Method and device for predicting hidden troubles in private line networking

Abstract

The present disclosure provides a private line networking hidden danger prediction method and device, relating to the technical field of artificial intelligence. The method comprises: obtaining communication analysis data for predicting private line networking hidden dangers, the communication analysis data comprising: fault analysis data, degradation analysis data and hidden danger analysis data; inputting the fault analysis data into a trained private line network fault prediction model to obtain a private line fault probability; inputting the degradation analysis data into a trained private line service performance degradation prediction model to obtain a private line degradation probability; if the private line fault probability is greater than a fault threshold and / or the private line degradation probability is higher than a degradation threshold, inputting the hidden danger analysis data into a trained private line networking hidden danger real-time prediction model to obtain a private line networking hidden danger probability. Thus, using artificial intelligence models can predict private line network fault probability, private line service performance degradation probability and private line networking hidden danger probability, achieving real-time, comprehensive and rapid prediction of private line networking hidden dangers, and greatly reducing labor costs.

Description

Technical Field

[0001] This disclosure relates to the field of artificial intelligence technology, and in particular to a method and apparatus for predicting potential risks in dedicated line networking. Background Technology

[0002] Network vulnerability prediction technology is a network security technology designed to identify, predict, and discover potential risks and vulnerabilities in a network. Common methods used in network vulnerability prediction include expert rules, fault trees, and simulation.

[0003] Among these methods, the following approaches are employed: Expert rules are used: By defining network topology types and potential risks for each type, an expert rule system is used for matching and prediction; Fault trees are used: By analyzing network topology and potential device failures, fault trees are used for prediction; and Simulation is used: By simulating the operation of network topology and devices, potential risks are observed and predicted.

[0004] However, network vulnerability prediction technologies such as expert rules, fault trees, and simulations primarily focus on predicting and analyzing static network structural vulnerabilities, making it difficult to accurately reflect the dynamic impact of the actual network environment and various uncertainties on network structural vulnerabilities. These technologies require significant manpower and time for analysis and maintenance, and struggle to cover all vulnerability types and scenarios, which is a problem that urgently needs to be addressed. Summary of the Invention

[0005] This disclosure provides a method and device for predicting potential risks in dedicated line networking.

[0006] In a first aspect, this disclosure provides a method for predicting hidden dangers in leased line networking, comprising: acquiring communication analysis data for predicting hidden dangers in leased line networking, wherein the communication analysis data includes: fault analysis data, degradation analysis data, and hidden danger analysis data; inputting the fault analysis data into a trained leased line network fault prediction model to obtain the leased line fault probability; inputting the degradation analysis data into a trained leased line service performance degradation prediction model to obtain the leased line degradation probability; if the leased line fault probability is greater than a fault threshold, and / or the leased line degradation probability is higher than a degradation threshold, inputting the hidden danger analysis data into a trained real-time prediction model for hidden dangers in leased line networking to obtain the leased line networking hidden danger probability.

[0007] In some embodiments, acquiring communication analysis data for predicting potential risks in leased line networking includes: acquiring real-time communication data of the leased line; acquiring analysis data for predicting potential risks in leased line networking from the real-time communication data of the leased line; and performing one-hot encoding conversion on the analysis data to generate communication analysis data.

[0008] In some embodiments, the above method further includes: obtaining a fault analysis sample dataset; training a leased network fault prediction model based on the fault analysis sample dataset to generate a trained leased network fault prediction model.

[0009] In some embodiments, obtaining a fault analysis sample dataset includes: obtaining first leased line data, wherein the first leased line data includes leased line basic data, leased line historical performance data, leased line historical fault data, leased line alarm data, and leased line network element device log data; determining fault sample data from the first leased line data; performing one-hot encoding conversion on the fault sample data; padding missing values ​​in the fault sample data after one-hot encoding conversion with zeros, and deduplicating duplicate data to obtain a fault analysis sample dataset.

[0010] In some embodiments, the above method further includes: obtaining a degradation analysis sample dataset; training a dedicated line service performance degradation prediction model based on the degradation analysis sample dataset to generate a trained dedicated line service performance degradation prediction model.

[0011] In some embodiments, obtaining a degradation analysis sample dataset includes: obtaining second leased line data, wherein the second leased line data includes transmission leased line performance data, Internet leased line performance data, and voice leased line performance data; determining degradation sample data from the second leased line data; performing one-hot encoding conversion on the degradation sample data; padding missing values ​​in the degradation sample data after one-hot encoding conversion with zeros; and deduplicating duplicate data to obtain a degradation analysis sample dataset.

[0012] In some embodiments, the above method further includes: obtaining a hidden danger analysis sample dataset; training a real-time prediction model for hidden dangers in dedicated line networking based on the hidden danger analysis sample dataset, and generating a trained real-time prediction model for hidden dangers in dedicated line networking.

[0013] In some embodiments, obtaining a hazard analysis sample dataset includes: obtaining third leased line data, wherein the third leased line data includes network model data corresponding to the leased line service protection level, leased line network management network topology data, and routing segment quality inspection data; determining hazard sample data from the third leased line data; performing one-hot encoding conversion on the hazard sample data; padding missing values ​​in the hazard sample data after one-hot encoding conversion with zeros, and deduplicating duplicate data to obtain a hazard analysis sample dataset.

[0014] Secondly, this disclosure provides a leased line network vulnerability prediction device, comprising: a data acquisition unit for acquiring communication analysis data for predicting leased line network vulnerabilities, wherein the communication analysis data includes: fault analysis data, degradation analysis data, and vulnerability analysis data; a first processing unit for inputting the fault analysis data into a trained leased line network fault prediction model to obtain the leased line fault probability; a second processing unit for inputting the degradation analysis data into a trained leased line service performance degradation prediction model to obtain the leased line degradation probability; and a third processing unit for inputting the vulnerability analysis data into a trained leased line network vulnerability real-time prediction model to obtain the leased line network vulnerability probability if the leased line fault probability is greater than a fault threshold and / or the leased line degradation probability is higher than a degradation threshold.

[0015] Thirdly, this disclosure provides an electronic device, including: a processor, and a memory communicatively connected to the processor; the memory stores computer execution instructions; the processor executes the computer execution instructions stored in the memory to implement the leased line networking vulnerability prediction method disclosed in the embodiments of this disclosure.

[0016] Fourthly, this disclosure provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the leased line networking vulnerability prediction method disclosed in the embodiments of this disclosure.

[0017] Fifthly, this disclosure provides a computer program product, including a computer program that, when executed by a processor, implements the leased line networking vulnerability prediction method disclosed in the embodiments of this disclosure.

[0018] The technical solutions provided by the embodiments of this disclosure have at least the following beneficial effects:

[0019] In this embodiment of the disclosure, artificial intelligence models can be used to assess the status of leased lines, predict the probability of leased line network failures, the probability of leased line service performance degradation, and the probability of leased line networking risks, thereby achieving real-time, comprehensive, and rapid prediction of leased line networking risks and significantly reducing labor costs. Attached Figure Description

[0020] The accompanying drawings, which are incorporated in and form a part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure.

[0021] Figure 1 This is a flowchart illustrating a method for predicting potential risks in leased line networking according to an exemplary embodiment;

[0022] Figure 2 This is a schematic diagram illustrating a training method for a fault prediction model of a leased line network according to an exemplary embodiment;

[0023] Figure 3 This is a schematic diagram illustrating a training method for a dedicated line service performance degradation prediction model according to an exemplary embodiment.

[0024] Figure 4 This is a schematic diagram illustrating a training method for a real-time prediction model of potential hazards in dedicated line networking, according to an exemplary embodiment.

[0025] Figure 5 This is a flowchart illustrating a method for predicting potential risks in government and enterprise dedicated line networking, according to an exemplary embodiment.

[0026] Figure 6 This is a structural diagram illustrating a dedicated line network hidden danger prediction device according to an exemplary embodiment;

[0027] Figure 7 This is a structural diagram illustrating a dedicated line network hidden danger prediction device according to another exemplary embodiment;

[0028] Figure 8 This is a structural diagram illustrating a dedicated line network hidden danger prediction device according to yet another exemplary embodiment;

[0029] Figure 9 This is a structural diagram illustrating a dedicated line network hidden danger prediction device according to yet another exemplary embodiment;

[0030] Figure 10 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this disclosure.

[0031] The accompanying drawings have illustrated specific embodiments of this disclosure, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concepts of this disclosure to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0032] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this disclosure. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this disclosure as detailed in the appended claims.

[0033] It should be noted that the information (including but not limited to user device information, user personal information, etc.), data (including but not limited to data used for analysis, stored data, displayed data, etc.) and signals involved in the embodiments of this disclosure are all authorized by the user or fully authorized by all parties, and the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0034] The technical solutions of this disclosure and how they solve the aforementioned technical problems will be described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of this disclosure will now be described with reference to the accompanying drawings.

[0035] Among related technologies, network vulnerability prediction techniques such as expert rules, fault trees, and simulations mostly focus on predicting and analyzing static network structure vulnerabilities. They struggle to accurately reflect the dynamic impact of the actual network environment and various uncertainties on network structure vulnerabilities. These techniques require significant manpower and time for analysis and maintenance, and are unlikely to cover all vulnerability types and scenarios.

[0036] Based on this, this disclosure provides a method and apparatus for predicting potential risks in leased line networking. The method includes: acquiring communication analysis data for predicting potential risks in leased line networking, wherein the communication analysis data includes: fault analysis data, degradation analysis data, and risk analysis data; inputting the fault analysis data into a trained leased line network fault prediction model to obtain the leased line fault probability; inputting the degradation analysis data into a trained leased line service performance degradation prediction model to obtain the leased line degradation probability; if the leased line fault probability is greater than a fault threshold, and / or the leased line degradation probability is higher than a degradation threshold, inputting the risk analysis data into a trained real-time prediction model for potential risks in leased line networking to obtain the probability of potential risks in leased line networking. Therefore, using an artificial intelligence model, the leased line status can be evaluated, the leased line network fault probability, the leased line service performance degradation probability, and the probability of potential risks in leased line networking can be predicted, achieving real-time, comprehensive, and rapid prediction of potential risks in leased line networking, significantly reducing labor costs.

[0037] Figure 1 This is a flowchart illustrating a method for predicting potential risks in dedicated line networking according to an exemplary embodiment.

[0038] like Figure 1 As shown, the method for predicting potential risks in leased line networking provided in this embodiment includes, but is not limited to, the following steps:

[0039] S1, acquire communication analysis data for predicting potential risks in leased line networking.

[0040] The communication analysis data includes: fault analysis data, degradation analysis data, and potential hazard analysis data.

[0041] In this embodiment of the disclosure, the method for predicting hidden dangers in leased line networking is executed by a leased line networking hidden danger prediction device, wherein the leased line networking hidden danger prediction device can be an electronic device, such as a mobile phone, computer, tablet, etc.

[0042] In this embodiment of the disclosure, the method for predicting potential risks in dedicated line networking can be executed by installing a specific application. For example, if the electronic device is a mobile phone, a specific application can be installed on the mobile phone to execute the method for predicting potential risks in dedicated line networking.

[0043] Among them, fault analysis data is used to predict the probability of leased line network failures, degradation analysis data is used to predict the probability of leased line service performance degradation, and hidden danger analysis data is used to predict the probability of hidden dangers in leased line networks.

[0044] To address the increasingly demanding requirements of government and enterprise leased lines, it is crucial to build a high-quality end-to-end dedicated support network and provide valuable user service levels and perception. In this embodiment, communication analysis data is acquired to predict potential network vulnerabilities in government and enterprise leased lines. Specifically, fault analysis data is used to predict the probability of network failures in government and enterprise leased lines, degradation analysis data is used to predict the probability of performance degradation in government and enterprise leased line services, and vulnerability analysis data is used to predict the probability of potential vulnerabilities in government and enterprise leased line networks.

[0045] In some embodiments, S1, acquiring communication analysis data for predicting potential risks in leased line networking includes: acquiring real-time communication data of the leased line; acquiring analysis data for predicting potential risks in leased line networking from the real-time communication data of the leased line; and performing one-hot encoding conversion on the analysis data to generate communication analysis data.

[0046] In this embodiment of the disclosure, real-time communication data of the leased line is obtained, including at least one of leased line basic data, leased line performance data, leased line fault data, leased line alarm data, and leased line network element equipment log data. Analysis data for predicting potential risks in leased line networking is obtained from at least one of the leased line basic data, leased line performance data, leased line fault data, leased line alarm data, and leased line network element equipment log data. The analysis data is then converted using one-hot encoding to generate fault analysis data.

[0047] The basic data for dedicated lines includes at least one of the following: product number, product type, service protection level, activation time, product status, and customer code.

[0048] The product types include: APN IoT leased lines, SMS / MMS leased lines, internet leased lines, data leased lines, voice leased lines, and cloud network leased lines. Service assurance levels include A, AA, AAA, AAA+, and Standard. Product status includes archived and resource configured.

[0049] In this embodiment of the disclosure, analytical data for predicting potential network vulnerabilities of leased lines is obtained from the leased line basic data. At least one of the following can be obtained: product type, service assurance level, and product status. At least one of the following can be converted into one-hot encoding to generate fault analysis data.

[0050] Understandably, one-hot encoding is a feature encoding method primarily used to convert discrete features into continuous features so that machine learning algorithms can process them better.

[0051] The leased line performance data includes at least one of the following: downlink average bandwidth utilization, downlink peak bandwidth utilization, uplink average bandwidth utilization, uplink peak bandwidth utilization, A-end average latency, A-end packet loss rate, Z-end average latency, and Z-end packet loss rate.

[0052] In this embodiment of the disclosure, analysis data for predicting potential network hazards of leased lines is obtained from leased line performance data. At least one of the following can be obtained: downlink average bandwidth utilization, downlink peak bandwidth utilization, uplink average bandwidth utilization, uplink peak bandwidth utilization, A-end average latency, A-end packet loss rate, Z-end average latency, and Z-end packet loss rate. One-hot encoding conversion is then performed to generate fault analysis data.

[0053] The dedicated line fault data includes at least one of the following: fault work order number, customer name, product number, equipment name, dispatch time, archiving time, fault type, and network element name.

[0054] The fault types include network-related faults, client-related faults, customer-related faults, other faults, and no fault.

[0055] In this embodiment of the disclosure, analysis data for predicting potential problems in leased line networking is obtained from leased line performance data. At least one of the following can be obtained: fault work order number, customer name, product number, equipment name, dispatch time, archiving time, fault type, and network element name. These can be converted into one-hot encoding to generate fault analysis data.

[0056] The dedicated line alarm data includes at least one of the following: alarm fingerprint, alarm occurrence time, alarm title, network element name, alarm location object, active alarm count, and product number.

[0057] In this embodiment of the disclosure, analytical data for predicting potential problems in leased line networking is obtained from leased line performance data. At least one of the following can be obtained: alarm fingerprint, alarm occurrence time, alarm title, network element name, alarm location object, active alarm count, and product number. These data are then converted using one-hot encoding to generate fault analysis data.

[0058] The dedicated line network element equipment log data includes at least one of the following: network element equipment identifier, equipment name, CPU utilization, and board utilization.

[0059] In this embodiment of the disclosure, analysis data for predicting potential problems in leased line networking can be obtained from leased line performance data. At least one of the following can be obtained: network element device identifier, device name, CPU utilization, and board utilization. This data can be converted into one-hot encoding to generate fault analysis data.

[0060] In this embodiment of the disclosure, real-time communication data of the leased line is obtained, including performance data of the transmission leased line, performance data of the Internet leased line, and performance data of the voice leased line. Analysis data for predicting potential risks in leased line networking is obtained from the performance data of the transmission leased line, the Internet leased line, and the voice leased line. The analysis data is then converted by one-hot encoding to generate degradation analysis data.

[0061] The transmission leased line performance data includes at least one of the following: product number, customer code, customer name, data statistics start time, data statistics end time, transmitting bandwidth utilization, receiving bandwidth utilization, transmitting rate, receiving rate, A-end packet loss rate, A-end latency, A-end jitter latency, Z-end packet loss rate, Z-end latency, Z-end jitter latency, A-end transmitting optical power, A-end receiving optical power, A-end CRC error packet count, Z-end transmitting optical power, Z-end receiving optical power, and Z-end CRC error packet count.

[0062] In this embodiment of the disclosure, analytical data for predicting potential network hazards of leased lines is obtained from leased line performance data. This data can include at least one of the following: product number, customer code, customer name, data statistics start time, data statistics end time, transmitting bandwidth utilization, receiving bandwidth utilization, transmitting rate, receiving rate, A-end packet loss rate, A-end latency, A-end jitter latency, Z-end packet loss rate, Z-end latency, Z-end jitter latency, A-end transmitting optical power, A-end receiving optical power, A-end CRC error packet count, Z-end transmitting optical power, Z-end receiving optical power, and Z-end CRC error packet count. These data are then converted using one-hot encoding to generate degradation analysis data.

[0063] The performance data of the dedicated internet line includes at least one of the following: product number, customer code, customer name, data statistics start time, data statistics end time, downlink average bandwidth utilization, downlink peak bandwidth utilization, uplink average bandwidth utilization, uplink peak bandwidth utilization, ONUID, ONU optical power, packet loss rate, maximum latency, minimum latency, average latency, maximum jitter, minimum jitter, average jitter, uplink traffic, and downlink traffic.

[0064] In this embodiment of the disclosure, analytical data for predicting potential network hazards of leased lines is obtained from leased line performance data. This data can include at least one of the following: product number, customer code, customer name, data statistics start time, data statistics end time, downlink average bandwidth utilization, downlink peak bandwidth utilization, uplink average bandwidth utilization, uplink peak bandwidth utilization, ONUID, ONU optical power, packet loss rate, maximum latency, minimum latency, average latency, maximum jitter, minimum jitter, average jitter, uplink traffic, and downlink traffic. These data are then converted using one-hot encoding to generate degradation analysis data.

[0065] The voice leased line performance data includes at least one of the following: product number, customer code, customer name, data statistics start time, data statistics end time, number of calls, number of calls connected, call setup duration, call connection rate, average call setup duration, initial call setup delay, final call setup delay, and final call connection delay.

[0066] In this embodiment of the disclosure, analytical data for predicting potential risks in leased line networking is obtained from leased line performance data. This data can be obtained by performing one-hot encoding conversion on at least one of the following: product number, customer code, customer name, data statistics start time, data statistics end time, number of calls, number of call connections, call setup duration, call connection rate, average call setup duration, initial call setup delay, final call setup delay, and final call connection delay, to generate degradation analysis data.

[0067] In this embodiment of the disclosure, real-time communication data of the leased line is obtained, including network model data corresponding to the leased line service guarantee level, network topology data of the leased line network management system, and routing segment quality inspection data. Analysis data for predicting potential risks in the leased line network is obtained from the network model data corresponding to the leased line service guarantee level, network topology data of the leased line network management system, and routing segment quality inspection data. One-hot encoding conversion is performed on the analysis data to generate potential risk analysis data.

[0068] The network model data corresponding to the dedicated line service guarantee level includes at least one of the following: service guarantee level, product type, and network model requirements.

[0069] The product types include: APN IoT leased lines, SMS / MMS leased lines, internet leased lines, data leased lines, voice leased lines, and cloud network leased lines. Service protection levels are categorized as A, AA, AAA, AAA+, and Standard.

[0070] In this embodiment of the disclosure, analysis data for predicting potential network vulnerabilities of leased lines is obtained from leased line performance data. This can be achieved by obtaining network model data, including at least one of service assurance level, product type, and network model requirements, and performing one-hot encoding conversion to generate vulnerability analysis data.

[0071] The dedicated line network management network topology data includes at least one of the following: node identifier (ID), node type, line ID, and line type.

[0072] The node types include OBD, OLT, ONU, PTN, etc. The line types include optical fiber cables, pigtails, etc.

[0073] In this embodiment of the disclosure, analysis data for predicting potential risks in leased line networking can be obtained from leased line performance data. At least one of the following can be obtained: node ID, node type, line ID, and line type. One-hot encoding conversion can be performed to generate potential risk analysis data.

[0074] The routing segment quality inspection data includes at least one of the following: product number, product type, service assurance level, number of power supplies, number of circuit boards, and number of routing breakpoints.

[0075] In this embodiment of the disclosure, analysis data for predicting potential risks in leased line networking is obtained from leased line performance data. At least one of the following can be obtained: product number, product type, service assurance level, number of power supplies, number of circuit boards, and number of routing breakpoints. This data can be converted into unique hot-swap encoding to generate potential risk analysis data.

[0076] S2, input the fault analysis data into the trained dedicated line network fault prediction model to obtain the dedicated line fault probability.

[0077] In this embodiment of the disclosure, after obtaining the fault analysis data, the fault analysis data is input into the trained dedicated line network fault prediction model to obtain the dedicated line fault probability.

[0078] The trained dedicated line network fault prediction model can be obtained by training the xgboost model.

[0079] In some embodiments, the above method further includes: obtaining a fault analysis sample dataset; training a leased network fault prediction model based on the fault analysis sample dataset to generate a trained leased network fault prediction model.

[0080] In some embodiments, obtaining a fault analysis sample dataset includes: obtaining first leased line data, wherein the first leased line data includes leased line basic data, leased line historical performance data, leased line historical fault data, leased line alarm data, and leased line network element device log data; determining fault sample data from the first leased line data; performing one-hot encoding conversion on the fault sample data; padding missing values ​​in the fault sample data after one-hot encoding conversion with zeros, and deduplicating duplicate data to obtain a fault analysis sample dataset.

[0081] In this embodiment of the disclosure, the first leased line data is first obtained, including: leased line basic data, leased line historical performance data, leased line historical fault data, leased line alarm data and leased line network element equipment log data. As shown in Table 1 below, fault sample data including fault time, fault type, alarm time, service impact, latency, packet loss, etc. are selected and used as input features (i.e. fault sample data) for model training, and preliminary data cleaning is performed.

[0082] Table 1

[0083]

[0084]

[0085] It should be noted that in the column for converting to training data features, fields indicated as "not input into the training model" are not used for model training, fields indicated as "directly input into the model" can be directly used as input data for model training, and fields indicated as "undergo One-hot encoding conversion" are used as input data for model training after One-hot encoding conversion.

[0086] Secondly, the input data (i.e., fault sample data) undergoes data preprocessing. For nominal data, one-hot encoding is performed. One-hot encoding uses an N-bit state register to encode N states, each state having its own independent register bit, and at any given time, only one bit is valid. One-hot encoding of the fault sample data preserves the data's feature information and also expands its features. Missing values ​​are padded with zeros. All data is deduplicated to ensure the uniqueness of the training data, resulting in the fault analysis sample dataset.

[0087] Finally, the fault analysis sample dataset was divided into training, validation, and test sets in proportions of 0.7, 0.2, and 0.1. The XGBoost model was trained using the labels of whether the fault occurred, whether the fault deteriorated, and whether the fault affected the business. The trained dedicated line network fault prediction model was then used to predict the probability of dedicated line network faults.

[0088] For example, such as Figure 2 As shown, taking government and enterprise leased lines as an example, the process begins with sample collection to obtain basic leased line data, historical leased line performance data, historical leased line fault data, leased line alarm data, and leased line network element log data. Data cleaning and preprocessing are then performed, followed by training the XGBoost model to obtain a trained leased line network fault prediction model for predicting the probability of government and enterprise leased line network faults.

[0089] S3. Input the degradation analysis data into the trained dedicated line service performance degradation prediction model to obtain the dedicated line degradation probability.

[0090] In this embodiment of the disclosure, after obtaining the degradation analysis data, the degradation analysis data is input into the trained dedicated line service performance degradation prediction model to obtain the dedicated line failure probability.

[0091] Among them, the well-trained dedicated line service performance degradation prediction model can be obtained by training the xgboost model.

[0092] In some embodiments, the above method further includes: obtaining a degradation analysis sample dataset; training a dedicated line service performance degradation prediction model based on the degradation analysis sample dataset to generate a trained dedicated line service performance degradation prediction model.

[0093] In some embodiments, obtaining a degradation analysis sample dataset includes: obtaining second leased line data, wherein the second leased line data includes transmission leased line performance data, Internet leased line performance data, and voice leased line performance data; determining degradation sample data from the second leased line data; performing one-hot encoding conversion on the degradation sample data; padding missing values ​​in the degradation sample data after one-hot encoding conversion with zeros; and deduplicating duplicate data to obtain a degradation analysis sample dataset.

[0094] In this embodiment of the disclosure, the second leased line data is first obtained, including transmission leased line performance data, Internet leased line performance data and voice leased line performance data. Then, the data is determined to be degraded according to relevant rules. The degradation criteria are as follows: if any of the following indicators is abnormal, it is considered degraded.

[0095] ① An HTTP request success rate of less than 90% is abnormal; ② A TCP request success rate of less than 90% is abnormal; ③ A PING latency of >50ms for key customers is abnormal; ④ A PING packet loss rate of >3% is abnormal; ⑤ A peak uplink instantaneous rate lower than 70% of the average of past peaks is abnormal; ⑥ A peak downlink instantaneous rate lower than 70% of the average of past peaks is abnormal.

[0096] As shown in Table 2 below, select the fields that can be used for model training as input features (i.e., degraded sample data) and perform preliminary data cleaning.

[0097] Table 2

[0098]

[0099]

[0100]

[0101] It should be noted that in the column for converting to training data features, fields indicated as "not input into the training model" are not used for model training, fields indicated as "directly input into the model" can be directly used as input data for model training, and fields indicated as "undergo One-hot encoding conversion" are used as input data for model training after One-hot encoding conversion.

[0102] Secondly, the input data (i.e., the degraded sample data) undergoes data preprocessing. For nominal data, one-hot encoding is performed. One-hot encoding uses an N-bit state register to encode N states, each state having its own independent register bit, and at any given time, only one bit is valid. One-hot encoding preserves the data's feature information and also expands its features. Missing values ​​are padded with zeros. All data is deduplicated to ensure the uniqueness of the training data, resulting in the degraded analysis sample dataset.

[0103] Finally, the degradation analysis sample dataset was divided into training, validation, and test sets in proportions of 0.7, 0.2, and 0.1. The XGBoost model was trained using the degradation status as a label, and the trained dedicated line service performance degradation prediction model was used to predict the probability of dedicated line service performance degradation.

[0104] For example, such as Figure 3 As shown, taking government and enterprise leased lines as an example, the process begins with sample collection to obtain performance data for transmission leased lines, Internet leased lines, and voice leased lines. Data cleaning and preprocessing are then performed, followed by training the XGBoost model to obtain a trained leased line service performance degradation prediction model for predicting the probability of performance degradation of government and enterprise leased line services.

[0105] S4. If the probability of a dedicated line failure is greater than the failure threshold, and / or the probability of a dedicated line degradation is higher than the degradation threshold, the hidden danger analysis data is input into the trained dedicated line networking hidden danger real-time prediction model to obtain the probability of dedicated line networking hidden dangers.

[0106] In this embodiment of the disclosure, if the probability of a dedicated line failure is greater than the failure threshold, and / or the probability of a dedicated line degradation is higher than the degradation threshold, and if the hidden danger analysis data is obtained, the hidden danger analysis data is input into the trained dedicated line networking hidden danger real-time prediction model to obtain the probability of dedicated line networking hidden danger.

[0107] Among them, the trained real-time prediction model for potential hidden dangers in dedicated line networking can be trained using the xgboost model.

[0108] In some embodiments, the above method further includes: obtaining a hidden danger analysis sample dataset; training a real-time prediction model for hidden dangers in dedicated line networking based on the hidden danger analysis sample dataset, and generating a trained real-time prediction model for hidden dangers in dedicated line networking.

[0109] In some embodiments, obtaining a hazard analysis sample dataset includes: obtaining third leased line data, wherein the third leased line data includes network model data corresponding to the leased line service protection level, leased line network management network topology data, and routing segment quality inspection data; determining hazard sample data from the third leased line data; performing one-hot encoding conversion on the hazard sample data; padding missing values ​​in the hazard sample data after one-hot encoding conversion with zeros, and deduplicating duplicate data to obtain a hazard analysis sample dataset.

[0110] In this embodiment of the disclosure, the third leased line data is first obtained, including the network model data corresponding to the leased line service guarantee level, the leased line network management network topology data, and the routing segment quality inspection data. As shown in Table 3 below, the fields that can be used for model training are selected as input features (i.e., hidden danger sample data) for preliminary data cleaning.

[0111] Table 3

[0112]

[0113]

[0114] It should be noted that in the column for converting to training data features, fields indicated as "not input into the training model" are not used for model training, fields indicated as "directly input into the model" can be directly used as input data for model training, and fields indicated as "undergo One-hot encoding conversion" are used as input data for model training after One-hot encoding conversion.

[0115] Secondly, the input data (i.e., the hazard sample data) undergoes data preprocessing. For nominal data, one-hot encoding is performed. One-hot encoding uses an N-bit state register to encode N states, each state having its own independent register bit, and at any given time, only one bit is valid. One-hot encoding preserves the data's feature information and also expands its features. Missing values ​​are padded with zeros. All data is deduplicated to ensure the uniqueness of the training data, resulting in the hazard sample dataset.

[0116] Finally, the potential hazard sample dataset was divided into training, validation, and test sets in proportions of 0.7, 0.2, and 0.1. The presence or absence of potential hazards was used as a label to train the XGBoost model, resulting in a trained real-time prediction model for potential hazards in dedicated line networks, which was then used to predict the probability of potential hazards in dedicated line networks.

[0117] For example, such as Figure 4 As shown, taking government and enterprise leased lines as an example, the process begins with sample collection to obtain network model data, leased line network management network topology data, and routing segment quality inspection data corresponding to the leased line service assurance level. Data cleaning and preprocessing are then performed, followed by training the XGBoost model to obtain a trained real-time prediction model for leased line network vulnerabilities, which is used to predict the probability of network vulnerabilities in government and enterprise leased lines.

[0118] For example, such as Figure 5 As shown, taking a government and enterprise leased line as an example, this embodiment of the invention introduces a method for predicting potential network risks in leased lines. First, real-time data collection of the leased line is performed, as shown in Tables 1, 2, and 3 above. Then, based on the data in Table 1, leased line network fault prediction is performed. This involves calling an API to request the government and enterprise leased line network fault prediction model to obtain the predicted fault probability based on the data in Table 1, and returning the data via the API. Similarly, based on the data in Table 2, leased line service performance degradation prediction is performed. This involves calling an API to request the government and enterprise leased line service performance degradation prediction model to obtain the predicted degradation probability based on the data in Table 2, and returning the data via the API. Specifically, when the predicted fault probability and the predicted degradation probability are both higher than a threshold, a probability prediction of potential network risks in the leased line is performed. Based on the data in Table 3, an API is called to request the government and enterprise leased line network risk real-time prediction model to obtain the predicted probability of potential network risks based on the data in Table 3, and returning the data via the API, thus obtaining the probability of potential network risks in the leased line.

[0119] This disclosure provides a real-time prediction capability for potential network vulnerabilities in government and enterprise private lines based on historical performance indicators of private lines. This capability consists of the following parts: training of a government and enterprise private line network fault prediction model, training of a government and enterprise private line service performance degradation prediction model, training of a real-time prediction model for potential network vulnerabilities in government and enterprise private lines, and real-time prediction of potential network vulnerabilities in government and enterprise private lines.

[0120] 1. Training a fault prediction model for government and enterprise leased network: This process integrates leased line fault data, leased line performance data, and leased line network element log data, in addition to using leased line alarm data. These four types of data are preprocessed and input into an intelligent model, which is then trained using an improved XGBoost method. This comprehensively assesses the leased line status and predicts the probability of leased line faults. This fills the gap in predicting the impact of leased line performance, fault conditions, and network element equipment status on leased line faults, representing a unique technology in the proposal.

[0121] 2. Training of the performance degradation prediction model for government and enterprise leased line services: This process involves cleaning and preprocessing performance data from transmission leased lines, internet leased lines, and voice leased lines. Based on expert experience rules, the data is categorized as degraded. An improved XGBoost algorithm is then used to train the performance degradation prediction model for government and enterprise leased line services to predict the probability of performance degradation. This process extracts relevant data features from various leased lines, obtains multiple pieces of information about the leased line network, and improves the accuracy of performance degradation prediction.

[0122] 3. Training a real-time prediction model for potential risks in government and enterprise leased line networks: This process utilizes both static elements of the leased line network, such as leased line network model data, leased line network management topology data, and routing segment quality inspection data, and dynamic elements, such as potential risks and trend changes caused by leased line faults and performance degradation, using methods like government and enterprise leased line network fault prediction and government and enterprise leased line service performance degradation prediction. Based on an improved XGBoost algorithm, a real-time prediction model for potential risks in government and enterprise leased line networks is trained to predict the probability of such risks. This unique technological innovation of the proposal lies in its integration of both static elements of the leased line network and dynamic elements such as potential risks and trend changes related to faults and performance degradation, using an intelligent model to predict the probability of potential risks in leased line networks.

[0123] 4. Real-time prediction of network vulnerabilities in government and enterprise leased lines: This process uses a government and enterprise leased line network fault prediction model and a government and enterprise leased line service performance degradation prediction model to predict faults and degradation in government and enterprise leased lines. If the probability of a fault or degradation prediction exceeds a preset threshold, the real-time prediction model for network vulnerabilities in government and enterprise leased line network data is used to predict the probability of network vulnerabilities in real time. By combining three XGBoost models, real-time and rapid prediction of multi-dimensional field data is achieved, significantly reducing labor costs while utilizing a large amount of relevant data features to achieve cost reduction and efficiency improvement.

[0124] Taking government and enterprise leased lines as the research object, this paper trains network fault prediction models and service performance degradation prediction models for government and enterprise leased lines using the XGBoost algorithm to predict the probability of faults and degradation. When the predicted probability of faults or degradations exceeds a preset threshold, a real-time prediction model for network vulnerabilities in government and enterprise leased line network data is used to predict the probability of network vulnerabilities in real time. This method enables network fault prediction for customer service equipment and routing at the service and access layers, and combines leased line network resource data to automatically inspect network vulnerabilities and predict service performance degradation, forming an automatic inspection capability and repair mechanism based on network data capabilities, ensuring high-quality service networks and timely handling of vulnerabilities.

[0125] During implementation, for training the fault prediction model of government and enterprise leased line networks, the preferred approach is to process the leased line basic data, leased line performance data, leased line fault data, leased line alarm data, and leased line network element log data through basic data cleaning, missing value imputation, and preprocessing, and then use the xgboost algorithm to train the fault prediction model of government and enterprise leased line networks.

[0126] For training a performance degradation prediction model for government and enterprise leased line services, the preferred approach is to process the performance data of the transmission leased line, the Internet leased line, and the voice leased line through data cleaning and preprocessing, and then use the XGBoost algorithm to train the performance degradation prediction model for government and enterprise leased line services.

[0127] For training a real-time prediction model for potential network vulnerabilities in government and enterprise private lines, the preferred approach is to clean and preprocess the data, using the failure probability of the government and enterprise private line network fault prediction model, the performance degradation probability of government and enterprise private line services, and the network model data, private line network management network topology data, and routing segment quality inspection data corresponding to the service protection level of the private line. Then, the xgboost algorithm is used to train the real-time prediction model for potential network vulnerabilities in government and enterprise private lines.

[0128] For real-time prediction of potential network risks in government and enterprise private lines, the preferred approach is to use a government and enterprise private line network fault prediction model and a government and enterprise private line service performance degradation prediction model for fault prediction and degradation prediction. When the fault prediction probability or degradation prediction probability is higher than a preset threshold, the real-time prediction model for potential network risks in government and enterprise private line network data is used to predict the probability of potential network risks in real time.

[0129] The beneficial technical effects of the embodiments disclosed herein are as follows: by combining the network resource data of the leased line network, network fault prediction is performed on customer service equipment and routing from the service layer and access layer, forming an automatic inspection capability and repair mechanism based on network data capabilities, ensuring that the service network quality is excellent and that hidden dangers are handled in a timely manner.

[0130] Figure 6 This is a structural diagram of a dedicated line network hidden danger prediction device according to an exemplary embodiment.

[0131] like Figure 6 As shown, the dedicated line network hidden danger prediction device 1 includes: a data acquisition unit 11, a first processing unit 12, a second processing unit 13 and a third processing unit 14.

[0132] The data acquisition unit 11 is used to acquire communication analysis data for predicting potential risks in leased line networking. The communication analysis data includes: fault analysis data, degradation analysis data, and potential risk analysis data.

[0133] The first processing unit 12 is used to input fault analysis data into the trained dedicated line network fault prediction model to obtain the dedicated line fault probability.

[0134] The second processing unit 13 is used to input the degradation analysis data into the trained dedicated line service performance degradation prediction model to obtain the dedicated line degradation probability.

[0135] The third processing unit 14 is used to input the hidden danger analysis data into the trained real-time prediction model of hidden dangers in dedicated line networking if the probability of dedicated line failure is greater than the failure threshold and / or the probability of dedicated line degradation is higher than the degradation threshold, so as to obtain the probability of hidden dangers in dedicated line networking.

[0136] In some embodiments, the data acquisition unit 11 is specifically used to acquire real-time communication data of the leased line; acquire analysis data from the real-time communication data of the leased line for predicting potential risks in the leased line network; and perform one-hot encoding conversion on the analysis data to generate communication analysis data.

[0137] like Figure 7 As shown, in some embodiments, the dedicated line network hidden danger prediction device 1 further includes: a first dataset acquisition unit 15 and a first model training unit 16.

[0138] The first dataset acquisition unit 15 is used to acquire the fault analysis sample dataset.

[0139] The first model training unit 16 is used to train the dedicated line network fault prediction model based on the fault analysis sample dataset, and generate a trained dedicated line network fault prediction model.

[0140] In some embodiments, the first dataset acquisition unit 15 is specifically used to acquire first leased line data, wherein the first leased line data includes leased line basic data, leased line historical performance data, leased line historical fault data, leased line alarm data, and leased line network element device log data; determine fault sample data from the first leased line data; perform one-hot encoding conversion on the fault sample data; fill missing values ​​in the fault sample data after one-hot encoding conversion with zeros, and perform duplicate data deduplication processing to obtain a fault analysis sample dataset.

[0141] like Figure 8 As shown, in some embodiments, the dedicated line network hidden danger prediction device 1 further includes: a second dataset acquisition unit 17 and a second model training unit 18.

[0142] The second dataset acquisition unit 17 is used to acquire the degradation analysis sample dataset.

[0143] The second model training unit 18 is used to train the dedicated line service performance degradation prediction model based on the degradation analysis sample dataset, and generate a trained dedicated line service performance degradation prediction model.

[0144] In some embodiments, the second dataset acquisition unit 17 is specifically used to acquire second leased line data, wherein the second leased line data includes transmission leased line performance data, Internet leased line performance data and voice leased line performance data; determine degraded sample data from the second leased line data; perform one-hot encoding conversion on the degraded sample data; fill missing values ​​in the degraded sample data after one-hot encoding conversion with zeros, and perform duplicate data deduplication processing to obtain a degraded analysis sample dataset.

[0145] like Figure 9 As shown, in some embodiments, the dedicated line network hidden danger prediction device 1 further includes: a third dataset acquisition unit 19 and a third model training unit 20.

[0146] The third dataset acquisition unit 19 is used to acquire the sample dataset for hazard analysis.

[0147] The third model training unit 20 is used to train the real-time prediction model of hidden dangers in dedicated line networking based on the hidden danger analysis sample dataset, and generate the trained real-time prediction model of hidden dangers in dedicated line networking.

[0148] In some embodiments, the third dataset acquisition unit 19 is specifically used to acquire third leased line data, wherein the third leased line data includes network model data corresponding to the leased line service protection level, leased line network management network topology data, and routing segment quality inspection data; determine hidden danger sample data from the third leased line data; perform one-hot encoding conversion on the hidden danger sample data; fill missing values ​​in the hidden danger sample data after one-hot encoding conversion with zeros, and perform duplicate data deduplication processing to obtain a hidden danger analysis sample dataset.

[0149] It should be noted that, in this embodiment of the present disclosure, the relevant description of the dedicated line networking hidden danger prediction device 1 can be found in the relevant description of the dedicated line networking hidden danger prediction method described above, and will not be repeated here.

[0150] In this embodiment of the disclosure, the beneficial effects achieved by the dedicated line network hidden danger prediction device are the same as those achieved by the above-described method, and can be found in the relevant descriptions in the above-described method, which will not be repeated here.

[0151] According to an embodiment of this disclosure, an electronic device is also provided, including: a processor; and a memory for storing processor-executable instructions, wherein the processor is configured to: implement the leased line networking vulnerability prediction method disclosed in the embodiments of this disclosure.

[0152] To implement the above embodiments, this disclosure also proposes a storage medium.

[0153] When the instructions in the storage medium are executed by the processor, the processor is able to execute the leased line networking vulnerability prediction method disclosed in this embodiment.

[0154] To implement the above embodiments, this disclosure also provides a computer program product.

[0155] When the computer program product is executed by the processor of the electronic device, the electronic device is able to execute the leased line networking vulnerability prediction method disclosed in the embodiments of this disclosure.

[0156] Figure 10 This is a structural block diagram of an electronic device according to an exemplary embodiment. Figure 7 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments disclosed herein.

[0157] like Figure 10 As shown, the electronic device 1000 includes a processor 111, which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 112 or a program loaded from memory 116 into random access memory (RAM) 113. The RAM 113 also stores various programs and data required for the operation of the electronic device 1000. The processor 111, ROM 112, and RAM 113 are interconnected via a bus 114. An input / output (I / O) interface 115 is also connected to the bus 114.

[0158] The following components are connected to I / O interface 115: memory 116 including hard disks, etc.; and communication section 117 including network interface cards such as local area network (LAN) cards, modems, etc., communication section 117 performs communication processing via a network such as the Internet; and driver 118 is also connected to I / O interface 115 as needed.

[0159] In particular, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 117. When the computer program is executed by processor 111, it performs the functions defined in the methods of this disclosure.

[0160] In an exemplary embodiment, a storage medium including instructions is also provided, such as a memory including instructions, which can be executed by the processor 111 of the electronic device 1000 to perform the above-described method. Optionally, the storage medium may be a non-transitory computer-readable storage medium, such as a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device.

[0161] In this disclosure, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in connection with an instruction execution system, apparatus, or device. In this disclosure, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, which can transmit, propagate, or transfer a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, wireline, optical fiber, RF, etc., or any suitable combination thereof.

[0162] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This disclosure is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the following claims.

[0163] It should be understood that this disclosure is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this disclosure is limited only by the appended claims.

Claims

1. A method for predicting potential risks in dedicated line networking, characterized in that, include: Acquire communication analysis data for predicting potential risks in leased line networking, wherein the communication analysis data includes: fault analysis data, degradation analysis data, and potential risk analysis data; The fault analysis data is input into the trained dedicated line network fault prediction model to obtain the dedicated line fault probability. The degradation analysis data is input into the trained dedicated line service performance degradation prediction model to obtain the dedicated line degradation probability. If the failure probability of the dedicated line is greater than the failure threshold, and / or the degradation probability of the dedicated line is higher than the degradation threshold, the hidden danger analysis data is input into the trained dedicated line networking hidden danger real-time prediction model to obtain the dedicated line networking hidden danger probability.

2. The method according to claim 1, characterized in that, The acquisition of communication analysis data for predicting potential risks in leased line networking includes: Obtain real-time communication data from the dedicated line; Analytical data for predicting potential risks in dedicated line networking is obtained from the real-time communication data of the dedicated line; The analysis data is converted into communication analysis data by performing one-hot encoding.

3. The method according to claim 1 or 2, characterized in that, The method further includes: Obtain the fault analysis sample dataset; The dedicated line network fault prediction model is trained based on the fault analysis sample dataset to generate the trained dedicated line network fault prediction model.

4. The method according to claim 3, characterized in that, The acquisition of the fault analysis sample dataset includes: Acquire first leased line data, wherein the first leased line data includes leased line basic data, leased line historical performance data, leased line historical fault data, leased line alarm data, and leased line network element device log data; Determine fault sample data from the first dedicated line data; Perform one-hot encoding conversion on the fault sample data; The missing values ​​in the fault sample data after one-hot encoding conversion are padded with zeros, and duplicate data are removed to obtain the fault analysis sample dataset.

5. The method according to claim 1 or 2, characterized in that, The method further includes: Obtain the degradation analysis sample dataset; The dedicated line service performance degradation prediction model is trained based on the degradation analysis sample dataset to generate the trained dedicated line service performance degradation prediction model.

6. The method according to claim 5, characterized in that, The process of obtaining the degradation analysis sample dataset includes: Acquire second leased line data, wherein the second leased line data includes transmission leased line performance data, Internet leased line performance data, and voice leased line performance data; Degraded sample data were determined from the second dedicated line data; The degraded sample data is converted using one-hot encoding. The missing values ​​in the degraded sample data after one-hot encoding conversion are padded with zeros, and duplicate data are removed to obtain the degraded analysis sample dataset.

7. The method according to claim 1 or 2, characterized in that, The method further includes: Obtain the sample dataset for hazard analysis; The real-time prediction model for hidden dangers in dedicated line networking is trained based on the hidden danger analysis sample dataset to generate the trained real-time prediction model for hidden dangers in dedicated line networking.

8. The method according to claim 7, characterized in that, The acquisition of the hazard analysis sample dataset includes: Acquire third leased line data, wherein the third leased line data includes network model data corresponding to the leased line service guarantee level, leased line network management network topology data, and routing segment quality inspection data; Determine potential hazard sample data from the third dedicated line data; The hazard sample data is converted using one-hot encoding. The missing values ​​in the hazard sample data after one-hot encoding conversion are padded with zeros, and duplicate data are removed to obtain the hazard analysis sample dataset.

9. A device for predicting potential risks in dedicated line networking, characterized in that it comprises: The data acquisition unit is used to acquire communication analysis data for predicting potential risks in leased line networking, wherein the communication analysis data includes: fault analysis data, degradation analysis data, and potential risk analysis data; The first processing unit is used to input the fault analysis data into the trained dedicated line network fault prediction model to obtain the dedicated line fault probability. The second processing unit is used to input the degradation analysis data into the trained dedicated line service performance degradation prediction model to obtain the dedicated line degradation probability. The third processing unit is used to input the hidden danger analysis data into the trained real-time prediction model of hidden dangers in dedicated line networking if the probability of failure of the dedicated line is greater than the failure threshold and / or the probability of degradation of the dedicated line is higher than the degradation threshold, so as to obtain the probability of hidden dangers in dedicated line networking.

10. An electronic device, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the method as described in any one of claims 1 to 8.

Images