CN113128693A

CN113128693A - Information processing method, device, equipment and storage medium

Info

Publication number: CN113128693A
Application number: CN201911408475.3A
Authority: CN
Inventors: 高爱丽; 吕万; 刘阳; 秦文丽; 杨晓青
Original assignee: China Mobile Communications Group Co Ltd; China Mobile Group Beijing Co Ltd
Current assignee: China Mobile Communications Group Co Ltd; China Mobile Group Beijing Co Ltd
Priority date: 2019-12-31
Filing date: 2019-12-31
Publication date: 2021-07-16

Abstract

The embodiment of the invention discloses an information processing method, an information processing device, information processing equipment and a storage medium, wherein the method comprises the following steps: acquiring a time sequence of a target network element, wherein the time sequence comprises characteristic data of a service index of the target network element at a plurality of times; inputting the time sequence into a network element prediction model to obtain a predicted value and an alarm threshold value of a service index; the network element prediction model comprises a monthly cycle fitting condition corresponding to the time sequence; and monitoring the target network element according to the actual value, the predicted value and the alarm threshold value of the target network element. The method and the device solve the problems that misjudgment and missed judgment of the running state of the network element are caused because the fixed threshold early warning cannot deal with the actual situation that the site is complicated and changeable.

Description

Information processing method, device, equipment and storage medium

Technical Field

The embodiment of the invention relates to the technical field of data processing, in particular to an information processing method and device, electronic equipment and a storage medium.

Background

Machine Learning (ML) is a one-domain-multiple-domain interdisciplinary, and it is studied how a computer simulates or implements human learning behavior. With the advent of the big data era, machine learning, particularly deep learning suitable for large-scale data, is gaining more and more attention and application, for example, monitoring and early warning the operating state of each network element according to massive business indexes.

At present, when monitoring and early warning are carried out on the running state of a network element, fixed threshold early warning is adopted, and abnormal changes of service indexes caused by factors such as important holidays, manual parameter modification, cell faults and the like can be identified. However, the single fixed threshold early warning cannot cope with the complex and variable actual situation of the site, and cannot really and effectively control the abnormal event situation, so that misjudgment and missed judgment often occur.

Disclosure of Invention

Embodiments of the present invention provide an information processing method, an information processing apparatus, an electronic device, and a storage medium, so as to solve the problem in the related art that misjudgment and missed judgment occur in the operation state of a network element due to the fact that a fixed threshold early warning cannot cope with a complex and changeable actual situation on a site.

In order to solve the technical problem, the invention is realized as follows:

in a first aspect, an embodiment of the present invention provides an information processing method, where the method may include:

acquiring a time sequence of a target network element, wherein the time sequence comprises characteristic data of a service index of the target network element at a plurality of times;

inputting the time sequence into a network element prediction model to obtain a predicted value and an alarm threshold value of a service index; the network element prediction model comprises a monthly cycle fitting condition corresponding to the time sequence;

and monitoring the target network element according to the actual value, the predicted value and the alarm threshold value of the target network element.

In a possible embodiment, the mentioned "feature data" is at least one of the following data:

trend data, periodic data, random data; wherein the content of the first and second substances,

the trend data is used for representing the variation trend of the time series within a first preset time; the periodic data is used for representing the variation trend of the season or periodicity of the time series; the random data is used for representing the variation trend of various factors of the time sequence;

and the periodic data correspond to the monthly cycle fitting condition in the network element prediction model.

Further, the trend data, the periodic data, or the random data includes at least one of the following: total XDR number, quality ratio and scene perception index data of instant messaging.

In another possible embodiment, the "network element prediction model" referred to above comprises a growth trend model and/or a seasonal trend model; wherein the content of the first and second substances,

the growth trend model is used for determining the change trend of the time series in a second preset time and/or estimating the transformation trend of the time series in a third preset time;

the seasonal trend model is used for processing the time sequence according to the month cycle fitting condition, the day cycle fitting condition and the year cycle fitting condition in the seasonal trend model to obtain preset values corresponding to the month cycle fitting condition, the day cycle fitting condition or the year cycle fitting condition.

Further, the characteristic value of the monthly cycle characteristics in the monthly cycle fitting condition is 30.5; and the monthly cycle characteristic value is used for fitting the time sequence.

In yet another possible embodiment, before the step of "obtaining the time series of the target network element" mentioned above, the method may further include:

determining abnormal data in the initial time sequence according to the initial time sequence;

and determining the abnormal data sequence not included as a time sequence.

In another possible embodiment, the step of "determining abnormal data in the initial time sequence according to the initial time sequence" may specifically include:

calculating the standard deviation of the initial time sequence according to the initial time sequence;

and determining the data with the standard deviation meeting the preset standard deviation condition as abnormal data.

In another possible embodiment, the step of "inputting the time sequence into the network element prediction model to obtain the predicted value of the service indicator and the alarm threshold value" may specifically include:

inputting the time sequence into a network element prediction model, and determining a target time point in the time sequence;

calculating an average value of the target time point, a previous time point related to the target time point and a next time;

smoothing the time sequence according to the average value to obtain a corrected time sequence;

and fitting the corrected time sequence to obtain a predicted value and an alarm threshold value of the service index.

In another possible embodiment, the step of monitoring the target network element according to the actual value, the predicted value, and the alarm threshold value of the target network element may specifically include:

calculating a target difference value between the actual value and the predicted value;

comparing the target difference value with an alarm threshold value to obtain a comparison result;

and when the comparison result shows that the target difference value does not meet the alarm threshold value, generating prompt information, wherein the prompt information is used for representing the abnormality of the target network element.

Or when the comparison result shows that the target difference value meets the alarm threshold value, generating information for representing the normal operation of the target network element.

In a second aspect, an embodiment of the present invention provides an information processing apparatus, which may include:

the acquisition module is used for acquiring a time sequence of the target network element, wherein the time sequence comprises characteristic data of a service index of the target network element at a plurality of times;

the processing module is used for inputting the time sequence into the network element prediction model to obtain a predicted value and an alarm threshold value of the service index; the network element prediction model comprises a monthly cycle fitting condition corresponding to the time sequence;

and the monitoring module is used for monitoring the target network element according to the actual value, the predicted value and the alarm threshold value of the target network element.

In a third aspect, an embodiment of the present invention provides an electronic device, which includes a processor, a memory, and a computer program stored in the memory and executable on the processor, and when executed by the processor, the computer program implements the information processing method according to the first aspect.

In a fourth aspect, there is provided a computer-readable storage medium having stored thereon a computer program for causing a computer to execute the information processing method according to the first aspect if the computer program is executed in the computer.

In the embodiment of the invention, the time sequence is input into the network element prediction model comprising the monthly cycle fitting condition corresponding to the time sequence so as to obtain the predicted value and the alarm threshold value of the service index. Therefore, the dynamic and period fitting condition can be adopted and is changed along with the change of the month, so that the time sequence with the characteristics of long-term trend, multiple periods and the like can be processed, and the method has the advantages of high efficiency and good robustness. And in addition, monitoring the target network element according to the actual value of the target network element, and the predicted value and the alarm threshold value determined by the network element prediction model. Therefore, the alarm threshold value can be dynamically adjusted according to the network element prediction model, and therefore the method provided by the embodiment of the invention can be used for dealing with the complex and changeable actual situation on site and really and effectively mastering the abnormal event situation, so that the misjudgment and the judgment missing situation caused by using a fixed threshold are avoided.

Drawings

The present invention will be better understood from the following description of specific embodiments thereof taken in conjunction with the accompanying drawings, in which like or similar reference characters designate like or similar features.

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

FIG. 2 is a diagram illustrating a functional variation of trend data according to an embodiment of the present invention;

FIG. 3 is a diagram illustrating a function variation corresponding to periodic data according to an embodiment of the present invention;

FIG. 4 is a diagram illustrating a function variation corresponding to random data according to an embodiment of the present invention;

FIG. 5 is a diagram illustrating a normal distribution density function according to an embodiment of the present invention;

fig. 6 is a schematic diagram of a change of a downlink packet loss rate of VOLTE in an overall 3.1-5.10 period of a city-divided company according to an embodiment of the present invention;

fig. 7 is a schematic diagram of variation of the XDR ratio in dimension of a city-divided company according to an embodiment of the present invention;

fig. 8 is a schematic diagram illustrating a variation trend of a predicted value within a prediction time according to an embodiment of the present invention;

FIG. 9 is a diagram illustrating a fitting effect provided by an embodiment of the present invention;

FIG. 10 is a flowchart of a method for implementing information processing according to an embodiment of the present invention;

FIG. 11 is a diagram of a time series data list according to an embodiment of the present invention;

FIG. 12 is a schematic diagram of an embodiment of the present invention showing a comparison result;

FIG. 13 is a diagram illustrating a modified time series fit according to an embodiment of the present invention;

fig. 14 is a schematic diagram of a prompt message visualization provided in the embodiment of the present invention;

fig. 15 is a schematic diagram of an early warning display of the total number of XDRs in hour level of instant messaging of a delay branch company according to an embodiment of the present invention;

fig. 16 is a schematic diagram of warning 5.1 abnormal day-level instant messaging traffic (total XDR number) of the urban trisection company according to the embodiment of the present invention;

fig. 17 is a schematic diagram of a city trisection company 5.1 early warning of an abnormal day-level instant messaging quality ratio;

fig. 18 is a schematic diagram of an early warning of 5.1 abnormal day-level instant messaging traffic (total XDR number) of a delay branch company according to an embodiment of the present invention;

fig. 19 is a schematic diagram of a 5.1 abnormal day-level instant messaging quality ratio early warning of a delay branch company according to an embodiment of the present invention;

fig. 20 is a schematic diagram of an early warning of an instant messenger hour-level instant messenger traffic (total XDR number) of a delay branch company according to an embodiment of the present invention;

fig. 21 is a schematic diagram of an early warning of a delay branch company, i.e., an instant messaging small-level instant messaging quality difference ratio warning according to an embodiment of the present invention;

fig. 22 is a schematic diagram of a kyanite WeChat class II sub-scenario, namely, an early warning of real-time communication total daily traffic (GB), according to an embodiment of the present invention;

FIG. 23 is a schematic diagram of a cell distribution covering a large state trade hotel banquet hall according to an embodiment of the present invention;

fig. 24 is a schematic diagram of a total XDR number warning based on the 21863681 click picture of the third-phase country trade hotel banquet hall HLM-1 in fig. 23 according to an embodiment of the present invention;

fig. 25 is a schematic diagram of warning of the total XDR number based on the clicked picture in fig. 24 according to the embodiment of the present invention;

fig. 26 is a schematic diagram of warning of total XDR numbers of click pictures based on the 21863691 korean national trade third-phase hotel banquet hall HLM-11 in fig. 23 according to an embodiment of the present invention;

fig. 27 is a schematic diagram of an early warning based on the ratio of the quality differences of the click pictures in fig. 26 according to an embodiment of the present invention;

fig. 28 is a schematic diagram of warning of total XDR number based on 21863701 click pictures of the third-phase country trade hotel party hall HLM-21 in fig. 23 according to an embodiment of the present invention;

fig. 29 is a schematic diagram of an early warning based on the ratio of the shot picture quality in fig. 28 according to an embodiment of the present invention;

FIG. 30 is a diagram illustrating a comparison of actual data provided by an embodiment of the present invention;

fig. 31 is a schematic diagram of an actual value of a total XDR number of an instant messaging "× circle click picture" in a whole time period from 5.1 to 7.21 according to an embodiment of the present invention;

FIG. 32 is a diagram illustrating actual value changes within a time period of the quality difference ratio indicator from 5.1 to 7.21 according to an embodiment of the present invention;

fig. 33 is a schematic diagram of a cell coverage area based on table 2 according to an embodiment of the present invention;

fig. 34 is a schematic diagram of warning total XDR numbers of circle-click pictures of ZL1 friends of the department of electronics industry of seafood in accordance with an embodiment of the present invention;

fig. 35 is a schematic diagram of warning total XDR numbers of circle-click pictures of ZL4 friends of the department of electronics industry of seafood in accordance with an embodiment of the present invention;

fig. 36 is a schematic diagram of total XDR number warning of click pictures of friend circles in a hai lake kaider crystal shopping center 1ZLM-203 according to an embodiment of the present invention;

fig. 37 is a schematic diagram of total XDR number warning of click pictures of friend circles in a hai lake kaider crystal shopping center 1ZLM-211 according to an embodiment of the present invention;

FIG. 38 is a diagram illustrating a variation of total xdr number of clicked pictures according to an embodiment of the present invention;

fig. 39 is a schematic diagram illustrating a variation of the cell quality xdr ratio according to an embodiment of the present invention;

fig. 40 is a device diagram of an information processing device according to an embodiment of the present invention;

fig. 41 is a schematic diagram of a hardware structure of an electronic device according to an embodiment of the present invention.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

Currently, conventional approaches to index alarms may include fixed threshold analysis and equal-or ring-ratio analysis. Wherein, the fixed threshold is a fixed threshold specified by human experience or examination requirements. For example, it is considered that the wireless side quality difference XDR ratio for setting the instant messaging ". x.circle click on picture" service is not higher than 5%, and the like. Another common way is a comparison or a comparison-and-loop analysis, for example, manufacturers monitor the indexes before and after upgrading by comparison, and generally monitor the indexes by comparison in the same hour before and after upgrading, i.e., by a comparison method, if the indexes are obviously degraded, the indexes are considered to be abnormally changed due to upgrading and need to be checked in time; the ring ratio method is mostly applied to the prediction of the flow and the number of users in the high-load cell.

However, the fixed threshold method is inherently static and post hoc. On one hand, the change trend of the index cannot be reflected; on the other hand, even if the threshold is set differently in different scenes, the threshold is still a knife in nature, and the embodiment of time dimension is lacked. Regardless of the same ratio or the ring ratio, only the long-term trend of the sequence variation can be explained, and the period trend is almost zero. In the time series fitting or prediction of long-term trend and coexistence of multiple periods (day/week/month/year), the traditional analysis method has great disadvantages and cannot be used for index dynamic monitoring and early warning. The most common time series analysis model is an autoregressive integrated moving average (ARIMA) model, which has the greatest disadvantage that it requires time series data to be stable or stable after differentiation, and extracts information with a fixed period during differential operation, and requires a complex data preprocessing process during application, which is difficult to meet the requirements of real data and analysis.

The embodiment of the invention provides an information processing method, an information processing device, an electronic device and a storage medium, aiming at solving the problems that in the related art, the operation state of a network element is misjudged and misjudged due to the fact that fixed threshold early warning cannot deal with the actual situation that a field is complex and changeable.

The information processing method provided by the embodiment of the invention is explained in detail below.

Fig. 1 is a flowchart of an information processing method according to an embodiment of the present invention.

As shown in fig. 1, the information processing method may specifically include steps 110 to 130, which are specifically as follows:

step 110, acquiring a time sequence of a target network element, wherein the time sequence comprises characteristic data of a service index of the target network element at a plurality of times;

step 120, inputting the time sequence into a network element prediction model to obtain a predicted value of the service index and an alarm threshold value;

and step 130, monitoring the target network element according to the actual value, the predicted value and the alarm threshold value of the target network element.

Therefore, the time sequence can be input into the network element prediction model comprising the monthly cycle fitting condition corresponding to the time sequence, so as to obtain the predicted value and the alarm threshold value of the service index. Therefore, the dynamic and period fitting condition can be adopted and is changed along with the change of the month, so that the time sequence with the characteristics of long-term trend, multiple periods and the like can be processed, and the method has the advantages of high efficiency and good robustness. And in addition, monitoring the target network element according to the actual value of the target network element, and the predicted value and the alarm threshold value determined by the network element prediction model. Therefore, the alarm threshold value can be dynamically adjusted according to the network element prediction model, and therefore the method provided by the embodiment of the invention can be used for dealing with the complex and changeable actual situation on site and really and effectively mastering the abnormal event situation, so that the misjudgment and the judgment missing situation caused by using a fixed threshold are avoided.

Based on the above steps 110 to 130, each step is explained in detail as follows:

it is referred to a step 110 of,

the characteristic data in the embodiment of the present invention may be at least one type of data among the following types of data: trend data, periodic data, random data.

The trend data is used for representing the variation trend of the time series within a first preset time; the periodic data is used for representing the variation trend of the season or periodicity of the time series; the random data is used for representing the variation trend of various factors of the time sequence; and the periodic data correspond to the monthly cycle fitting condition in the network element prediction model.

For example, the following steps are carried out: the time sequence in the embodiment of the invention can be a sequence formed by arranging the numerical values of the same statistical index according to the occurrence time sequence. The main purpose of time series analysis is to predict the future based on existing historical data. A common analysis method in Time Series analysis theory is Time Series Decomposition (Decomposition of Time Series), the core of which is an additive regression model, i.e. the Time Series f (x) can be expressed as a function of three factors, f (x) ═ t (x) + s (x) + r (x).

The variation trend of the trend data t (x) is shown in fig. 2, the variation trend of the periodic data s (x) is shown in fig. 3, and the variation trend of the random data r (x) is shown in fig. 4. Besides the form of addition, there are also forms of multiplication, namely: f (x) x (t), (x) x s (x) x r (x).

In addition, f (x) ═ t (x) + s (x) + r (x) may be equivalent to lnf (x) ═ lnt (x) + lns (x) + lnr (x), i.e., in the case of the prediction model, the form of multiplication can be obtained by taking the logarithm first and then performing time-series decomposition. The addition model is more prone to make up each factor of the model independent, and the multiplication model is more prone to influence among each factor, especially for applications where periodic variations are obvious. The embodiment of the invention adopts an addition model in the process of calculating the network element prediction model.

In a possible embodiment, before the step of acquiring the time series, the method may further include: determining abnormal data in the initial time sequence according to the initial time sequence; and determining the abnormal data sequence not included as a time sequence.

The embodiment of the present invention provides a method for determining abnormal data, which may specifically refer to the following implementation manners:

It should be noted that, based on the above example, data of past n cycles (n is a positive integer greater than or equal to 1) of a single index can be decomposed into three functions by time-series decomposition, where the trend and the cycle are determined (stable) by the system, and random terms are of particular interest. Normally, the range of the random term should be within a reasonable range, and if the range is beyond the reasonable range, the random term indicates an abnormality.

As shown in fig. 5, it is assumed here that the abnormality follows a normal distribution (this assumption is relatively easy to verify when the factors are large and independent of each other), and is derived from a normal distribution density function. Referring to fig. 5, the distribution within 1 σ accounts for 68.3% of the total and within 3 σ accounts for 99.7% of the whole, and in practice, the distribution outside 3 σ is concerned (usually 3.5 σ to 4.5 σ in practical application), which should not occur theoretically, or should not occur according to the change rule of the historical time extending to the prediction time, and the point beyond-3 σ (4 σ)/+3 σ (4 σ) is a point in the sequence that cannot be explained by the trend data t (x) and the periodic data s (x), and may be caused by other burst factors, and is determined as abnormal data.

It is referred to a step 120 of,

the network element prediction model comprises a monthly cycle fitting condition corresponding to the time sequence. Further, the characteristic value of the monthly cycle characteristics in the monthly cycle fitting condition is 30.5; and the monthly cycle characteristic value is used for fitting the time sequence.

Here, the network element prediction model in the embodiment of the present invention includes a growth trend model and/or a seasonal trend model; wherein the content of the first and second substances,

the seasonal trend model is used for processing the time sequence according to the month cycle fitting condition, the day cycle fitting condition and the year cycle fitting condition in the seasonal trend model to obtain preset values corresponding to the month cycle fitting condition, the day cycle fitting condition or the year cycle fitting condition. In addition, in a possible embodiment, the above-mentioned network element prediction model may further include a holiday model.

For example, the network element prediction model is mainly a Prophet model, and the overall construction of the model is shown in the following formula (1):

y(t)＝g(t)+s(t)+h(t)+ε_t (1)

wherein, formula (1) is wholly composed of three parts: the influence of growth trend models (growth), seasonal trend models (seaselectivity), and holiday models (holidabys) on predicted values. Wherein g (t) represents a growth function for fitting aperiodic variations of the predicted values in the time series; s (t) is used to indicate periodic variations, such as weekly, monthly (which may be understood herein as monthly cycle fit conditions), and the seasons of the year, etc.; h (t) represents the impact of those potential holidays with non-fixed periods in the time series on the predicted values. Last epsilon_tFor the noise term, representing the fluctuation of the model not predicted, let us assume_tIs gaussian distributed.

Here, the growth trend model (growth) is the core of the entire model, which represents how the entire time series is considered to grow, and how it is expected to grow in the future time. This section provides two models: a Non-Linear growth model (Non-Linear growth) and a Linear growth (Linear growth), in the embodiment of the present invention, a Linear growth is adopted as a model for fitting a long-term trend, and a formula of the Linear model is described in the following formula (2):

g(t)＝(k+a(t^T)δ)t+((m+a(t)^T)γ) (2)

the algorithm in the seasonal trend model (seacoast) mainly relies on a fourier series to construct a flexible periodic model, and P may be set to be a regular period length of a preset time series (for example, when the time series is a unit of day, P is 365.25 for annual data and P is 7 for weekly data), or of course, P is 30.5 for a month period of the time series fourier series (different days per month, this is taken as a trade-off). Thus, an estimate of any smoothing cycle effect is obtained, as shown by equation (3):

the holiday model (holidabys) for holidays or some major events has a large impact on the time series, and there is often no periodicity at these time points. Analysis of these points is extremely necessary, even though sometimes it is of far greater importance than the usual points. The algorithm supports the consideration of the effects of different holidays at different time points as independent models.

Therefore, based on the influence of the growth trend model (growth), the seasonal trend model (seasonal) and the holiday model (holidays) on the predicted value, the following describes that the method in the embodiment of the present invention determines the predicted value of the service index more accurately, in comparison with the prior art, and the specific manner is as follows:

taking the cheng fen company as an example, the change of the daily-level VOLTE downlink packet loss rate of the cheng fen company in 3.1-5.10 periods as a whole is shown in fig. 6 as follows: and the instant messaging of the city-one company dimension ". x.circle click on picture" the quality XDR ratio change is shown in fig. 7. Based on fig. 6 and 7, the variation of the KQI index with the service attribute has a significant end-of-month effect compared with the KPI, and the end-of-month effect is shown in: (1) the end of the month is influenced by factors such as unlimited set meals and the like, and the index deterioration is most obvious; (2) the indexes show a gradual deterioration trend in the whole month and recover in the beginning of the next month; (3) the larger the dimension of the analysis object, the more obvious the end-of-month effect.

If the original Prophet model is still used to fit the sequence, the following results are obtained: the index has a month end effect, and the index tends to deteriorate in the whole 3 months, particularly the deterioration at the end of the month is obvious, but the fitted curve identifies the normal index deterioration as an abnormal value, only presents a single cycle characteristic, and the fitting accuracy of the whole curve is influenced by the partial value of the deterioration at the end of the month, so that the whole curve is deteriorated. Thus, when the index change at the end of 4 months is predicted, the index still extends in the prediction time according to the trained single-week-period trend, and the actual index at the end of 4 months still presents a month end effect with obvious month end deterioration, and easily exceeds the upper boundary to trigger false alarm as shown in fig. 8.

For time series with the end-of-month effect, or in essence with multi-cycle characteristics (i.e. day cycle, week cycle, month cycle), the original Prophet model needs to be improved, and the month cycle setting P of adding the time series fourier series is 30.5, i.e. the month cycle fitting condition (different days per month, here taking a compromise). With this added point, the improved Prophet model of the present example shows the following fitting effect as shown in fig. 9. Through analysis, index changes of all branch companies/SGW/IP dimensions have obvious month end effects, time sequence changes of partial second-class WeChat scenes (universities, scenic spots, traffic trunks and the like) have no obvious month end effects (for example, specific scene attributes can form specific user behaviors in partial scenes such as Chinese science and technology museums), the time sequence of a cell level is influenced by cell level data fluctuation, and periodic characteristics are further weakened integrally. In addition, although the monthly period setting of the time series Fourier series is increased in the embodiment of the invention, the fitting precision of the time series is not influenced.

Therefore, the dynamic and period fitting condition can be adopted and changes along with the change of the month, so that the time sequence with the characteristics of long-term trend, multiple periods and the like can be processed, and the method has the advantages of high efficiency and good robustness.

In addition, based on one possible embodiment of step 110, step 120 may specifically include:

It is referred to a step 130 of,

In summary, in the embodiment of the present invention, the time sequence is input into the network element prediction model including the monthly cycle fitting condition corresponding to the time sequence, so as to obtain the predicted value of the service indicator and the alarm threshold value. Therefore, the dynamic and period fitting condition can be adopted and is changed along with the change of the month, so that the time sequence with the characteristics of long-term trend, multiple periods and the like can be processed, and the method has the advantages of high efficiency and good robustness. And in addition, monitoring the target network element according to the actual value of the target network element, and the predicted value and the alarm threshold value determined by the network element prediction model. Therefore, the alarm threshold value can be dynamically adjusted according to the network element prediction model, and therefore the method provided by the embodiment of the invention can be used for dealing with the complex and changeable actual situation on site and really and effectively mastering the abnormal event situation, so that the misjudgment and the judgment missing situation caused by using a fixed threshold are avoided.

In order to facilitate understanding of the method provided by the embodiment of the present invention, based on the above contents, the instant messaging index (which can be extended to any end-to-end index) of the monitoring network element in each dimension of the current network is taken as an input, and an improved network element prediction model, i.e., a Prophet model, is taken as an example, so as to exemplify the information processing method provided by the embodiment of the present invention.

Fig. 10 is a flowchart of a method for implementing information processing according to an embodiment of the present invention.

As shown in fig. 10, the method may include steps 1010-1050, which are specifically as follows:

step 1010, an initial time series is obtained.

In which the raw data, i.e. the initial time series, required for the algorithmic analysis may be obtained by a certain platform.

For example: and extracting the instant messaging indexes of the network elements to be early-warned as an initial time sequence, and actually selecting the total XDR number and the quality difference ratio of the instant messaging click pictures to respectively train. In one possible embodiment, as shown in FIG. 11, the input time series data is simply required to be divided into three columns: and distinguishing the IDs of different network elements, the timestamp of each network element time sequence and a corresponding real numerical value, and meanwhile, the algorithm supports data loss to a certain degree, and the program can carry out smoothing treatment. Next, as shown in fig. 11, the length of the training sequence and the prediction period may be set in the configuration file, and currently, for the algorithm verification stage, for example, inputting the actual data of 3.26-5.5 days, verifying whether the data of 5.1-5.5 days has an abnormality, it is necessary to set the data of 3.26-4.30 days as the training data, i.e. testdays as 5, so that the actual data of 5.1-5.5 are only used as the test set for result comparison, and refer to fig. 12 specifically.

Step 1020, determining a time sequence of the target network element according to the initial time sequence, where the time sequence includes characteristic data of the service index of the target network element at multiple times.

Here, the time series of the target network element is determined in the embodiment of the present invention is essentially unsupervised learning because it is not known in advance which data are abnormal, for example, a 7% quality difference index may be normal for a long-term high-load cell, and an abnormal calculation for a cell with almost no quality difference cannot distinguish positive and negative samples according to an absolute standard. And manually distinguishing abnormal data one by one aiming at each analysis object and removing the abnormal data, wherein the efficiency of the labeling process is too low to meet the requirement of real-time monitoring, so the method provided by the embodiment of the invention does not have a step of labeling the data.

From the result of the sequence fitting, if there are outliers in the history data of the prediction threshold without any intervention, the outliers will have a bad influence on the final threshold. Thus, the embodiment of the invention provides a feasible method for calculating the standard deviation of the input time series and eliminating the data beyond the range of 3/4 standard deviations. Specifically, two-stage calculation is adopted in threshold prediction, namely, algorithm calculation is performed twice, points with obvious abnormality are removed for the first time (exceeding a set standard deviation range and being set to be 3.5 sigma at present), for the removed points, an average value (capable of being set) of front and back continuous time points (hour/day) is taken as a filling value after removal for smoothing, and fitting is performed again by using a corrected time sequence.

As shown in fig. 13, the data (points in circles) at noon of day 6 and 4 months are identified as abnormal during the first operation of the algorithm, the algorithm is performed after the abnormal data are removed, the data are filled according to the numerical average of the sliding windows before and after the abnormal point, the second operation is performed, and finally the actual curve and the boundary are obtained by fitting according to the corrected time series.

Step 1030, inputting the time sequence into the network element prediction model to obtain a predicted value of the service index and an alarm threshold value.

Here, the program calls the Prophet model according to the corrected time sequence for fitting, the principle of this step is similar to that involved in step 120, and details can refer to step 120, which is not described herein again.

Step 1040, monitoring the target network element according to the actual value, the predicted value and the alarm threshold value of the target network element.

And 1050, when the comparison result of the difference value between the actual value and the predicted value of the target network element and the alarm threshold value shows that the target difference value does not meet the alarm threshold value, generating prompt information, wherein the prompt information is used for representing the abnormality of the target network element.

Here, the embodiment of the present invention may visualize the prompt information to be displayed to the user, and specifically may invoke a Matplotlib application program to output a visualization result, as specifically shown in fig. 14. Referring to fig. 14, black dots indicate each timestamp value of the originally entered time series, supporting day-level/hour-level granularity data entry; the blue curve represents the functional relationship of algorithm fitting after abnormal point elimination according to the black input time sequence; the plus sign is also the time sequence of input, but is set as a test set, does not participate in sequence training, and is only aligned as a result.

The program will calculate and output the upper and lower boundaries of reasonable index variation at the same time. The premise of successful time series decomposition is that the predictable part of the series is the superposition of long-term trend and multi-period trend, and the unpredictable part is controllable and meets the normal distribution. The interval-3 σ (4 σ) +3 σ (4 σ) satisfying the normal distribution can be calculated as a reasonable range and added to a predictable portion to obtain upper and lower boundaries of reasonable variation of the time series.

Assume that the fitted function relationship is f (t) ═ t (t) + s (t) + r (t), where the trend data t (t) and the periodicity data s (t) represent the trend function and periodicity data, respectively, in the predictable portion of the fit, and the training sequence of the original input is y (t), i.e., σ is the standard deviation (square root of the sum of the squares of the differences between the random sequence portions and the sequence mean) of the random sequence considered to be in accordance with normal distribution after y (t) — t (t) -s (t) ± 3/4 σ.

It should be noted that, in the visualization of data, all data statistics including the fitting values, the identified abnormal values, and the calculated upper and lower boundary values of the training set and the test set time periods may be output simultaneously. For example, as shown in fig. 15, the hour-level XDR total early warning display of the instant messaging of the celebration affiliate shows, and then the visualization result presented by the program can be restored by using excel drawing according to the program output value and the real value.

In the following, an improved Prophet time series algorithm is adopted to monitor and warn the whole network, the branch companies, the SGW, the IP, the key scenes and the day and hour instant messaging perception indexes of each dimension of the cell, and examples are respectively given below in terms of capacity burst limitation, manual parameter modification and cell fault warning caused by a hotspot event, specifically as follows:

(1) capacity burst limitation due to hot spot events

In daily optimization, most of the causes of the Quality difference of Key Quality Indicators (KQI) are high load. Most reasons for short-term sudden degradation of the index can also be summarized as local capacity burst, which is shown in that the network absorbs traffic exceeding the carrying capacity of the network, and abnormal events causing the local capacity burst include holidays, important conferences, concerts, important gatherings and the like.

For example, the early warning of the abnormal WeChat flow and the index in the holiday of 5 months and 1 days is carried out by comparing the actual early warning results of the city trisection company and the delay-celebration division company, the current stage still belongs to the algorithm verification effect, the input actual values of the hour level and the day level after 5.1 are used for comparing with the early warning boundary output by the algorithm, and do not participate in the time series training and influence the curve fitting. Referring to fig. 16, 17, 18 and 19, it can be seen from these 4 graphs that the overall instant messaging traffic and the quality ratio of the urban third-generation company are normal in the period of 5.1, and the actual index of the overall instant messaging traffic and the quality ratio of the delay branch company exceeds the early warning boundary in the period of 5.1 to trigger an alarm. The phenomenon is consistent with the actual user behavior, and the opening of the world meeting and the arrival of a 5.1 holiday make the delay celebration become a business hotspot area in a 5.1 period.

Then, further, the day-level granularity is refined to the hour level (the introduction of an hour-level time sequence can increase day period characteristics on the basis of original time sequence period characteristics), and early warning is performed on hour-level instant communication flow and index change of the delay celebration branch company, as shown in fig. 20 and 21, from the view of an hour-level early warning result, the whole instant communication flow and quality difference of the delay celebration branch company in the period of 5.1 account for the first hours, an alarm can be continuously triggered, and a capacity burst is taken as a cause and generally occurs before index degradation. 5.1 vacation traffic and index anomaly monitoring can be further extended to be applied to two types of sub-scenarios of instant messaging. As shown in fig. 22, the day-level traffic is abnormal during the 5.1 vacation for the kyanite instant messaging two-class sub-scenario.

In addition, the method provided by the embodiment of the invention provides intelligent early warning in a state trade internet peak meeting, for example, a 2019-Eri (Beijing) annual peak meeting is held in a state trade big hotel all day 6 and 13 in 2019, the meeting lasts from 9:30 in the morning to 16:00 in the afternoon, and the prophet algorithm is used for verifying the cell-level capacity and the sensitivity of index early warning.

As shown in fig. 23, four red-labeled cells are used as the cells covering the banquet hall of the national trade hotel, wherein 21863731 cells are missing and cannot be fitted, and three other cells and the rest of the cells can be fitted, and the total XDR number and the quality ratio of the instant messaging click pictures with time series of 4.1-6.12 days are input. From the deepened part of fig. 23 and in combination with fig. 24, 25, 26, 27, 28 and 29, it can be seen that the warning result, i.e., the total XDR number and the quality ratio of the ring click pictures, all 3 cells 6.13 covering the banquet hall trigger an alarm at the day level, and only 23569117 of other cells covering the national trade hotel trigger an alarm, which may be caused by other reasons, conforms to the actual user behavior of the peak meeting at the day.

(2) Manually changing parameters

Daily optimization can modify a large number of parameters, part of the parameters, such as common modification power, can affect the coverage area of a cell and further affect the service absorption capacity of the cell, in view of the instant communication quality difference ratio, the increase of the service absorption capacity of the cell does not necessarily cause the quality difference ratio to be degraded, and only when the service absorption capacity of the cell exceeds the self network bearing capacity or the edge users are greatly increased, the indexes begin to be obviously degraded.

Table 1 below is a HLM-211 cell power modification record for the city of the bridge east city, foal, tong, and RS power modification for the cell of day 7 and day 5.

TABLE 1

Training is carried out by using a time sequence of total XDR numbers of 5.1-6.30 days, a predicted value and an early warning upper and lower bounds of 7.1-7.7 are output by an algorithm, actual data of 7.1-7.7 days are used for comparison, and the output of the algorithm is shown in figure 30. Referring to fig. 30, the total XDR number of the instant messaging "x circle click picture" of the cell obviously exceeds the upper early warning boundary from 7.5 days, and the algorithm identifies that the flow anomaly is consistent with the power modification time.

In the whole time period from 5.1 to 7.21, the actual value of the total XDR number of the instant messaging "× circle click picture" in the cell changes, and the instant messaging flow rate at the beginning of 7.5 days in the cell is obviously increased, which can be considered as a single factor of modifying the power, and can be shown in fig. 31. The instant messaging of the cell "click the picture" change of the actual value in the time period of the quality difference ratio index 5.1-7.21, the external factor of the quality difference ratio and the modified power is not directly related, as shown in fig. 32.

(3) Cell fault warning

Table 2 below is a room sub-cell shutdown alarm for a mall in the cheidel crystal mall in the three areas of city, lasting 6.21-7.4 days:

TABLE 2

As shown in fig. 33, the mall has a macro station with the same coverage, 71585 ZL, department of electronics industry, hai lake, mainly 1,4 cells with the same coverage. And (3) carrying out early warning verification on data after 6.21(6.21-6.25) by using an algorithm: the significant flow rise (total XDR number of click pictures, as shown in fig. 34 and 35) is observed in 1 and 4 cells of the hai lake electronics industry ZL, while the flow decay is evident in the fault cells, such as 96133_ hai lake kaider crystal shopping center 1ZLM (96133), 203 and 211 cells, and the early warning is triggered as well, as shown in fig. 36 and 37, as shown in fig. 38, the total XDR number of click pictures, 4,27-7.20, in 1 cell of the hai lake electronics industry ZL, is evident after 6.21, and as shown in fig. 39, the cell quality difference XDR is proportional to the normal fluctuation.

In conclusion, the improved Prophet algorithm outputs the predicted value of the index and the alarm threshold, and essentially the time series emphasizes that the change of the dependent variable has the memory in continuous time, the memory shows that the data characteristic (the value is large or small) analyzed at the moment influences the data characteristic at the next moment with high probability, and the memory is weaker than the hour grade between the day grades.

The abnormal value is caused by the superposition influence of another external force and exceeds the normal variation range determined only by the influence of the original time change rule, so that essentially no sign exists before the time starting point of the abnormal event, the intelligent early warning does not tell in advance when the abnormal event occurs, but when the abnormal event occurs, the abnormal event can be rapidly identified by comparing the actual value with the early warning threshold, and the logic meaning of the early warning is faster than the analysis after human and not faster than the abnormal event.

In the existing network, the abnormal event may be that the local capacity is suddenly limited due to a hot event, the quality is poor due to sudden interference, the flow and index of the local cell and the surrounding cells are abnormal due to cell failure, and the flow and index of the cells are abnormal due to artificial parameter modification. From the correlation with the abnormal events, the change of the cell traffic flow is more closely related to the abnormal events, and the quality difference index presents certain fluctuation which is more prominent in the cell level. Although the index degradation and the flow rate abnormality do not necessarily occur simultaneously, the flow rate change often indicates a cause of the index degradation. The dynamic monitoring and early warning of each network element of a key cell or scene can quickly identify important abnormal events, and timely take measures, so that the practical significance is necessary compared with the practical significance of analyzing the quality reasons of the poor cell or network element according to a fixed threshold and adopting the traditional same-ratio ring ratio monitoring index.

Therefore, based on the above method, the embodiment of the present invention further provides an information processing apparatus, which is specifically described with reference to fig. 40.

Fig. 40 is a device diagram of an information processing device according to an embodiment of the present invention.

As shown in fig. 40, the information processing apparatus 40 may specifically include:

an obtaining module 401, configured to obtain a time sequence of a target network element, where the time sequence includes characteristic data of a service index of the target network element at multiple times;

a processing module 402, configured to input the time sequence into a network element prediction model, so as to obtain a predicted value of the service indicator and an alarm threshold value; the network element prediction model comprises a monthly cycle fitting condition corresponding to the time sequence;

and a monitoring module 403, configured to monitor the target network element according to the actual value, the predicted value, and the alarm threshold value of the target network element.

The characteristic data in the embodiment of the invention is at least one of the following data:

The network element prediction model in the embodiment of the invention comprises a growth trend model and/or a seasonal trend model; wherein the content of the first and second substances,

Furthermore, the information processing apparatus 40 in the embodiment of the present invention may further include: a determining module 405, configured to determine, according to the initial time sequence, abnormal data in the initial time sequence; and determining the abnormal data sequence not included as a time sequence.

In a possible embodiment, the determining module 405 may be specifically configured to calculate a standard deviation of the initial time sequence according to the initial time sequence; and determining the data with the standard deviation meeting the preset standard deviation condition as abnormal data.

The processing module 402 in the embodiment of the present invention may be specifically configured to input the time sequence into a network element prediction model, and determine a target time point in the time sequence;

In a possible embodiment, the monitoring module 403 in the embodiment of the present invention may specifically be configured to calculate a target difference between the actual value and the predicted value;

In another possible embodiment, the monitoring module 403 may be further configured to generate information representing that the target network element normally operates when the comparison result indicates that the target difference value satisfies the alarm threshold value.

Therefore, in the embodiment of the invention, the time sequence is input into the network element prediction model comprising the monthly cycle fitting condition corresponding to the time sequence, so as to obtain the predicted value and the alarm threshold value of the service index. Therefore, the dynamic and period fitting condition can be adopted and is changed along with the change of the month, so that the time sequence with the characteristics of long-term trend, multiple periods and the like can be processed, and the method has the advantages of high efficiency and good robustness. And in addition, monitoring the target network element according to the actual value of the target network element, and the predicted value and the alarm threshold value determined by the network element prediction model. Therefore, the alarm threshold value can be dynamically adjusted according to the network element prediction model, and therefore the method provided by the embodiment of the invention can be used for dealing with the complex and changeable actual situation on site and really and effectively mastering the abnormal event situation, so that the misjudgment and the judgment missing situation caused by using a fixed threshold are avoided.

The electronic device 4100 includes, but is not limited to: a radio frequency unit 4101, a network module 4102, an audio output unit 4103, an input unit 4104, a sensor 4105, a display unit 4106, a user input unit 4107, an interface unit 4108, a memory 4109, a processor 4110, and a power supply 4111. Those skilled in the art will appreciate that the electronic device configuration shown in fig. 41 does not constitute a limitation of the electronic device, and that the electronic device may include more or fewer components than shown, or some components may be combined, or a different arrangement of components. In the embodiment of the present invention, the electronic device includes, but is not limited to, a mobile phone, a tablet computer, a notebook computer, a palm computer, a vehicle-mounted terminal, a wearable device, a pedometer, and the like.

It should be understood that, in the embodiment of the present invention, the radio frequency unit 4101 may be configured to receive and send signals during a message sending and receiving process or a call process, and specifically, receive downlink resources from a base station and then process the received downlink resources to the processor 4110; in addition, the uplink resource is transmitted to the base station. In general, radio unit 4101 includes, but is not limited to, an antenna, at least one amplifier, a transceiver, a coupler, a low noise amplifier, a duplexer, and the like. In addition, the radio frequency unit 4101 may also communicate with a network and other devices through a wireless communication system.

The electronic device provides wireless broadband internet access to the user through the network module 4102, such as assisting the user in sending and receiving e-mail, browsing web pages, and accessing streaming media.

The audio output unit 4103 may convert an audio resource received by the radio frequency unit 4101 or the network module 4102 or stored in the memory 4109 into an audio signal and output as sound. Also, the audio output unit 4103 may also provide audio output related to a specific function performed by the electronic apparatus 4100 (e.g., a call signal reception sound, a message reception sound, etc.). The audio output unit 4103 includes a speaker, a buzzer, a receiver, and the like.

The input unit 4104 is for receiving an audio or video signal. The input Unit 4104 may include a Graphics Processing Unit (GPU) 41041 and a microphone 41042, the Graphics processor 41041 Processing image resources of still pictures or video obtained by an image capturing device (such as a camera) in a video capturing mode or an image capturing mode. The processed image frames may be displayed on the display unit 4107. The image frames processed by the graphic processor 41041 may be stored in the memory 4109 (or other storage medium) or transmitted via the radio frequency unit 4101 or the network module 4102. The microphone 41042 can receive sound and can process such sound into an audio asset. The processed audio resources may be converted to a format output transmittable to a mobile communication base station via the radio frequency unit 4101 in case of a phone call mode.

The electronic device 4100 also includes at least one sensor 4105, such as light sensors, motion sensors, and other sensors. Specifically, the light sensor includes an ambient light sensor that can adjust the brightness of the display panel 41061 according to the brightness of ambient light, and a proximity sensor that can turn off the display panel 41061 and/or backlight when the electronic device 4100 is moved to the ear. As one type of motion sensor, an accelerometer sensor can detect the magnitude of acceleration in each direction (generally three axes), detect the magnitude and direction of gravity when stationary, and can be used to identify the posture of an electronic device (such as horizontal and vertical screen switching, related games, magnetometer posture calibration), and vibration identification related functions (such as pedometer, tapping); the sensors 4105 may also include fingerprint sensors, pressure sensors, iris sensors, molecular sensors, gyroscopes, barometers, hygrometers, thermometers, infrared sensors, etc., which are not described in detail herein.

The display unit 4106 is used to display information input by a user or information provided to the user. The Display unit 4106 may include a Display panel 41061, and the Display panel 41061 may be configured in the form of a Liquid Crystal Display (LCD), an Organic Light-Emitting Diode (OLED), or the like.

The user input unit 4107 may be used to receive input numeric or character information and generate key signal inputs related to user settings and function control of the electronic device. Specifically, the user input unit 4107 includes a touch panel 41071 and other input devices 41072. The touch panel 41071, also referred to as a touch screen, may collect touch operations by a user on or near the touch panel 41071 (e.g., operations by a user on or near the touch panel 41071 using a finger, a stylus, or any suitable object or attachment). The touch panel 41071 may include two parts of a touch detection device and a touch controller. The touch detection device detects the touch direction of a user, detects a signal brought by touch operation and transmits the signal to the touch controller; the touch controller receives touch information from the touch detection device, converts the touch information into touch point coordinates, sends the touch point coordinates to the processor 4110, and receives and executes commands sent by the processor 4110. In addition, the touch panel 41071 can be implemented by various types such as a resistive type, a capacitive type, an infrared ray, and a surface acoustic wave. In addition to the touch panel 41071, the user input unit 4107 may include other input devices 41072. In particular, other input devices 41072 can include, but are not limited to, a physical keyboard, function keys (e.g., volume control keys, switch keys, etc.), a trackball, a mouse, and a joystick, which are not described in detail herein.

Further, the touch panel 41071 may be overlaid on the display panel 41061, and when the touch panel 41071 detects a touch operation on or near the touch panel 41071, the touch operation is transmitted to the processor 4110 to determine the type of the touch event, and then the processor 4110 provides a corresponding visual output on the display panel 41061 according to the type of the touch event. Although in fig. 41, the touch panel 41071 and the display panel 41061 are two independent components to implement the input and output functions of the electronic device, in some embodiments, the touch panel 41071 and the display panel 41061 may be integrated to implement the input and output functions of the electronic device, and is not limited herein.

The interface unit 4108 is an interface for connecting an external device to the electronic apparatus 4100. For example, the external device may include a wired or wireless headset port, an external power supply (or battery charger) port, a wired or wireless resource port, a memory card port, a port for connecting a device having an identification module, an audio input/output (I/O) port, a video I/O port, an earphone port, and the like. The interface unit 4108 may be used to receive input (e.g., resource information, power, etc.) from external devices and transmit the received input to one or more elements within the electronic apparatus 4100 or may be used to transmit resources between the electronic apparatus 4100 and external devices.

Memory 4109 may be used to store software programs as well as various resources. The memory 4109 may mainly include a storage program area and a storage resource area, wherein the storage program area may store an operating system, an application program required by at least one function (such as a sound playing function, an image playing function, etc.), and the like; the storage resource area may store resources (such as audio resources, a phonebook, etc.) created according to the use of the cellular phone, and the like. Further, memory 4109 can include high speed random access memory, and can also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other volatile solid state storage device.

The processor 4110 is a control center of the electronic device, connects various parts of the whole electronic device by using various interfaces and lines, and performs various functions and processing resources of the electronic device by running or executing software programs and/or modules stored in the memory 4109 and calling resources stored in the memory 4109, thereby performing overall monitoring of the electronic device. The processor 4110 may include one or more processing units; preferably, the processor 4110 may integrate an application processor and a modem processor, wherein the application processor mainly handles operating systems, user interfaces, application programs, and the like, and the modem processor mainly handles wireless communication. It is to be appreciated that the modem processor may not be integrated into processor 4110.

The electronic device 4100 may further comprise a power supply 4111 (such as a battery) for supplying power to various components, and preferably, the power supply 4111 is logically connected to the processor 4110 via a power management system, so as to manage charging, discharging, and power consumption management functions via the power management system.

In addition, the electronic device 4100 includes some functional modules that are not shown, and are not described in detail herein.

Embodiments of the present invention also provide a computer-readable storage medium on which a computer program is stored, which, when executed in a computer, causes the computer to perform the steps of the information processing method of an embodiment of the present invention.

It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.

Through the above description of the embodiments, those skilled in the art will clearly understand that the method of the above embodiments can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware, but in many cases, the former is a better implementation manner. Based on such understanding, the technical solutions of the present invention may be embodied in the form of a software product, which is stored in a storage medium (such as ROM/RAM, magnetic disk, optical disk) and includes instructions for enabling an electronic device (such as a mobile phone, a computer, a server, an air conditioner, or a network device) to execute the method according to the embodiments of the present invention.

While the present invention has been described with reference to the embodiments shown in the drawings, the present invention is not limited to the embodiments, which are illustrative and not restrictive, and it will be apparent to those skilled in the art that various changes and modifications can be made therein without departing from the spirit and scope of the invention as defined in the appended claims.

Claims

1. An information processing method applied to an electronic device, the method comprising:

inputting the time sequence into the network element prediction model to obtain a predicted value and an alarm threshold value of the service index; wherein the network element prediction model comprises a monthly cycle fitting condition corresponding to the time series;

2. The method of claim 1, wherein the characteristic data is at least one of:

the trend data is used for representing the variation trend of the time series within a first preset time; the periodic data is used for representing the variation trend of the time series season or periodicity; the random data is used for representing the variation trend of various factors of the time series;

wherein the periodic data corresponds to a monthly cycle fitting condition in the network element prediction model.

3. The method of claim 1 or 2, wherein the trend data, periodic data, or random data comprises at least one of: total XDR number, quality ratio and scene perception index data of instant messaging.

4. The method according to claim 1 or 2, characterized in that the network element prediction model comprises a growth trend model and/or a seasonal trend model; wherein the content of the first and second substances,

the seasonal trend model is used for processing the time sequence according to the monthly cycle fitting condition, the daily cycle fitting condition and the annual cycle fitting condition in the seasonal trend model to obtain preset values corresponding to the monthly cycle fitting condition, the daily cycle fitting condition or the annual cycle fitting condition.

5. The method of claim 4, wherein the characteristic value of the monthly cycle feature in the monthly cycle fitting condition is 30.5; and the month cycle characteristic value is used for fitting the time sequence.

6. The method of claim 1, wherein prior to the step of obtaining the time sequence of the target network element, the method further comprises:

determining that the abnormal data sequence is not included as the time sequence.

7. The method of claim 6, wherein determining the anomalous data in the initial time series from the initial time series comprises:

and determining the data with the standard deviation meeting a preset standard deviation condition as the abnormal data.

8. The method according to claim 6 or 7, wherein the inputting the time series into the network element prediction model to obtain the predicted value of the service indicator and the alarm threshold value comprises:

inputting the time sequence into the network element prediction model, and determining a target time point in the time sequence;

calculating an average value of the target time point, a previous time point and a next time point related to the target time point;

9. The method of claim 1, wherein the monitoring the target network element according to the actual value, the predicted value, and the alarm threshold value of the target network element comprises:

comparing the target difference value with the alarm threshold value to obtain a comparison result;

10. The method of claim 9, wherein when the comparison result indicates that the target difference satisfies the alarm threshold, generating information indicating that the target network element normally operates.

11. An information processing apparatus characterized by comprising:

an obtaining module, configured to obtain a time sequence of a target network element, where the time sequence includes characteristic data of a service indicator of the target network element at multiple times;

the processing module is used for inputting the time sequence into the network element prediction model to obtain a predicted value and an alarm threshold value of the service index; wherein the network element prediction model comprises a monthly cycle fitting condition corresponding to the time series;

12. An electronic device, comprising a processor, a memory, and a computer program stored on the memory and executable on the processor, the computer program, when executed by the processor, implementing the information processing method of claims 1-10.

13. A computer-readable storage medium, having stored thereon a computer program which, if executed in a computer, causes the computer to execute an information processing method according to claims 1 to 10.