Method and device for predicting network health degree
By predicting the health of mobile communication networks, collecting historical data and generating prediction models, the problem of traditional fault handling cannot be warned is solved, and the effect of discovering and handling network failures is achieved in advance.
Patent Information
- Application Number
- CN202311259706.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-09-26
- Publication Date
- 2025-03-28
Smart Images
Figure CN119728462A_ABST
Abstract
Description
Technical Field
[0001] The embodiments of the present invention relate to the field of communications, and in particular, to a method and device for predicting network health. Background Art
[0002] In recent years, with the development of 5G mobile communication networks, the core network structure has become more and more complex, the types of network elements have increased, and users have higher and higher requirements for network quality, so mobile network maintenance has become more and more important. Traditional responsive fault handling can no longer meet operators' requirements for network quality. How to grasp the real-time operation status of the network, actively warn of network performance risks, and move user complaints forward are important issues that operators urgently need to solve. Summary of the invention
[0003] The embodiments of the present invention provide a method and device for predicting network health, so as to at least solve the problem in the related art that it is impossible to give early warning of potential network performance risks.
[0004] According to an embodiment of the present invention, a method for predicting network health is provided, comprising:
[0005] Collect historical data on indicators of network health and obtain prediction models corresponding to the indicators;
[0006] Inputting the historical data of the indicator into the prediction model corresponding to the indicator to output the prediction data of the indicator by the prediction model;
[0007] A predicted value of the health of the network is determined based on the predicted data of the indicator.
[0008] According to another embodiment of the present invention, a network health prediction device is provided, comprising:
[0009] A first collection module is used to collect historical data of indicators for network health and obtain a prediction model corresponding to the indicator;
[0010] An input module, used to input the historical data of the indicator into the prediction model corresponding to the indicator, so as to output the prediction data of the prediction model for the indicator;
[0011] The determination module is used to determine the predicted value of the health of the network according to the predicted data of the indicator.
[0012] According to yet another embodiment of the present invention, a computer-readable storage medium is provided, in which a computer program is stored, wherein the computer program is configured to execute the steps of any one of the above method embodiments when run.
[0013] According to yet another embodiment of the present invention, there is provided an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.
[0014] Through the present invention, since the network health prediction value output by the prediction model can be used to judge the change trend of the network and provide early warning for abnormal situations, the problem that the related technology cannot provide early warning for network performance risks can be solved, so that network maintenance personnel can analyze and handle them as early as possible to reduce the occurrence of failures. BRIEF DESCRIPTION OF THE DRAWINGS
[0015] Figure 1 It is a hardware structure block diagram of a mobile terminal of a network health prediction method according to an embodiment of the present invention;
[0016] Figure 2 is a diagram of a network health prediction system architecture according to an embodiment of the present invention;
[0017] Figure 3 is a flow chart of a method for predicting network health according to an embodiment of the present invention;
[0018] Figure 4 is a network health statistics structure diagram according to an embodiment of the present invention;
[0019] Figure 5 It is a process timing diagram of a data sequence system;
[0020] Figure 6 It is a process sequence diagram of a health prediction system;
[0021] Figure 7 4 is a structural block diagram of a network health prediction device according to an embodiment of the present invention. DETAILED DESCRIPTION
[0022] Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings and in combination with the embodiments.
[0023] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.
[0024] The method embodiments provided in the embodiments of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Taking running on a mobile terminal as an example, Figure 1 FIG. 1 is a hardware structure block diagram of a mobile terminal of a network health prediction method according to an embodiment of the present invention. Figure 1 As shown, the mobile terminal may include one or more ( Figure 1Only one is shown in the figure) a processor 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA) and a memory 104 for storing data, wherein the mobile terminal may also include a transmission device 106 and an input / output device 108 for communication functions. It can be understood by those skilled in the art that Figure 1 The structure shown is for illustration only and does not limit the structure of the mobile terminal. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations shown.
[0025] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the network health prediction method in the embodiment of the present invention. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, to implement the above method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the mobile terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.
[0026] The transmission device 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the mobile terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.
[0027] Figure 2 is a diagram of the network health prediction system architecture according to an embodiment of the present invention. The embodiment of the present application can be run on Figure 2 In the system architecture shown in Figure 2As shown, the system architecture includes: a data training system and a health prediction system, wherein the data training system can be used to analyze the historical indicator data of network elements, and generate a health prediction model through training and model testing of the historical indicator data; the health prediction system can be used to obtain the prediction model and collect indicator data from the data training system, and input the collected indicator data into the prediction model to perform predictive analysis on the indicator data in the future and generate prediction information on the network health status.
[0028] The following is a further explanation of the specific structure of the data training system and health prediction system in the system architecture:
[0029] 1. Data training system
[0030] The data training system can train the model through historical data in the simulation environment to obtain a reliable network element health prediction model, which can include external display components and server components. Among them, the external display components of the data training system include a user interface module (UI) and a model export module; the server components of the data training system can include a model management module, a data processing module, and a data training module.
[0031] 1) Model management module
[0032] Used to maintain dimension indicator models and data training models.
[0033] Among them, the indicator model defines the key indicators that can constitute the measurement items of health. These indicators can form a prediction model for the indicator through data collection and data training.
[0034] The data training model can include data preprocessing rules and training models. The data preprocessing rules can refer to guiding the data training system to clean the historical data corresponding to the collected indicators and divide them into training samples through model definition; the training model can refer to the process of analyzing the change trend of the data and finally generating a prediction model.
[0035] 2) Data processing module
[0036] It includes a data acquisition submodule, a data cleaning submodule, a data aggregation submodule and a data slicing submodule, which are used for data acquisition, data cleaning, aggregation and slicing.
[0037] Among them, the data acquisition submodule can be used to obtain historical data corresponding to the indicator. The data acquisition submodule can connect to the network element layer in the simulation environment to collect simulation data in real time. It can also provide a data packet import function in the interface to import static data into the data training system.
[0038] The data cleaning submodule can be used for data standardization processing such as filtering missing values and outliers.
[0039] The data aggregation submodule can be used to aggregate the data of a time period into one or more records.
[0040] The data slicing submodule can be used to divide the collected historical data into at least one training set and at least one test set according to the time window deviation according to the training data scale train_size defined in the model. The sample data in the training set can be used for data training to generate a prediction model, and the sample data in the test set can be used to test the prediction model to verify the reliability of the prediction model.
[0041] 3) Data training module
[0042] The data training module is used to train the prediction model to be trained based on the sample data in the training set, generate the prediction model, and test the prediction model based on the sample data in the test set.
[0043] The data training module may include data annotation to clarify various types of indicator data. For example, the features in the historical data may be manually annotated in advance, and the annotated data may be used to train the prediction model so that the prediction model can accurately predict the new data.
[0044] 2. Health Prediction System
[0045] The health prediction system can be used to analyze the indicator data of the existing network environment using the prediction model, and predict the health trend changes in the short term in the future. Among them, each indicator corresponds to a prediction model, and the prediction indicator data for the future can be aggregated step by step, and finally the future network health can be generated and displayed externally. The health prediction system may include an external display component and a server-side component, wherein the external display component of the health prediction system may include a UI module and a prediction result display module; the server-side component of the health prediction system may include a model management module, a data acquisition module, and a health analysis module.
[0046] 1) Model management module
[0047] It can be used to maintain health models and prediction models. The health model is used to calculate network health based on indicator data of different dimensions; the prediction model is a prediction model for each indicator output by the data training system. After being imported into the health prediction system, it can predict the indicator data of the existing network.
[0048] 2) Data acquisition module
[0049] It can be used to collect indicator data of the existing network. For example, the query attribute lookup_size defined in the model can be used to determine how long the data series should be collected for review analysis.
[0050] 3) Health analysis module
[0051] It can be used to predict indicators through prediction models, and then aggregate the predicted values of indicators into network health through health models, and evaluate the health to analyze whether there are risks.
[0052] In this embodiment, a method for predicting the network health running on the above mobile terminal or network architecture is provided. Figure 3 is a flow chart of a method for predicting network health according to an embodiment of the present invention. Figure 3 As shown, the process includes the following steps:
[0053] Step S301, collect historical data of indicators for network health, and obtain a prediction model corresponding to the indicator.
[0054] Exemplarily, historical data of various indicators that affect the health of the current network may be collected, and a prediction model corresponding to each indicator may be obtained.
[0055] Step S302: inputting the historical data of the indicator into the prediction model corresponding to the indicator to output the prediction data of the indicator by the prediction model.
[0056] Exemplarily, the historical data of each indicator may be input into the corresponding prediction model respectively, and the prediction data of each indicator may be output respectively.
[0057] Step S303: Determine a predicted value of the health of the network according to the predicted data of the indicator.
[0058] For example, the health of the network can be calculated based on the predicted data of the indicator to achieve the prediction of the network health.
[0059] In an exemplary embodiment, the network includes at least one network element type, each of the network element type includes at least one network element, each of the network element includes at least one data dimension, and each of the data dimension includes at least one of the indicators.
[0060] As an example, the data of the network element can be dimensioned according to fault, configuration, accounting, performance and security (FCAPS). For example, the data dimensions may include: performance dimension, alarm dimension, configuration dimension, etc.
[0061] As an example, the data dimension may include one or more indicators. For example, the performance dimension may include indicators such as traffic, delay, and load.
[0062] Figure 4 is a network health statistics structure diagram according to an embodiment of the present invention, referring to Figure 4 The network may include at least one type of network element. For example, the network may include multiple network element types such as a session management function network element (SMF), an access and mobility management function network element (AMF), a user plane function network element (UPF), a registration and discovery function network element (NF Repository Function, NRF), etc.
[0063] Each network element type may include at least one network element. For example, the SMF network element type may include an SMF01 network element and an SMF02 network element.
[0064] Each network element may include at least one data dimension. For example, the SMF01 network element may include data dimensions such as performance, warning, log, and configuration.
[0065] Each data dimension may include at least one indicator. For example, the performance data dimension may include indicators such as traffic, latency, and load; the alarm data dimension may include indicators such as network element alarms, network alarms, and equipment alarms; the log data dimension may include indicators such as abnormal events; and the configuration data dimension may include configurations such as licenses.
[0066] The step S303 may specifically include:
[0067] Step S3031, obtaining the weight corresponding to the indicator, the weight corresponding to the data dimension, and the weight corresponding to the network element type;
[0068] Step S3032, determining a predicted health value of the data dimension corresponding to the indicator according to the predicted data of the indicator and the weight corresponding to the indicator;
[0069] Step S3033, determining the predicted health value of the network element corresponding to the data dimension according to the predicted health value of the data dimension and the weight corresponding to the data dimension;
[0070] Step S3034, determining the predicted health value of the network element type according to the predicted health value of the network element;
[0071] Step S3035: Determine the predicted value of the health of the network according to the predicted value of the health of the network element type and the corresponding weight of the network element type.
[0072] As an example, the weight can be the proportion of each indicator prediction data when calculating the network health according to the data dimension, the proportion of each data dimension when calculating the network health according to the network element, or the proportion of each network element type when calculating the network health according to the network.
[0073] As an example, the health algorithms at each level can be shown in Table 1 below:
[0074]
[0075] The network health can be calculated based on the algorithm in Table 1.
[0076] In an exemplary embodiment, after outputting the prediction data of the prediction model for the indicator, the method further includes: generating a predicted change trend of the indicator according to the prediction data of the indicator.
[0077] As an example, after each prediction model outputs prediction data for its own indicator, the prediction data for each indicator may be aggregated to generate a prediction change trend of the indicator.
[0078] In an exemplary embodiment, before collecting historical data of indicators for network health, the method further includes:
[0079] Step S3001, obtaining a preset indicator model and a data training model for network health; the preset indicator model includes at least one indicator;
[0080] Step S3002, collecting historical network element data of the network according to the preset indicator model;
[0081] Step S3003, preprocessing the historical network data to obtain a sample data set of the indicator;
[0082] Step S3004: training the data training model using the data of the sample data set of the indicator to obtain a prediction model corresponding to the indicator.
[0083] As an example, the user may import the indicator model affecting the network health and the data training model used for data training into the system in advance.
[0084] As an example, after obtaining the indicator model and the data training model, the system can collect historical network data corresponding to each indicator based on the indicator information in the indicator model, and can preprocess the historical network data to obtain a sample data set for each indicator.
[0085] As an example, the sample data sets of each indicator can be input into the corresponding data training model for training to generate the prediction model corresponding to each indicator.
[0086] In an exemplary embodiment, the data training model includes a training data scale parameter, and the preprocessing of the historical network data to obtain a sample data set of the indicator includes:
[0087] Obtaining the time series corresponding to the historical network element data;
[0088] According to the time series, the training data scale parameter and the indicators in the preset indicator model, the historical network data is segmented to obtain a sample data set of the indicator.
[0089] Machine learning itself is a process of classifying and regressing historical data, which is generally based on time series. In the embodiment of the present invention, the sample data set can be extracted according to the time series.
[0090] For example, historical network data can be decomposed into four parts: trend, cycle, period and unstable factors, and these factors can be combined for prediction, so as to emphasize the continuous observation of an area within a certain period of time, extract data-related features, and analyze its change process and development trend, that is, the change trend.
[0091] As an example, the changing trend may include long-term trend change, seasonal cycle change, cyclic change, random change, etc. Among them, long-term trend change refers to the influence of certain basic factors, the data changes over time and shows a certain tendency, it grows or decreases steadily according to certain rules, and the analysis methods that can be used are: moving average method, exponential smoothing method, model fitting method, etc.; seasonal cycle change refers to the influence of factors such as seasonal changes, the sequence changes in a fixed periodic regularity, also known as the business cycle, and the methods that can be used are: seasonal index, etc.; cyclic change refers to the fluctuation change with an irregular period; random change refers to the sequence change caused by many uncertain factors.
[0092] As an example, the data training model may include a training data scale parameter, which may indicate the size of a sample training set. The historical network data may be segmented according to the time series, the training data scale parameter, and the indicators in the preset indicator model to obtain a sample data set of the indicators.
[0093] In an exemplary embodiment, the segmenting of the historical network data according to the time series, the training data scale parameter, and the index in the preset index model to obtain a sample data set of the index includes:
[0094] Cleaning or aggregating the historical network data;
[0095] According to the time series, the training data scale parameter and the indicators in the preset indicator model, the historical network data after data cleaning or aggregation is segmented to obtain a sample data set of the indicator.
[0096] As an example, before segmenting the historical network data, the historical network data may be cleaned and / or aggregated, for example, missing values and abnormal values in the historical network data may be cleaned, and data of a time period may be aggregated into one or more records.
[0097] In an exemplary embodiment, the sample data set includes a sample training set, and the data training model is trained using the data of the sample data set of the indicator to obtain a prediction model corresponding to the indicator, including:
[0098] The data training model is trained using the data of the sample training set of the indicator to obtain a prediction model corresponding to the indicator.
[0099] In the embodiment of the present invention, the sample data set may include a sample training set and a sample test set.
[0100] As an example, the historical network data may be divided into a sample training set and a sample test set according to the time series, the training data scale parameter, and the indicators in the preset indicator model.
[0101] As an example, the sample training sets of each indicator after preprocessing can be respectively input into the corresponding data training model for training to obtain the prediction model corresponding to each indicator.
[0102] In an exemplary embodiment, the sample data set further includes a sample test set. After the data training model is trained using the sample training set of the indicator to obtain a prediction model corresponding to the indicator, the following further includes:
[0103] Input the data of the sample test set of the indicator into the prediction model corresponding to the indicator, and obtain the indicator prediction data corresponding to the data of the sample test set output by the prediction model; determine the prediction error based on the indicator prediction data and the data of the sample test set corresponding to the indicator prediction data; when the prediction error is less than or equal to the preset error threshold, determine that the prediction model training is completed.
[0104] As an example, the sample test sets of each indicator can be respectively input into the prediction models corresponding to each indicator to obtain the prediction data of each indicator output by each prediction model. The prediction error is calculated based on the prediction data of each indicator and the sample test sets corresponding to the prediction data of each indicator. When the prediction error is less than or equal to the preset error threshold, it is determined that the prediction model training is completed.
[0105] The following is further explained through an example:
[0106] Example 1:
[0107] Taking the simple moving average forecasting method as an example, let the time series be y t , taking the number of moving average items as n, the calculation formula for the forecast value of the t+1 period is:
[0108]
[0109] Among them, y t represents the actual value of period t, represents the moving average of the tth period, Represents the predicted value for the t+1th period.
[0110] The standard error of prediction is:
[0111]
[0112] Where N is the time series {y t The number of raw data contained in}. The value of the number of items n can be determined according to the characteristics of the time series. If n is too large, the sensitivity of the moving average will be reduced, affecting the accuracy of the prediction; if n is too small, the moving average will be easily affected by random changes and it will be difficult to reflect the actual trend. As an example, for a time series without seasonal and cyclical changes, n can be a relatively large number.
[0113] In an exemplary embodiment, when the prediction error is greater than a preset error threshold, the training data scale parameter is adjusted; the historical network data is segmented according to the time series, the adjusted training data scale parameter and the indicators in the preset indicator model, and the sample data set of the indicators is regenerated; and data training is performed again based on the regenerated sample data set of the indicators to generate a new prediction model.
[0114] As an example, if the prediction error is greater than the preset error threshold, the training data scale parameter can be adjusted, and the historical network data can be segmented according to the time series, the adjusted training data scale parameter and the indicators in the preset indicator model, and the sample data set of each indicator can be regenerated. Based on the regenerated sample data set of the indicator, data training is performed again to generate a new prediction model.
[0115] In the embodiment of the present invention, historical data of the index for network health is collected, and the prediction model corresponding to the index is obtained; the historical data of the index is input into the prediction model corresponding to the index to output the prediction data of the prediction model for the index; and the predicted value of the health of the network is determined according to the predicted data of the index. Therefore, the change trend of the network can be judged according to the predicted value of the health of the network, and an early warning can be issued for abnormal situations, so that network maintenance personnel can analyze and handle them as soon as possible to reduce the occurrence of faults, which solves the problem that traditional responsive fault handling cannot actively warn of network performance risks and move user complaints forward.
[0116] In order to facilitate the understanding of the present application scheme, some examples are given below to further illustrate:
[0117] Example 2:
[0118] Figure 5 It is a flow chart of a data sequence system, refer to Figure 5 , the execution process is as follows:
[0119] 1. The user imports the indicator model and data training model in the simulation environment;
[0120] 2. The data collection module collects data from the network element according to the indicator model, or the network element provides a static data import function to import static data packets;
[0121] 3. After data collection, abnormal data can be cleaned, or the collected data can be aggregated as required;
[0122] 4. Split the cleaned and aggregated data according to the time series and the train_size parameter configured in the model to generate training sets and test sets for each indicator;
[0123] 5. Use the data model training model to train the data of each indicator training set to generate a prediction model;
[0124] 6. Use test data to verify the indicator prediction model to verify the reliability and execution efficiency of the prediction model;
[0125] 7. If the prediction model fails to pass the verification, adjust the train_size to regenerate the training set and test set, and perform data training again to generate the prediction model until the verification passes;
[0126] 8. If the verification is successful, export the prediction model for each indicator.
[0127] Example 3:
[0128] Figure 6This is a process sequence diagram of a health prediction system. Figure 6 , the execution process is as follows:
[0129] 1. Import the model into the existing network system, including the indicator models corresponding to the health level and the prediction models corresponding to the indicators;
[0130] 2. Collect the existing network indicator data according to the indicator model;
[0131] 3. Input the collected data of the current network indicators into the corresponding prediction model, and output the predicted value of each indicator by each prediction model;
[0132] 4. Based on the predicted values of each indicator, generate the short-term change trend of each indicator in the future;
[0133] 5. Aggregate the predicted values of various indicators into network health for external display;
[0134] 6. Based on network health, assess future network health risk status and proactively issue external reminders.
[0135] The above-mentioned Example 1 and Example 2 analyze the trends of the historical data of the network elements and predict the network risks in the future by combining the current real-time data, so as to predict the health change trend of the overall network and individual network elements, and thus issue early warnings for abnormal situations so that network maintenance personnel can analyze and handle them as soon as possible to reduce the occurrence of failures.
[0136] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for enabling a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present invention.
[0137] In this embodiment, a network health prediction device is also provided, which is used to implement the above embodiments and preferred implementations, and will not be repeated here. As used below, the term "module" can implement a combination of software and / or hardware for a predetermined function. Although the device described in the following embodiments is preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.
[0138] Figure 7 is a structural block diagram of a network health prediction device according to an embodiment of the present invention. Figure 7 As shown, the device comprises:
[0139] The first collection module 701 is used to collect historical data of indicators for network health and obtain a prediction model corresponding to the indicator;
[0140] An input module 702, used to input the historical data of the indicator into the prediction model corresponding to the indicator, so as to output the prediction data of the prediction model for the indicator;
[0141] The determination module 703 is used to determine the predicted value of the health of the network according to the predicted data of the indicator.
[0142] In an exemplary embodiment, the network includes at least one network element type, each of the network element type includes at least one network element, each of the network element includes at least one data dimension, and each of the data dimensions includes at least one indicator; the determination module 703 includes:
[0143] A first acquisition submodule is used to acquire the weight corresponding to the indicator, the weight corresponding to the data dimension, and the weight corresponding to the network element type;
[0144] A first determination submodule, configured to determine a predicted health value of the data dimension corresponding to the indicator according to the predicted data of the indicator and the weight corresponding to the indicator;
[0145] A second determination submodule, configured to determine a predicted health value of the network element corresponding to the data dimension according to the predicted health value of the data dimension and a weight corresponding to the data dimension;
[0146] A third determination submodule, configured to determine a predicted health value of the network element type according to the predicted health value of the network element;
[0147] The fourth determination submodule is used to determine the predicted value of the health of the network according to the predicted value of the health of the network element type and the corresponding weight of the network element type.
[0148] In an exemplary embodiment, it also includes:
[0149] A generation module is used to generate a predicted change trend of the indicator according to the predicted data of the indicator after outputting the predicted data of the prediction model for the indicator.
[0150] In an exemplary embodiment, it also includes:
[0151] An acquisition module, used to acquire a preset indicator model and a data training model for network health before collecting historical data of indicators for network health; the preset indicator model includes at least one indicator;
[0152] A second collection module, used for collecting historical network element data of the network according to the preset indicator model;
[0153] A preprocessing module, used to preprocess the historical network data to obtain a sample data set of the indicator;
[0154] The training module is used to train the data training model using the data of the sample data set of the indicator to obtain a prediction model corresponding to the indicator.
[0155] In an exemplary embodiment, the data training model includes a training data scale parameter, and the preprocessing module includes:
[0156] The second acquisition submodule is used to acquire the time series corresponding to the historical network element data;
[0157] The segmentation submodule is used to segment the historical network data according to the time series, the training data scale parameter and the indicators in the preset indicator model to obtain a sample data set of the indicators.
[0158] In an exemplary embodiment, the cutting submodule comprises:
[0159] A data processing unit, used for cleaning or aggregating the historical network data;
[0160] The segmentation unit is used to segment the historical network data after data cleaning or aggregation according to the time series, the training data scale parameter and the indicators in the preset indicator model to obtain a sample data set of the indicator.
[0161] In an exemplary embodiment, the sample data set includes a sample training set, and the training module includes:
[0162] The training submodule is used to train the data training model using the data of the sample training set of the indicator to obtain a prediction model corresponding to the indicator.
[0163] In an exemplary embodiment, the sample data set further includes a sample test set, and the apparatus further includes:
[0164] An input submodule, for training the data training model through the sample training set of the indicator to obtain the prediction model corresponding to the indicator, and then inputting the data of the sample test set of the indicator into the prediction model corresponding to the indicator to obtain the indicator prediction data corresponding to the data of the sample test set output by the prediction model;
[0165] A fifth determination submodule, configured to determine a prediction error according to the indicator prediction data and data of the sample test set corresponding to the indicator prediction data;
[0166] The sixth determination submodule is used to determine that the prediction model training is completed when the prediction error is less than or equal to a preset error threshold.
[0167] In an exemplary embodiment, it also includes:
[0168] An adjustment submodule, configured to adjust the training data scale parameter when the prediction error is greater than a preset error threshold;
[0169] The regeneration submodule is used to segment the historical network data according to the time series, the adjusted training data scale parameters and the indicators in the preset indicator model, regenerate the sample data set of the indicators, and perform data training again based on the regenerated sample data set of the indicators to generate a new prediction model.
[0170] It should be noted that the above modules can be implemented by software or hardware. For the latter, it can be implemented in the following ways, but not limited to: the above modules are all located in the same processor; or the above modules are located in different processors in any combination.
[0171] An embodiment of the present invention further provides a computer-readable storage medium, in which a computer program is stored, wherein the computer program is configured to execute the steps of any of the above method embodiments when running.
[0172] In an exemplary embodiment, the computer-readable storage medium may include, but is not limited to, various media that can store computer programs, such as a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk or an optical disk.
[0173] An embodiment of the present invention further provides an electronic device, including a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.
[0174] In an exemplary embodiment, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.
[0175] For specific examples in this embodiment, reference may be made to the examples described in the above embodiments and exemplary implementation modes, and this embodiment will not be described in detail herein.
[0176] Obviously, those skilled in the art should understand that the above modules or steps of the present invention can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order than here, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.
[0177] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the principle of the present invention shall be included in the protection scope of the present invention.
Claims
1. A method for predicting network health, characterized in that: include: Collect historical data on indicators of network health and obtain prediction models corresponding to the indicators; Inputting the historical data of the indicator into the prediction model corresponding to the indicator to output the prediction data of the prediction model for the indicator; A predicted value of the health of the network is determined based on the predicted data of the indicator.
2. The method according to claim 1, characterized in that: The network includes at least one network element type, each of the network element types includes at least one network element, each of the network elements includes at least one data dimension, and each of the data dimensions includes at least one of the indicators; and determining the predicted value of the health of the network according to the predicted data of the indicators includes: Obtain the weight corresponding to the indicator, the weight corresponding to the data dimension, and the weight corresponding to the network element type; Determine the predicted health value of the data dimension corresponding to the indicator according to the predicted data of the indicator and the weight corresponding to the indicator; Determining a predicted health value of the network element corresponding to the data dimension according to the predicted health value of the data dimension and a weight corresponding to the data dimension; Determining a predicted health value of the network element type according to the predicted health value of the network element; The predicted value of the health of the network is determined according to the predicted value of the health of the network element type and the corresponding weight of the network element type.
3. The method according to claim 1, characterized in that After outputting the prediction data of the prediction model for the indicator, the method further includes: Based on the predicted data of the indicator, a predicted change trend of the indicator is generated.
4. The method according to claim 1, characterized in that: Before collecting historical data for network health indicators, also include: Obtaining a preset indicator model and a data training model for network health; the preset indicator model includes at least one indicator; Collecting historical network element data of the network according to the preset indicator model; Preprocessing the historical network data to obtain a sample data set of the indicator; The data training model is trained using the data of the sample data set of the indicator to obtain a prediction model corresponding to the indicator.
5. The method according to claim 4, characterized in that The data training model includes a training data scale parameter, and the preprocessing of the historical network data to obtain a sample data set of the indicator includes: Obtaining the time series corresponding to the historical network element data; According to the time series, the training data scale parameter and the indicators in the preset indicator model, the historical network data is segmented to obtain a sample data set of the indicator.
6. The method according to claim 5, characterized in that The segmenting of the historical network data according to the time series, the training data scale parameter and the index in the preset index model to obtain a sample data set of the index includes: Performing data cleaning or aggregation on the historical network data; According to the time series, the training data scale parameter and the indicators in the preset indicator model, the historical network data after data cleaning or aggregation is segmented to obtain a sample data set of the indicator.
7. The method according to claim 6, characterized in that The sample data set includes a sample training set, and the data training model is trained using the data of the sample data set of the indicator to obtain a prediction model corresponding to the indicator, including: The data training model is trained using the data of the sample training set of the indicator to obtain a prediction model corresponding to the indicator.
8. The method according to claim 7, characterized in that The sample data set also includes a sample test set. After the data training model is trained by the sample training set of the indicator to obtain the prediction model corresponding to the indicator, the following is further included: Inputting the data of the sample test set of the indicator into the prediction model corresponding to the indicator, and obtaining the indicator prediction data corresponding to the data of the sample test set output by the prediction model; Determining a prediction error based on the indicator prediction data and the data of the sample test set corresponding to the indicator prediction data; When the prediction error is less than or equal to a preset error threshold, it is determined that the prediction model training is completed.
9. The method according to claim 7, characterized in that: Also includes: When the prediction error is greater than a preset error threshold, adjusting the training data scale parameter; According to the time series, the adjusted training data scale parameter and the indicators in the preset indicator model, the historical network data is segmented, the sample data set of the indicator is regenerated, and data training is performed again based on the regenerated sample data set of the indicator to generate a new prediction model.
10. A network health prediction device, characterized in that: include: A first collection module is used to collect historical data of indicators for network health and obtain a prediction model corresponding to the indicator; An input module, used to input the historical data of the indicator into the prediction model corresponding to the indicator, so as to output the prediction data of the prediction model for the indicator; The determination module is used to determine the predicted value of the health of the network according to the predicted data of the indicator.
11. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, wherein the computer program implements the steps of the method described in any one of claims 1 to 7 when executed by a processor.
12. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method described in any one of claims 1 to 5 are implemented.