Network load prediction method and device, electronic equipment and storage medium

Abstract

Embodiments of the present application provide a network load prediction method and device, electronic equipment and storage medium, and relate to the technical field of information processing. In the embodiments of the present application, based on the actual network load data of the current time period and the pre-trained prediction model, the network load prediction data of the next time period is obtained. Thus, the embodiments of the present application can realize network load prediction based on real data, get rid of the dependence on professional experience of managers, and avoid misjudgment caused by professional experience. In the embodiments of the present application, the network load is analyzed and predicted by a big data intelligent method, uninterrupted monitoring and analysis of the network load condition are realized, and real-time analysis with high efficiency and high accuracy is realized. In the embodiments of the present application, when the network load is predicted, N exponential smoothing components are considered, so that the influence of seasonal, periodic and occasional factors can be excluded in the fluctuating historical network load data, and the real network load change condition can be accurately identified.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of information processing technology, and in particular to a network load prediction method, apparatus, electronic device and storage medium. Background Technology

[0002] With the promotion of network services, the number of network users and business traffic has increased significantly, and network load has repeatedly reached new highs. In this context of rapid network service development, the requirements for network resource expansion and construction are becoming increasingly stringent. Insufficient network resource capacity can lead to delays and other impacts on service experience during peak traffic periods.

[0003] In traditional operating models, network size and capacity assessments largely rely on the maintenance experience of administrators, who subjectively provide capacity recommendations. Furthermore, due to the long cycles of network expansion and adjustment, administrators typically need to predict business capacity growth. This method of subjective prediction by administrators has at least the following problems: 1) it fails to effectively utilize real-world data; 2) subjective predictions may have accuracy issues.

[0004] Therefore, there is an urgent need for a new method for network load forecasting to improve the accuracy of network load forecasting. Summary of the Invention

[0005] This invention provides a network load prediction method, apparatus, electronic device, and storage medium to at least partially solve the problems existing in related technologies.

[0006] The first aspect of this invention provides a network load prediction method, the method comprising:

[0007] Collect actual network load data for the current time period;

[0008] Based on the actual network load data for the current time period, the Nth exponentially smoothed actual component for the current time period is obtained, where N is any integer between 1 and 3.

[0009] The actual network load data for the current time period and the Nth exponentially smoothed actual component for the current time period are input into a pre-trained prediction model to obtain the Nth exponentially smoothed predicted component for the next time period.

[0010] Based on the Nth exponentially smoothed prediction component of the next time period, the network load prediction data for the next time period is obtained.

[0011] Optionally, the training steps of the prediction model include:

[0012] Collect actual network load data for multiple historical time periods, including at least a first time period and a second time period;

[0013] Based on the actual network load data for the first time period, the Nth exponentially smoothed actual component for the first time period is obtained.

[0014] The actual network load data for the first time period and the Nth exponentially smoothed actual component of the first time period are input into a preset model to obtain the Nth exponentially smoothed predicted component for the second time period.

[0015] Based on the Nth exponentially smoothed prediction component of the second time period, the network load prediction data for the second time period is obtained;

[0016] The model parameters of the preset model are adjusted based on the difference between the network load prediction data and the actual network load data for the second time period.

[0017] Using actual network load data from multiple historical time periods following the second time period, the model parameters of the preset model are further adjusted.

[0018] Optionally, it also includes:

[0019] Null and / or zero values ​​in the collected actual network load data are filled with previous values, where the previous values ​​are: actual network load data collected within a historical time period prior to the collected actual network load data; or

[0020] The null and / or zero values ​​in the collected actual network load data are filled with the mean value from previous periods. The mean value from previous periods is the average value of the actual network load data collected over multiple historical time periods prior to the collected actual network load data.

[0021] Optionally, it also includes:

[0022] The historical weighted offset coefficient is determined based on the offset between the actual network load data of two adjacent time periods in multiple historical time periods.

[0023] The offset coefficient is determined based on the actual network load data of the first time period and the actual network load data of the second time period. The second historical time period and the first historical time period are corresponding time periods in history.

[0024] Based on the historical weighted offset coefficient and the offset coefficient, outliers in the actual network load data for the first time period are determined;

[0025] Outliers in the actual network load data for the first time period are filled with earlier values, where earlier values ​​are actual network load data collected in historical time periods prior to the first time period.

[0026] Optionally, the method further includes:

[0027] Set a first threshold value and a second threshold value, wherein the first threshold value is greater than the second threshold value;

[0028] Based on the network load forecast data for the next time period and the relationship between the first threshold and the second threshold, output expansion-related prompts.

[0029] Optionally, based on the network load forecast data for the next time period and the relationship between the first threshold and the second threshold, expansion-related prompts are output, including:

[0030] If the predicted network load for the next time period is not greater than the second threshold, the network load for the next time period is determined to be normal, and a prompt indicating that no capacity expansion is needed is output.

[0031] Optionally, based on the network load forecast data for the next time period and the relationship between the first threshold and the second threshold, expansion-related prompts are output, including:

[0032] If the predicted network load for the next time period is greater than the second threshold but not greater than the first threshold, it is determined that the network load for the next time period is approaching saturation, and a suggestion to expand capacity is output.

[0033] Optionally, based on the network load forecast data for the next time period and the relationship between the first threshold and the second threshold, expansion-related prompts are output, including:

[0034] If the predicted network load for the next time period is greater than the first threshold, it is determined that the network load for the next time period exceeds the network capacity, and a capacity expansion prompt is output.

[0035] A second aspect of the present invention provides a network load prediction device, the device comprising:

[0036] The device includes:

[0037] The data acquisition module is used to collect actual network load data for the current time period.

[0038] The component determination module is used to obtain the Nth exponentially smoothed actual component of the current time period based on the actual network load data of the current time period, where N is any integer between 1 and 3.

[0039] The component prediction module is used to input the actual network load data of the current time period and the Nth exponentially smoothed actual component of the current time period into the pre-trained prediction model to obtain the Nth exponentially smoothed predicted component of the next time period.

[0040] The network load forecasting module is used to obtain the network load forecasting data for the next time period based on the Nth exponential smoothing forecast component of the next time period.

[0041] Optionally, the training steps of the prediction model include:

[0042] Collect actual network load data for multiple historical time periods, including at least a first time period and a second time period;

[0043] Based on the actual network load data for the first time period, the Nth exponentially smoothed actual component for the first time period is obtained.

[0044] The actual network load data for the first time period and the Nth exponentially smoothed actual component of the first time period are input into a preset model to obtain the Nth exponentially smoothed predicted component for the second time period.

[0045] Based on the Nth exponentially smoothed prediction component of the second time period, the network load prediction data for the second time period is obtained;

[0046] The model parameters of the preset model are adjusted based on the difference between the network load prediction data and the actual network load data for the second time period.

[0047] Using actual network load data from multiple historical time periods following the second time period, the model parameters of the preset model are further adjusted.

[0048] Optionally, it also includes:

[0049] The first filling module is used to fill null and / or zero values ​​in the collected actual network load data with previous values. The previous values ​​are: actual network load data collected within a historical time period prior to the collected actual network load data; or...

[0050] The null and / or zero values ​​in the collected actual network load data are filled with the mean value from previous periods. The mean value from previous periods is the average value of the actual network load data collected over multiple historical time periods prior to the collected actual network load data.

[0051] Optionally, it also includes:

[0052] The second filling module is used to determine a historical weighted offset coefficient based on the offset between the actual network load data of every two adjacent time periods in multiple historical time periods; determine an offset coefficient based on the actual network load data of the first time period and the actual network load data of the second time period, wherein the second historical time period and the first historical time period are corresponding time periods in history; determine outliers in the actual network load data of the first time period based on the historical weighted offset coefficient and the offset coefficient; and fill the outliers in the actual network load data of the first time period with previous values, wherein the previous values ​​are: actual network load data collected in historical time periods before the first time period.

[0053] Optionally, the device further includes:

[0054] The setting module is used to set a first threshold value and a second threshold value, wherein the first threshold value is greater than the second threshold value;

[0055] The output module is used to output expansion-related prompts based on the network load prediction data for the next time period and the relationship between the first threshold value and the second threshold value.

[0056] Optionally, the output module is specifically used for:

[0057] If the predicted network load for the next time period is not greater than the second threshold, the network load for the next time period is determined to be normal, and a prompt indicating that no capacity expansion is needed is output.

[0058] Optionally, the output module is specifically used for:

[0059] If the predicted network load for the next time period is greater than the second threshold but not greater than the first threshold, it is determined that the network load for the next time period is approaching saturation, and a suggestion to expand capacity is output.

[0060] Optionally, the output module is specifically used for:

[0061] If the predicted network load for the next time period is greater than the first threshold, it is determined that the network load for the next time period exceeds the network capacity, and a capacity expansion prompt is output.

[0062] A third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps described in the first aspect of the present invention.

[0063] A fourth aspect of the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executed, implements the steps of the method described in the first aspect of the present invention.

[0064] In this embodiment of the invention, network load prediction data for the next time period can be obtained based on actual network load data for the current time period and a pre-trained prediction model. Therefore, this embodiment of the invention enables network load prediction based on real data, eliminating reliance on the professional experience of management personnel and avoiding misjudgments caused by such experience. In this embodiment of the invention, by using big data intelligent methods to analyze and predict network load, uninterrupted monitoring and analysis of network load can be achieved, enabling high-efficiency and high-accuracy real-time analysis. In this embodiment of the invention, when performing network load prediction, an Nth-order exponential smoothing component is considered, thereby eliminating the influence of seasonal, periodic, and occasional factors from fluctuating historical network load data and accurately identifying the true network load changes. Attached Figure Description

[0065] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0066] Figure 1 This is a flowchart of a network load prediction method according to an embodiment of the present invention;

[0067] Figure 2 This is a schematic diagram of a gimbal architecture corresponding to a network load prediction method according to an embodiment of the present invention.

[0068] Figure 3 This is a flowchart of another network load prediction method according to an embodiment of the present invention;

[0069] Figure 4 This is a flowchart of another network load prediction method according to an embodiment of the present invention;

[0070] Figure 5 This is a structural block diagram of a network load prediction device according to an embodiment of the present invention. Detailed Implementation

[0071] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0072] Reference Figure 1The flowchart illustrates a network load prediction method according to an embodiment of the present invention. The network load prediction method provided by the embodiment of the present invention may include the following steps:

[0073] S101, collect actual network load data for the current time period.

[0074] In this embodiment of the invention, time can be divided according to year, month, week or day to obtain multiple time periods, and the actual network load data within the current time period can be collected.

[0075] In this embodiment of the invention, the actual network load data can be the average value within the current time period, or it can be a sequence of actual network load data corresponding to multiple times within the current time period.

[0076] In this embodiment of the invention, each time period can be further divided to obtain multiple moments. For example, if the current time period is the current month, then the current month can be divided, with each day as a moment. Similarly, if the current time period is the current day, then the current day can be divided, with each hour as a moment.

[0077] In this embodiment of the invention, the actual network load data is used to describe the network resource load situation, which can be specifically reflected in the number of online users, network traffic value, base station throughput, etc.

[0078] S102, based on the actual network load data for the current time period, obtain the Nth exponentially smoothed actual component for the current time period, where N is any integer between 1 and 3.

[0079] Exponential smoothing (ES) can highlight the influence of recent data while taking into account factors from longer-term data, in addition to full averaging and moving averages. When using historical data to predict the future, the weight of historical data decreases exponentially over time, with more recent data having a higher weight. Commonly used methods include single-stage, double-stage, and triple-stage exponential smoothing. Single-stage exponential smoothing is suitable for predicting time series without obvious trends or seasonality, double-stage exponential smoothing can be used for predicting time series with trends, and triple-stage exponential smoothing is suitable for time series containing horizontal, trend, and seasonal characteristics. In this embodiment of the invention, single-stage, double-stage, or triple-stage exponential smoothing can be used, with triple-stage exponential smoothing being the optimal choice.

[0080] In this embodiment of the invention, the exponential smoothing method can be used to decompose the actual network load data for the current time period to obtain the Nth exponentially smoothed actual components for the current time period.

[0081] Specifically, taking triple exponential smoothing as an example, the actual network load data for the current time period can be decomposed into three series: trend, seasonal, and horizontal component (white noise).

[0082] S103, input the actual network load data of the current time period and the Nth exponentially smoothed actual component of the current time period into the pre-trained prediction model to obtain the Nth exponentially smoothed prediction component of the next time period.

[0083] In this embodiment of the invention, the pre-trained prediction model is obtained based on the exponential smoothing method. By inputting the actual network load data of the current time period into the trained prediction model, the Nth exponential smoothing prediction component for the next time period can be obtained.

[0084] S104, based on the Nth exponential smoothing prediction component of the next time period, obtain the network load prediction data for the next time period.

[0085] In this embodiment of the invention, the network load prediction data for the next time period can be obtained by multiplying or adding the Nth exponentially smoothed prediction components of the next time period. Therefore, based on this predicted network load data, it can be determined whether network resource expansion is needed.

[0086] Reference Figure 2 This diagram illustrates the training steps of the prediction model used in the network load prediction method provided in this embodiment of the invention. The prediction model trained using the training steps provided in this embodiment of the invention can be applied to the aforementioned step S103. Specifically, the training steps include:

[0087] S201, collect actual network load data for multiple historical time periods, wherein the multiple historical time periods include at least a first time period and a second time period.

[0088] The division of the current time period in step S101 can correspond to the division of historical time periods in the embodiments of the present invention. For example, if the historical time period is divided in days, then the current time period is today.

[0089] S202, based on the actual network load data including the first time period, obtain the Nth exponentially smoothed actual component of the first time period.

[0090] In this embodiment of the invention, an exponential smoothing method is used to smooth the actual network load data for a first time period, obtaining the corresponding Nth exponentially smoothed actual component for the first time period. In this embodiment, the initial Nth exponentially smoothed actual component for the first time period can be set as the actual network load data for the first time period.

[0091] S203, input the actual network load data of the first time period and the Nth exponentially smoothed actual component of the first time period into the preset model to obtain the Nth exponentially smoothed predicted component of the second time period.

[0092] In this embodiment of the invention, the actual network load data of the first time period and the Nth exponentially smoothed actual component of the first time period are used as independent variables and input into a preset model to obtain the Nth exponentially smoothed predicted component of the second time period, and the Nth exponentially smoothed predicted component of the second time period is used as the dependent variable.

[0093] S204. Based on the Nth exponential smoothing prediction component of the second time period, obtain the network load prediction data for the second time period.

[0094] S205, adjust the model parameters of the preset model based on the difference between the network load prediction data and the actual network load data for the second time period.

[0095] In this embodiment of the invention, during the model adjustment process, the actual network load data for the second time period is known. Therefore, the known value can be compared with the network load prediction data for the second time period obtained by the model, and the model parameters of the preset model can be adjusted according to the comparison result.

[0096] S206, using actual network load data from multiple historical time periods following the second time period, the model parameters of the preset model are further adjusted.

[0097] In this embodiment of the invention, predictions and comparisons can be continuously performed to adjust the model parameters of the preset model until the optimal parameters are obtained. In specific applications, the optimal parameters can be obtained through AI analysis and gradient descent.

[0098] To facilitate understanding, the training steps of the prediction model provided in the embodiments of the invention will be illustrated with an example, specifically, the first exponential smoothing method:

[0099] Let L t Y is the smoothed value of time t. t-1 It is the actual value of time t-1, L t -1 is the smoothed value at time t-1, and a is the damping coefficient of the horizontal component, ranging from 0 to 1. The formula for first-order exponential smoothing is:

[0100] L t =aY t-1 +(1-a)L t-1

[0101] In this embodiment of the invention, a value 'a' can be preset. The actual network load data (actual value at time t-1) for the first time period and the predicted value of the horizontal component of the first time period (smoothed value at time t-1) are input into the exponential smoothing formula to obtain the predicted network load data (smoothed value at time t) for the second time period. Then, based on the difference between the predicted network load data (smoothed value at time t) and the actual network load data (actual value at time t) for the second time period, the model parameter 'a' of the preset model is adjusted until the optimal model parameter 'a' is obtained.

[0102] Specifically, taking the triple exponential smoothing method as an example:

[0103] Let L t It is the horizontal component, hb t It is the trend component, S t-p If it is a seasonal component, then the triple exponential smoothing formula after superimposing the trend component and the seasonal component can be decomposed as follows:

[0104] Actual value: Y t+1 =L t +hb t +S t-p

[0105] Horizontal component: L t =α(Y t -S t-p )+(1-α)(L t-1 +hb t-1 )

[0106] Trend component: hb t =β(L t -L t-1 )+(1-β)hb t-1

[0107] Periodic component: S t =γ(Y) t -L t )+(1-γ)S t-p

[0108] Through the above decomposition, the embodiments of the present invention can decompose real data into three types of sequences: trend, seasonality, and horizontal component (white noise). Specifically, the Holt-Winter algorithm obtains the predicted value by weighting the three components. Among them, the three parameters α, β, and γ all take values ​​in the range [0, 1].

[0109] Similar to the first exponential smoothing method, the values ​​of three parameters α, β, and γ can be preset. The actual network load data for the first time period (the actual value at time t-1) is decomposed to obtain the actual component of the first time period with triple exponential smoothing. The actual network load data for the first time period and the actual component of the first time period with triple exponential smoothing are input into the preset model to obtain the predicted component of the second time period with triple exponential smoothing (the predicted component at time t). Then, based on the predicted component of the triple exponential smoothing (the predicted component at time t), the difference between the predicted data of the second time period with the predicted value at time t and the actual data of the second time period with the actual value at time t is obtained. The model parameters α, β, and γ of the preset model are adjusted until the optimal model parameters α, β, and γ are obtained.

[0110] In this embodiment of the invention, when training the model, training can be based on existing historical data, or the time series can be incrementally updated based on real-time data generated in real time, and the updated data can be used to continuously optimize the model.

[0111] Reference Figure 3 The flowchart illustrates a network load prediction method according to an embodiment of the present invention. The network load prediction method provided by the embodiment of the present invention may include the following steps:

[0112] S301, collect actual network load data for multiple historical time periods, wherein the multiple historical time periods include at least a first time period and a second time period.

[0113] Step S301 is similar to step S201 described above, and will not be repeated here in this embodiment of the invention.

[0114] S302, Fill in null and / or zero data.

[0115] In this embodiment of the invention, null and / or zero data may occur during network load data collection. Null data refers to data where the actual network load data for a specific historical time period may be missing, resulting in a null value. Zero data refers to data where the actual network load data for a specific historical time period may be incorrect, resulting in a zero value.

[0116] In this embodiment of the invention, the previous value filling method or the previous mean filling method can be used to fill in the null value data and the 0 value data, so as to improve the time series composed of the actual network load data of the multiple historical time periods.

[0117] Specifically, step S302 may include:

[0118] Null and / or zero values ​​in the collected actual network load data can be filled with previous values, where the previous values ​​are actual network load data collected in historical time periods prior to the collected actual network load data; or the average of previous values ​​can be filled with the null and / or zero values ​​in the collected actual network load data, where the average of previous values ​​is the average of actual network load data collected in multiple historical time periods prior to the collected actual network load data.

[0119] In this embodiment of the invention, after discovering null or zero values ​​in the actual network load data, the previous values ​​of that data can be verified to determine whether the previous values ​​are normal. Specifically, the verification method can be to verify whether the difference between the previous value and the average of its historical values ​​for the same period is normal.

[0120] In this embodiment of the invention, when the previous values ​​are normal, the previous values ​​are used to fill in the null and / or zero values ​​in the collected actual network load data.

[0121] In this embodiment of the invention, when the previous value is not a normal value, the previous average value is used to fill the null value data and / or zero value data in the actual network load data collected.

[0122] In this embodiment of the invention, the multiple historical time periods corresponding to the previous average can be determined according to the granularity of the historical time period division. For example, if the granularity of the historical time period division is days, then the average of the actual network load data collected in the previous 10 days can be taken as the previous average.

[0123] S303, Outlier Correction.

[0124] In this embodiment of the invention, outliers may occur during network load data collection. Outliers refer to abnormal data with large deviations.

[0125] In this embodiment of the invention, outliers can be corrected to ensure the validity and accuracy of historical data.

[0126] Specifically, the following steps are included:

[0127] S3031, determine the historical weighted offset coefficient based on the offset between the actual network load data of every two adjacent time periods in multiple historical time periods.

[0128] In this embodiment of the invention, the historical weighted offset coefficient corresponds to the overall time series and is generally a relatively stable value.

[0129] Specifically, in this embodiment of the invention, the offset between pairs of data in the historical data is set to F1 to F... t The historical weighted offset is

[0130] S3032, determine the offset coefficient based on the actual network load data of the first time period and the actual network load data of the second time period, wherein the second historical time period and the first historical time period are corresponding time periods in history.

[0131] In this embodiment of the invention, the corresponding time period in history refers to the corresponding time period in different time periods, such as the same month in different years, or the same hour in different days.

[0132] Specifically, let t be the current data, Yt be the current value, and Yt-1 be the previous value, which can be the data from the same period of the previous day. Set the offset coefficient Ft, then:

[0133] S3033, Based on the historical weighted offset coefficient and the offset coefficient, determine the outliers in the actual network load data for the first time period.

[0134] In this embodiment of the invention, if the offset coefficient F t Same as historical weighted offset coefficient If the difference is too large, the actual network load data for the first time period is judged to be an anomaly.

[0135] Specifically, if Therefore, it is determined that the offset coefficient Ft is the same as the historical weighted offset coefficient. If the difference is too large, the actual network load data for the first time period is determined to be an anomaly.

[0136] S3034, use previous values ​​to fill out the outliers in the actual network load data of the first time period. The previous values ​​are: actual network load data collected in historical time periods before the first time period.

[0137] In this embodiment of the invention, after determining an outlier, it is possible to continue to determine whether its previous value is an outlier. If its previous value is an outlier, it is possible to continue to determine whether the previous value of the previous value is an outlier, until a reasonable previous value is determined and the outlier is filled with the reasonable previous value.

[0138] S304, Based on the actual network load data including the first time period, obtain the Nth exponentially smoothed actual component of the first time period.

[0139] S305, input the actual network load data of the first time period and the Nth exponentially smoothed actual component of the first time period into the preset model to obtain the Nth exponentially smoothed predicted component of the second time period.

[0140] S306. Based on the Nth exponential smoothing prediction component of the second time period, obtain the network load prediction data for the second time period.

[0141] S307, Adjust the model parameters of the preset model based on the difference between the network load prediction data and the actual network load data for the second time period.

[0142] S308, using actual network load data from multiple historical time periods following the second time period, the model parameters of the preset model are further adjusted.

[0143] Steps S304-S308 are similar to the aforementioned steps S202-S206, and will not be described again in this embodiment of the invention.

[0144] Reference Figure 4 The flowchart illustrates a network load prediction method according to an embodiment of the present invention. The network load prediction method provided by the embodiment of the present invention may include the following steps:

[0145] S401 collects actual network load data for the current time period.

[0146] In this embodiment of the invention, after collecting the actual network load data for the current time period, the null value data and / or zero value data filling and outlier correction steps provided in steps S302-S303 above can be used to adjust the actual network load data for the current time period to obtain more accurate actual network load data.

[0147] S402, based on the actual network load data for the current time period, obtain the Nth exponentially smoothed actual component for the current time period, where N is any integer between 1 and 3.

[0148] S403, input the actual network load data of the current time period and the Nth exponentially smoothed actual component of the current time period into the pre-trained prediction model to obtain the Nth exponentially smoothed predicted component of the next time period.

[0149] S404, based on the Nth exponential smoothing prediction component of the next time period, obtain the network load prediction data for the next time period.

[0150] Steps S401-S404 are similar to the aforementioned steps S101-S104, and will not be described again in this embodiment of the invention.

[0151] S405, set a first threshold value and a second threshold value, wherein the first threshold value is greater than the second threshold value.

[0152] In this embodiment of the invention, the first threshold value and the second threshold value can remain fixed and are determined based on the maximum network load of the network system.

[0153] In this embodiment of the invention, considering that the maximum network load of the network system may change, it can also be updated periodically by technicians. Alternatively, an intelligent algorithm can be used to obtain the first and second threshold values ​​based on the maximum network load of the same historical period.

[0154] S406, based on the network load prediction data for the next time period and the relationship between the first threshold value and the second threshold value, output expansion-related prompts.

[0155] In this embodiment of the invention, the need for capacity expansion is determined based on the network load prediction data for the next time period and the relationship between the first threshold value and the second threshold value.

[0156] Specifically, S406 includes the following three cases:

[0157] If the predicted network load for the next time period is not greater than the second threshold, the network load for the next time period is determined to be normal, and a prompt indicating that no capacity expansion is needed is output.

[0158] If the predicted network load for the next time period is greater than the second threshold but not greater than the first threshold, it is determined that the network load for the next time period is approaching saturation, and a suggestion to expand capacity is output.

[0159] In this embodiment of the invention, if the network load approaches saturation in the next time period, the current network may not be expanded. However, considering that the network cannot cope with sudden network load when the network load is saturated, a suggestion to expand capacity can be output to ensure that the network load is at a healthy level. Technicians can further determine whether expansion is necessary based on the actual situation.

[0160] If the predicted network load for the next time period is greater than the first threshold, it is determined that the network load for the next time period exceeds the network capacity, and a capacity expansion prompt is output to ensure that the network capacity is within a reasonable range.

[0161] In practical applications, the monitoring program can be determined based on the first threshold value and the second threshold value. The monitoring program can identify the network capacity status according to the threshold judgment logic in step S406 above, and push reminders according to the rules.

[0162] Based on the same inventive concept, embodiments of the present invention provide a network load prediction device, with reference to... Figure 5 , Figure 5 This is a schematic diagram of the network load prediction device provided in an embodiment of the present invention. Figure 5 As shown, the device includes:

[0163] The data acquisition module 501 is used to collect actual network load data for the current time period.

[0164] The component determination module 502 is used to obtain the Nth exponentially smoothed actual component of the current time period based on the actual network load data of the current time period, where N is any integer between 1 and 3.

[0165] The component prediction module 503 is used to input the actual network load data of the current time period and the Nth exponentially smoothed actual component of the current time period into the pre-trained prediction model to obtain the Nth exponentially smoothed predicted component of the next time period.

[0166] The network load prediction module 504 is used to obtain the network load prediction data for the next time period based on the Nth exponential smoothing prediction component of the next time period.

[0167] Optionally, the training steps of the prediction model include:

[0168] Collect actual network load data for multiple historical time periods, including at least a first time period and a second time period;

[0169] Based on the actual network load data for the first time period, the Nth exponentially smoothed actual component for the first time period is obtained.

[0170] The actual network load data for the first time period and the Nth exponentially smoothed actual component of the first time period are input into a preset model to obtain the Nth exponentially smoothed predicted component for the second time period.

[0171] Based on the Nth exponentially smoothed prediction component of the second time period, the network load prediction data for the second time period is obtained;

[0172] The model parameters of the preset model are adjusted based on the difference between the network load prediction data and the actual network load data for the second time period.

[0173] Using actual network load data from multiple historical time periods following the second time period, the model parameters of the preset model are further adjusted.

[0174] Optionally, it also includes:

[0175] The first filling module is used to fill null and / or zero values ​​in the collected actual network load data with previous values. The previous values ​​are: actual network load data collected within a historical time period prior to the collected actual network load data; or...

[0176] The null and / or zero values ​​in the collected actual network load data are filled with the mean value from previous periods. The mean value from previous periods is the average value of the actual network load data collected over multiple historical time periods prior to the collected actual network load data.

[0177] Optionally, it also includes:

[0178] The second filling module is used to determine a historical weighted offset coefficient based on the offset between the actual network load data of every two adjacent time periods in multiple historical time periods; determine an offset coefficient based on the actual network load data of the first time period and the actual network load data of the second time period, wherein the second historical time period and the first historical time period are corresponding time periods in history; determine outliers in the actual network load data of the first time period based on the historical weighted offset coefficient and the offset coefficient; and fill the outliers in the actual network load data of the first time period with previous values, wherein the previous values ​​are: actual network load data collected in historical time periods before the first time period.

[0179] Optionally, the device further includes:

[0180] The setting module is used to set a first threshold value and a second threshold value, wherein the first threshold value is greater than the second threshold value;

[0181] The output module is used to output expansion-related prompts based on the network load prediction data for the next time period and the relationship between the first threshold value and the second threshold value.

[0182] Optionally, the output module is specifically used for:

[0183] If the predicted network load for the next time period is not greater than the second threshold, the network load for the next time period is determined to be normal, and a prompt indicating that no capacity expansion is needed is output.

[0184] Optionally, the output module is specifically used for:

[0185] If the predicted network load for the next time period is greater than the second threshold but not greater than the first threshold, it is determined that the network load for the next time period is approaching saturation, and a suggestion to expand capacity is output.

[0186] Optionally, the output module is specifically used for:

[0187] If the predicted network load for the next time period is greater than the first threshold, it is determined that the network load for the next time period exceeds the network capacity, and a capacity expansion prompt is output.

[0188] As the device embodiment is basically similar to the method embodiment, the description is relatively simple, and relevant parts can be found in the description of the method embodiment.

[0189] Based on the same inventive concept, embodiments of the present invention provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the network load prediction method described in any of the above embodiments.

[0190] Based on the same inventive concept, embodiments of the present invention provide an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the steps of the network load prediction method described in any of the above embodiments.

[0191] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0192] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, apparatus, or computer program products. Therefore, embodiments of the present invention can take the form of entirely hardware embodiments, entirely software embodiments, or embodiments combining software and hardware aspects. Furthermore, embodiments of the present invention can take the form of computer program products implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0193] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, terminal devices (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing terminal device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing terminal device, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0194] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing terminal device to operate in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0195] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal equipment, causing a series of operational steps to be performed on the computer or other programmable terminal equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable terminal equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0196] Although preferred embodiments of the present invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of the present invention.

[0197] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes said element.

[0198] The above provides a detailed description of the network load prediction method, apparatus, electronic device, and storage medium provided by the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A network load prediction method characterized by, The method comprises: collecting network load actual data of a current time period; obtaining an Nth exponential smoothing actual component of the current time period according to the network load actual data of the current time period, N being any integer between 1 and 3; inputting the network load actual data of the current time period and the Nth exponential smoothing actual component of the current time period into a pre-trained prediction model to obtain an Nth exponential smoothing prediction component of a next time period; obtaining network load prediction data of the next time period according to the Nth exponential smoothing prediction component of the next time period; The method further comprises: The historical weighted offset coefficient is determined according to an offset between network load actual data of each two adjacent time periods in a plurality of historical time periods; the historical weighted offset coefficient is wherein, an average value of the offset between the network load actual data of each two adjacent time periods in the plurality of historical time periods. According to the network load actual data of the first time period and the network load actual data of the second time period, which is a corresponding time period in history of the first time period, an offset coefficient is determined, and the offset coefficient is , wherein, is the network load actual data of the first time period, is the network load actual data of the second time period. According to the historical weighted offset coefficient and the offset coefficient, an abnormal value in network load actual data of a first time period is determined; wherein, if 2, it is determined that the network load actual data of the first time period is an abnormal value; filling abnormal values in the network load actual data of the first time period with previous values, the previous values being network load actual data collected in a historical time period before the first time period.

2. The method of claim 1, wherein, The training step of the prediction model comprises: collecting network load actual data of a plurality of historical time periods, the plurality of historical time periods at least comprising a first time period and a second time period; obtaining an Nth exponential smoothing actual component of the first time period according to the network load actual data of the first time period; inputting the network load actual data of the first time period and the Nth exponential smoothing actual component of the first time period into a preset model to obtain an Nth exponential smoothing prediction component of the second time period; obtaining network load prediction data of the second time period according to the Nth exponential smoothing prediction component of the second time period; adjusting model parameters of the preset model according to a difference between the network load prediction data of the second time period and the network load actual data of the second time period; continuing to adjust the model parameters of the preset model by using network load actual data of a plurality of historical time periods after the second time period.

3. The method according to claim 1 or 2, characterized in that, Further comprising: filling null data and / or zero data in the collected network load actual data with previous values, the previous values being network load actual data collected in a historical time period before the collected network load actual data; Or filling null data and / or zero data in the collected network load actual data with previous mean values, the previous mean values being mean values of network load actual data collected in a plurality of historical time periods before the collected network load actual data.

4. The method of claim 1, wherein, The method further comprises: setting a first threshold value and a second threshold value, the first threshold value being greater than the second threshold value; outputting an expansion-related prompt according to a size relationship between the network load prediction data of the next time period and the first threshold value and the second threshold value.

5. The method of claim 4, wherein, Outputting an expansion-related prompt according to a size relationship between the network load prediction data of the next time period and the first threshold value and the second threshold value comprises: determining that the network load of the next time period is normal and outputting a no expansion prompt in a case where the network load prediction data of the next time period is not greater than the second threshold value.

6. The method of claim 4, wherein, Outputting an expansion-related prompt according to a size relationship between the network load prediction data of the next time period and the first threshold value and the second threshold value comprises: In a case where the network load prediction data of the next time period is greater than the second threshold value and is not greater than the first threshold value, it is determined that the network load of the next time period is close to saturation, and a suggestion expansion prompt is output.

7. The method of claim 4, wherein, According to the size relationship between the network load prediction data of the next time period and the first threshold value and the second threshold value, an expansion related prompt is output, including: In a case where the network load prediction data of the next time period is greater than the first threshold value, it is determined that the network load of the next time period exceeds the network capacity, and an expansion prompt is output.

8. A network load prediction device, characterized by comprising: The device comprises: a collection module configured to collect network load actual data of a current time period; a component determination module configured to obtain N exponential smoothing actual components of the current time period according to the network load actual data of the current time period, N being any integer between 1 and 3; a component prediction module configured to input the network load actual data of the current time period and the N exponential smoothing actual components of the current time period into a pre-trained prediction model to obtain N exponential smoothing prediction components of a next time period; a network load prediction module configured to obtain network load prediction data of the next time period according to the N exponential smoothing prediction components of the next time period; The device further comprises: The second filling module is configured to determine a historical weighted offset coefficient according to an offset between network load actual data of each two adjacent time periods in the plurality of historical time periods; the historical weighted offset coefficient is wherein, represents an average of the offset between the network load actual data of each two adjacent time periods in the plurality of historical time periods; an offset coefficient is determined according to network load actual data of a first time period and network load actual data of a second time period, the second time period being a corresponding time period in history with the first time period; the offset coefficient is , wherein, is the network load actual data of the first time period, is the network load actual data of the second time period; an abnormal value in the network load actual data of the first time period is determined according to the historical weighted offset coefficient and the offset coefficient; wherein, if 2, it is determined that the network load actual data of the first time period is an abnormal value; an abnormal value in the network load actual data of the first time period is filled with a previous value, the previous value being network load actual data collected in a historical time period before the first time period.

9. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor implements the steps of the network load prediction method of any one of claims 1-7 when executing the program.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, The program is executed by the processor to implement the steps of the network load prediction method of any one of claims 1-7.

Images