount of time sequence data containing a plurality of long service periods are not needed, therefore, the determining efficiency of the service scene time sequence data is improved.

In order to implement the method provided by the embodiment of the present invention, an embodiment of the present invention provides a service scene time series data determining system based on multi-time scale fusion, as shown in fig. 2, the system includes: decomposition module 21, prediction module 22, and acquisition module 23.

And the decomposition module 21 is configured to perform trend and local periodic two-component mode decomposition on the minute-level time series data after exception handling to obtain a trend partial sequence and a local periodic mode partial sequence.

The decomposition module 21 is further configured to perform data integration on the trend partial sequence by taking days as sampling units, and perform secondary decomposition on the integrated day-level time sequence data through a preset decomposition algorithm to obtain a smoothness trend component and a smooth volatility component corresponding to the day-level time sequence data.

And the prediction module 22 is configured to perform time sequence prediction on the smoothness trend component and the stationary volatility component respectively to obtain a smoothness trend component estimation sequence, a stationary volatility component estimation sequence, and a residual statistic estimation value of the stationary volatility component.

And the obtaining module 23 is configured to obtain trend prediction data corresponding to the day-level service time sequence data according to the smoothness trend component estimation sequence, the stationary volatility component estimation sequence, and the residual statistic estimation value of the stationary volatility component.

The obtaining module 23 is further configured to obtain an average value of data sampling frequencies in a day according to the day-level trend prediction data, and merge the average value with the minute-level time sequence data local period mode partial sequence with time sequence data of one period by one period to obtain a minute-level long-time traffic estimation value.

Further, the system further comprises: a correction module 24;

the correction module 24 is configured to obtain time sequence data of a minute-level sampling frequency; and when the minute-level time sequence data has partial sampling point data loss or obvious abnormal points, correcting the abnormal data by using the average value of the minute-level time sequence data.

Further, the decomposition module 21 is specifically configured to perform secondary decomposition on the integrated day-level time sequence data through an empirical mode decomposition algorithm EMD. .

Further, the prediction module 22 is specifically configured to perform, by using an ARIMA algorithm, prediction on the smoothness trend component T _ smooth with a length np, and generate a smoothness trend component estimation sequence Tpred _ smooth, where np is a prediction time step in units of days; predicting the length np of the stationary volatility component T _ station by using a Singular Spectrum Analysis (SSA) algorithm to generate a stationary volatility component estimation sequence Tpred _ station and a data fitting part Tsit _ station; and carrying out statistical calculation aiming at an estimation error to obtain a residual statistic estimation value of the stationary fluctuation component, wherein the estimation error is calculated according to a formula Terr _ station-T _ station-Tf _ station.

Further, the obtaining module 23 is specifically configured to obtain trend prediction data corresponding to the day-level service time series data according to a formula Tpred _ tend + Tpred _ station + u, where u is a mean value in statistic estimation performed on a residual sequence of a stationary volatility component; and calculating upper and lower bound error limits corresponding to the antenna-level service time sequence data according to formulas Tpred _ trend _ upper and Tpred _ trend + std and Tpred _ trend _ low as Tpred _ trend-std, wherein std is a standard deviation in statistic estimation aiming at a residual sequence of the stationary volatility component.

Further, the obtaining module 23 is specifically configured to perform calculation according to a formula tpred _ i ═ trand _ i + t _ session, where tpred _ i is a traffic volume estimation value in minutes, trand _ i is a minute-level average value of day-level trend prediction, and t _ session is day mode traffic data.

The service scene time sequence data determining system based on multi-time scale fusion provided by the embodiment of the invention utilizes historical service time sequence data and related data, decomposes data of two different time scales of a minute scale and a day scale by utilizing the historical service time sequence data and the related data, respectively adopts different methods to predict time sequences according to the characteristics of the decomposed time sequence data, and finally realizes a fine service volume prediction task aiming at a long period of the service time sequence data and a short time scale of the minute scale by fusing the time sequence data of the two time scales, thereby realizing more stable long period trend fluctuation prediction by taking the day as a time unit, and also realizing fine-grained service volume prediction in a future time period by utilizing an extracted 24-hour time sequence mode of the minute scale, and because the system adopts a signal decomposition step, the whole calculation process is more concise and stable, a complex parameter adjusting process and calculation of a large amount of time sequence data containing a plurality of long service periods are not needed, therefore, the determining efficiency of the service scene time sequence data is improved.

It should be understood that the specific order or hierarchy of steps in the processes disclosed is an example of exemplary approaches. Based upon design preferences, it is understood that the specific order or hierarchy of steps in the processes may be rearranged without departing from the scope of the present disclosure. The accompanying method claims present elements of the various steps in a sample order, and are not intended to be limited to the specific order or hierarchy presented.

In the foregoing detailed description, various features are grouped together in a single embodiment for the purpose of streamlining the disclosure. This method of disclosure is not to be interpreted as reflecting an intention that the claimed embodiments of the subject matter require more features than are expressly recited in each claim. Rather, as the following claims reflect, invention lies in less than all features of a single disclosed embodiment. Thus, the following claims are hereby expressly incorporated into the detailed description, with each claim standing on its own as a separate preferred embodiment of the invention.

The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. To those skilled in the art; various modifications to these embodiments will be readily apparent, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the disclosure. Thus, the present disclosure is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

What has been described above includes examples of one or more embodiments. It is, of course, not possible to describe every conceivable combination of components or methodologies for purposes of describing the aforementioned embodiments, but one of ordinary skill in the art may recognize that many further combinations and permutations of various embodiments are possible. Accordingly, the embodiments described herein are intended to embrace all such alterations, modifications and variations that fall within the scope of the appended claims. Furthermore, to the extent that the term "includes" is used in either the detailed description or the claims, such term is intended to be inclusive in a manner similar to the term "comprising" as "comprising" is interpreted when employed as a transitional word in a claim. Furthermore, any use of the term "or" in the specification of the claims is intended to mean a "non-exclusive or".

Those of skill in the art will further appreciate that the various illustrative logical blocks, units, and steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both. To clearly illustrate the interchangeability of hardware and software, various illustrative components, elements, and steps have been described above generally in terms of their functionality. Whether such functionality is implemented as hardware or software depends upon the particular application and design requirements of the overall system. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present embodiments.

The various illustrative logical blocks, or elements, described in connection with the embodiments disclosed herein may be implemented or performed with a general purpose processor, a digital signal processor, an Application Specific Integrated Circuit (ASIC), a field programmable gate array or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein. A general-purpose processor may be a microprocessor, but in the alternative, the processor may be any conventional processor, controller, microcontroller, or state machine. A processor may also be implemented as a combination of computing devices, e.g., a digital signal processor and a microprocessor, a plurality of microprocessors, one or more microprocessors in conjunction with a digital signal processor core, or any other similar configuration.

The steps of a method or algorithm described in connection with the embodiments disclosed herein may be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. A software module may be stored in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art. For example, a storage medium may be coupled to the processor such the processor can read information from, and write information to, the storage medium. In the alternative, the storage medium may be integral to the processor. The processor and the storage medium may reside in an ASIC, which may be located in a user terminal. In the alternative, the processor and the storage medium may reside in different components in a user terminal.

In one or more exemplary designs, the functions described above in connection with the embodiments of the invention may be implemented in hardware, software, firmware, or any combination of the three. If implemented in software, the functions may be stored on or transmitted over as one or more instructions or code on a computer-readable medium. Computer-readable media includes both computer storage media and communication media that facilitate transfer of a computer program from one place to another. Storage media may be any available media that can be accessed by a general purpose or special purpose computer. For example, such computer-readable media can include, but is not limited to, RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to carry or store program code in the form of instructions or data structures and which can be read by a general-purpose or special-purpose computer, or a general-purpose or special-purpose processor. Additionally, any connection is properly termed a computer-readable medium, and, thus, is included if the software is transmitted from a website, server, or other remote source via a coaxial cable, fiber optic cable, twisted pair, Digital Subscriber Line (DSL), or wirelessly, e.g., infrared, radio, and microwave. Such discs (disk) and disks (disc) include compact disks, laser disks, optical disks, DVDs, floppy disks and blu-ray disks where disks usually reproduce data magnetically, while disks usually reproduce data optically with lasers. Combinations of the above may also be included in the computer-readable medium.

The above-mentioned embodiments are intended to illustrate the objects, technical solutions and advantages of the present invention in further detail, and it should be understood that the above-mentioned embodiments are merely exemplary embodiments of the present invention, and are not intended to limit the scope of the present invention, and any modifications, equivalent substitutions, improvements and the like made within the spirit and principle of the present invention should be included in the scope of the present invention. .

14页详细技术资料下载

上一篇：一种医用注射器针头装配设备

下一篇：基于雨洪时空过程相似性挖掘的洪水动态预估方法及系统

Service scene time sequence data determination method and system based on multi-time scale fusion

相关技术

网友询问留言