Broadband fault root cause analysis method and device based on Markov chain algorithm

Abstract

The invention relates to a broadband fault root cause analysis method and device based on a Markov chain algorithm, and belongs to the field of network service support. Comprising the steps of obtaining historical alarm data; preprocessing the historical alarm data; performing correlation analysis on the network equipment characteristics and the network service characteristics of the preprocessed historical alarm data through a Pearson's correlation coefficient; according to a correlation analysis result, generating a root cause analysis rule of the home broadband service through a Markov chain algorithm; acquiring real-time alarm data; matching the real-time alarm data with the root cause analysis rule through a matching algorithm; and outputting the root cause fault of the household broadband according to the matching result. The alarm reason of the household broadband is accurately analyzed, the work order compression ratio is improved, the maintenance efficiency is improved, and the operation and maintenance cost is reduced.

Description

Technical Field

[0001] The present invention belongs to the technical field of network business support, and in particular relates to a broadband fault root cause analysis method and device based on a Markov chain algorithm. Background Art

[0002] The Markov chain algorithm refers to the situation where certain factors of a system are in transition, and the Nth result is only affected by the N-1th result, and is only related to the current state, and has nothing to do with other things. The broadband fault root cause analysis method refers to the use of the Markov chain algorithm to model the transition and association of different alarm states in the broadband fault root cause analysis, so as to identify and predict the potential causes of the fault.

[0003] Analyzing the root causes of home broadband failures can effectively reduce the manual effort of O&M experts to sort out alarm association rules, improve the accuracy of home broadband service failure identification, and minimize the number of home broadband service failure work orders. Fast and accurate fault analysis is of great significance to improving user experience, optimizing O&M costs, ensuring network stability, realizing intelligent O&M, and gaining a stronger market advantage in competition.

[0004] However, the existing broadband fault analysis method is mainly based on the experience of maintenance personnel, and sets preliminary judgment conditions for home broadband faults, but does not achieve refined fault analysis and root cause judgment. With the rapid development of home broadband services, various types of broadband faults emerge in an endless stream, resulting in an excessive number of fault work orders, thereby increasing the maintenance workload, resulting in the inability to quickly and accurately locate the causes of large-scale home broadband failures, and unable to improve maintenance efficiency, reduce operation and maintenance costs, and reduce user experience. Summary of the invention

[0005] In view of the above shortcomings of the prior art, the purpose of the invention is to provide a broadband fault root cause analysis method and device based on Markov chain algorithm, which can improve user experience, optimize operation and maintenance costs, ensure network stability and realize intelligent operation and maintenance, and gain a stronger market advantage in competition.

[0006] In a first aspect of the present invention, a method for analyzing root causes of broadband faults based on a Markov chain algorithm is proposed, the method comprising:

[0007] S1, obtain historical alarm data;

[0008] S2, preprocessing historical alarm data;

[0009] S3, performing correlation analysis on the network device characteristics and network service characteristics of the pre-processed historical alarm data through the Pearson correlation coefficient;

[0010] S4, generating root cause analysis rules for home broadband services through a Markov chain algorithm based on the correlation analysis results;

[0011] S5, obtain real-time alarm data;

[0012] S6, matching the real-time alarm data with the root cause analysis rules through a matching algorithm;

[0013] S7, based on the matching result, outputs the root cause of the home broadband failure.

[0014] Furthermore, the historical alarm data includes: ONU alarm data, PON port alarm data, OLT uplink interruption alarm data, power environment alarm data and BRAS alarm data.

[0015] Furthermore, preprocessing includes data cleaning, multi-dimensional data supplementation and data time series analysis.

[0016] Furthermore, the network device feature is specifically the number of times the physical machine network port is down, and the network service feature is specifically the unreachable destination of the network element link.

[0017] Furthermore, the Pearson correlation coefficient is specifically:

[0018]

[0019] Where X represents the number of physical machine network port down failures, Y represents the number of network element link destination unreachable failures, and ρ X,Y represents the Pearson correlation coefficient between the number of physical machine network port down failures and the number of network element link destination unreachable failures, cov(X,Y) represents the covariance between the number of physical machine network port down failures and the number of network element link destination unreachable failures, σ X Indicates the standard deviation of the number of physical machine network port down failures, σ Y represents the standard deviation of the number of network element link destination unreachable failures, E represents the expected value symbol, μ X Indicates the average number of physical machine network port down failures, μ Y Indicates the average number of NE link destination unreachable failures.

[0020] Furthermore, S4 specifically includes:

[0021] S401, constructing a time series model from historical alarm data, in which different alarm types are regarded as different states of the system, and changes in alarm states represent state transitions;

[0022] S402, simulating the state transition process between various alarm types in the time series model through a Markov chain algorithm;

[0023] S403, calculating the state transition probability between various alarm types according to the historical alarm data;

[0024] S404: Generate a root cause analysis rule according to the state transition frequency.

[0025] In a second aspect of the present invention, a broadband fault root cause analysis device is provided, which is used to implement any broadband fault root cause analysis method based on a Markov chain algorithm according to the first aspect, and the device comprises:

[0026] The first acquisition module is used to acquire historical alarm data;

[0027] The processing module pre-processes the historical alarm data;

[0028] The analysis module uses the Pearson correlation coefficient to perform correlation analysis on the network device characteristics and network service characteristics of the pre-processed historical alarm data;

[0029] A generation module, used to generate root cause analysis rules of home broadband services through a Markov chain algorithm based on the correlation analysis results;

[0030] The second acquisition module is used to obtain real-time alarm data;

[0031] The matching module matches the real-time alarm data with the root cause analysis rules through the matching algorithm;

[0032] The output module is used to output the root cause fault of the home broadband according to the matching result.

[0033] Furthermore, the historical alarm data includes: ONU alarm data, PON port alarm data, OLT uplink interruption alarm data, power environment alarm data and BRAS alarm data.

[0034] Furthermore, preprocessing includes data cleaning, multi-dimensional data supplementation and data time series analysis.

[0035] Furthermore, the network device feature is specifically that the physical machine network port is down, and the network service feature is specifically that the network element link destination is unreachable.

[0036] Furthermore, the Pearson correlation coefficient is specifically:

[0037]

[0038] Where X represents the number of physical machine network port down failures, Y represents the number of network element link destination unreachable failures, and ρ X,Yrepresents the Pearson correlation coefficient between the number of physical machine network port down failures and the number of network element link destination unreachable failures, cov(X,Y) represents the covariance between the number of physical machine network port down failures and the number of network element link destination unreachable failures, σ X Indicates the standard deviation of the number of physical machine network port down failures, σ Y represents the standard deviation of the number of network element link destination unreachable failures, E represents the expected value symbol, μ X Indicates the average number of physical machine network port down failures, μ Y Indicates the average number of NE link destination unreachable failures.

[0039] Furthermore, the generation module is used to:

[0040] The historical alarm data is constructed into a time series model. Different alarm types in the time series model are regarded as different states of the system, and the change of alarm state indicates the state transition.

[0041] Through the Markov chain algorithm, the state transition process between various alarm types in the time series model is simulated;

[0042] Based on historical alarm data, calculate the state transition probability between various alarm types;

[0043] Generate root cause analysis rules based on the state transition frequency.

[0044] A third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program can be loaded and executed by a processor to provide a broadband fault root cause analysis method based on a Markov chain algorithm according to the first aspect.

[0045] The beneficial effects of the present invention are as follows:

[0046] In an embodiment of the present invention, the acquired historical alarm data is preprocessed, and then the Pearson correlation coefficient is used to perform correlation analysis on the network device characteristics and network service characteristics of the preprocessed historical alarm data. Based on the correlation analysis results, the root cause analysis rules of the home broadband service are generated through the Markov chain algorithm. Finally, the real-time alarm data is matched with the root cause analysis rules through the matching algorithm to output the root cause failure of the home broadband. The method automatically learns from the massive alarm data of the home broadband by using the Markov chain machine learning algorithm, continuously refines and summarizes the RCA rules related to the home broadband, performs alarm compression across network elements through the RCA rules, and performs root cause analysis across network elements, accurately analyzes the alarm causes of the home broadband, improves the work order compression ratio, improves maintenance efficiency, reduces operation and maintenance costs, reduces maintenance workload, and enhances user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The accompanying drawings are only used to illustrate specific embodiments and are not considered to limit the present invention. In the entire drawings, the same reference symbols represent the same components. Obviously, the drawings described below are only some embodiments recorded in the embodiments of the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings.

[0048] Figure 1 It is a flow chart of a broadband fault root cause analysis method based on a Markov chain algorithm provided by an embodiment of the present invention;

[0049] Figure 2 This is a schematic diagram of a home bandwidth topology provided by an embodiment of the present invention;

[0050] Figure 3 It is a structural schematic diagram of a broadband fault root cause analysis method based on a Markov chain algorithm provided by an embodiment of the present invention;

[0051] Figure 4 This is a schematic diagram of a root cause alarm presentation method provided by an embodiment of the present invention;

[0052] Figure 5 It is a schematic diagram of a derivative alarm presentation method provided by an embodiment of the present invention;

[0053] Figure 6 It is a structural schematic diagram of a broadband fault root cause analysis device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0054] In order to enable those skilled in the art to better understand the technical solutions in the embodiments of the present invention, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. It should be understood that these descriptions are only exemplary and are not intended to limit the scope of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in the field without making creative work should fall within the scope of protection of the present invention.

[0055] Furthermore, in the following description, descriptions of well-known structures and technologies are omitted to avoid unnecessarily obscuring the concepts disclosed in the present invention.

[0056] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the orientation or position relationship indicated by the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inside", "outside", etc. is based on the orientation or position relationship shown in the drawings, which is only for the convenience of describing the present invention and simplifying the description, and does not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation of the present invention. In addition, the terms "first", "second", and "third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance. The terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be a connection between the two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0057] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Instead, they are merely examples of methods and systems consistent with some aspects of the present invention as detailed in the appended claims.

[0058] The present invention proposes a broadband fault root cause analysis method and device based on a Markov chain algorithm, in order to solve the problems that the existing broadband fault analysis method is mainly based on the experience of maintenance personnel, sets preliminary judgment conditions for home broadband faults, fails to achieve refined fault analysis and root cause judgment, resulting in the inability to quickly and accurately locate the causes of large-scale home broadband faults, unable to improve maintenance efficiency, reduce operation and maintenance costs, and reduce user experience.

[0059] Method Embodiment

[0060] Reference Figure 1 , shows a flow chart of a broadband fault root cause analysis method based on a Markov chain algorithm provided in an embodiment of the present invention.

[0061] An embodiment of the present invention provides a broadband fault root cause analysis method based on a Markov chain algorithm, the method comprising:

[0062] Specifically, the method includes steps S1 to S7.

[0063] S1, obtain historical alarm data.

[0064] Among them, historical alarm data refers to various alarm information recorded by the network system during its past operation. This information contains data on failures or abnormal events that occurred at different times, in different network devices, and in different business scenarios. By collecting and summarizing a large amount of historical alarm data, we can establish a rich data accumulation and form a comprehensive understanding of broadband network failures.

[0065] In a possible implementation manner, the historical alarm data includes: ONU alarm data, PON port alarm data, OLT uplink interruption alarm data, power environment alarm data and BRAS alarm data.

[0066] Among them, ONU alarm data refers to the alarm information that occurs during the operation of the optical network unit (ONU) device; PON port alarm data refers to the alarm data of the optical distribution network (PON) interface; OLT uplink interruption alarm data refers to the uplink interruption alarm information data generated when the OLT uplink (the link that transmits data to the higher-layer network) is interrupted or abnormal; power environment alarm data refers to alarm data related to the equipment operating environment, including power supply, battery temperature, humidity, computer room environment and other data; BRAS alarm data includes information data such as equipment overload, processing abnormalities or connection interruptions.

[0067] Reference Figure 2 , showing a schematic diagram of a home bandwidth topology provided by an embodiment of the present invention.

[0068] Figure 2 In the figure, the OLT in the aggregation room refers to the optical line terminal located in the central room. It is the management core of the entire optical network. As the interface between the service provider and the end user, it sends and receives user data through the optical fiber network. The optical fiber supports wireless transmission through the base station to connect the OLT to the remote location. The optical signal from the OLT is divided into multiple signals through the splitter and distributed to multiple end users for connection. ONU refers to the optical network unit. The ONU is located at the user end and converts the received optical signal into an electrical signal for use by user devices (such as TVs, computers and phones). Among them, ONU supports various services, including IPTV, broadband Internet access and voice services.

[0069] Reference Figure 3 , shows a structural schematic diagram of a broadband fault root cause analysis method based on a Markov chain algorithm provided in an embodiment of the present invention.

[0070] Figure 3In the process, the alarm data subscription service first subscribes to the historical alarm data of the entire network in the past month and injects it into the mining platform. The mining platform performs data preprocessing, feature engineering and machine learning on the alarm samples to generate RCA rules for home broadband and save the rules to the rule base. The production platform obtains real-time alarms through the alarm data subscription service, monitors the alarms, matches RCA rules from the rule base, and generates root alarms for home broadband services through the rules to achieve root cause analysis of faults. RCA (Root Cause Analysis) rules are a structured problem-solving method that aims to gradually find the root cause of the problem and solve it, rather than just focusing on the representation of the problem. This method includes steps such as determining and analyzing the cause of the problem, finding a solution to the problem, and formulating preventive measures to form a systematic problem-solving process.

[0071] It should be noted that the main function of the alarm data subscription module is to subscribe to relevant alarms from the fault management system, including: on the one hand, subscribing to and synchronizing the full amount of historical alarms in the past month and injecting them into the mining platform; on the other hand, subscribing to and obtaining real-time alarms in real time, and transferring real-time alarms to the production platform.

[0072] The mining platform is a big data processing platform based on machine learning technology. The platform combines the injected alarm information sample data with the resource data of the energy regulation system to supplement and enrich it in multiple dimensions, cleans the junk alarm data, and refines the alarm data through conversion and extraction to generate detailed tables of different dimensions, such as manufacturer details, model details, bandwidth details, city details, circuit details, temperature details, CPU details, memory details, CRC details, port status details, board status details, traffic details, etc. Through feature capture, the alarms of each dimension are strongly correlated and calculated to generate an alarm theme library. Through time feature analysis, the Markov chain machine learning algorithm is used to generate RCA rules, and the rules are stored in the rule library.

[0073] The production platform obtains alarms in the existing network environment in real time through the alarm subscription service, and generates corresponding root alarms based on the RCA rules in the model and matching rule library to determine the root cause network elements and faults of home broadband. The RCA rule library mainly stores generated and trained rules.

[0074] S2, preprocessing the historical alarm data.

[0075] Among them, preprocessing refers to the process of cleaning and structuring the collected historical alarm data to ensure the accuracy, completeness and reliability of the data analysis.

[0076] In a possible implementation, preprocessing includes data cleaning, multi-dimensional data supplementation, and data time series analysis.

[0077] Among them, data cleaning refers to removing duplicate, erroneous and invalid alarm records and reducing data noise. Multi-dimensional data supplementation refers to adding relevant contextual information to the alarm data, such as time, geographic location, device type, etc., in order to more comprehensively understand each alarm record. Data time series analysis is to organize the alarm data into a time series format to facilitate the identification of dynamic characteristics such as the frequency, duration and trend changes of alarm events.

[0078] It should be noted that preprocessing greatly improves the utilization value and analysis efficiency of data, laying a solid foundation for subsequent accurate root cause analysis.

[0079] First, the full alarm data of the past month is injected. The full alarm data includes not only the alarm data of ONU, PON port, and OLT, but also all the alarm data of power environment, BRAS, etc., so as to conduct end-to-end root cause analysis of home broadband services. Then, the alarm data is cleaned, and the key attributes of the data can be interpolated and supplemented according to the alarm network element, network element type, alarm type, alarm level, alarm title, service ID, etc., so as to ensure the quality of the data for machine learning. At the same time, the Pearson correlation coefficient is used to compare the network equipment characteristics with the network service characteristics. Correlation analysis is performed to find out the eigenvalues ​​with strong correlation, and feature selection is performed by comparing the variance of the strongly correlated features. Based on the feature selection results, data mining, machine learning and other methods are used based on AI technology to mine the rules and knowledge behind the alarm data, generate RCA rules for home broadband services, and finally monitor the real-time alarms obtained and match them according to the RCA rules to output the root cause of the fault. The root cause fault only displays the root alarm of the fault, the alarm information is compressed, and the topological relationship between the root alarm and the derived alarm is visualized, thereby realizing the rapid location of the fault.

[0080] Specifically, the RCA rules for home broadband services are generated using the Markov chain machine learning algorithm. Alarm data is essentially time series data. Different alarm events occur at different time points, and the order and interval of alarms can often reveal potential causes of failures. In order to accurately capture the transfer patterns of these alarm events, the system constructs the alarm data into a time series model. In the time series model, different alarm types (such as equipment failure, link unreachability, etc.) are regarded as different states of the system. Changes in the alarm state (i.e., the occurrence or disappearance of an alarm event) indicate state transitions. The probability of each alarm state transition is called the transition probability. These probabilities will be used in the subsequent rule generation process to infer the root cause of future alarms. Then, the Markov chain is used to model the alarm transfer. The Markov chain is a statistical model commonly used in time series with "no aftereffect". ", that is, the current state only depends on the previous state, but not on the earlier historical state. Based on this property, the system regards the occurrence process of alarm events as a discrete-time Markov process. At the same time, each device (such as ONU, PON port, OLT, BRAS, etc.) and each alarm type corresponds to a Markov chain state. For example, the alarm at a certain moment may be "ONU offline", and the next moment may be "PON port failure". The sequence and transition probability of these states will be used to analyze the root cause of the failure. Finally, based on the historical alarm data, the platform will count the transition frequencies between various alarms. For example, how often "PON port failure" occurs immediately after "ONU offline", or whether "OLT equipment failure" often occurs before "BRAS equipment failure". These historical frequencies will be used to calculate the transition probabilities between various states.

[0081] Once the state and transition probability of the Markov chain are trained, the platform can use this information to generate RCA rules. RCA rules are a set of conditional rules that represent the logic of alarm transfer, which are used to infer the root cause of a certain type of alarm event. At the same time, each RCA rule contains a time series pattern of a series of alarm events. For example, a rule may say "If in the past hour, 'ONU offline' occurred first, and then 'PON port failure' occurred, then the root cause of the problem may be 'OLT port abnormality'." This rule can be used to quickly analyze the cause of future alarms. In order to ensure the effectiveness of the rules, the system will continue to learn new alarm data, regularly update the Markov chain model and RCA rule base, and if the system identifies a new alarm pattern or a new alarm type, it will generate new rules and add them to the rule base, so that the system can be continuously optimized.

[0082] When new alarm data is input in real time, the system will use the generated RCA rules for matching and quickly determine whether these alarms comply with the existing rules. If the match is successful, the system can infer the root cause of the fault, generate a root alarm and notify the relevant operation and maintenance personnel to handle it. Since the network environment and equipment status are constantly changing, the mode and frequency of the alarm will also change accordingly. The system can not only perform static root cause analysis based on historical data, but also dynamically adjust the RCA rules to adapt to new changes in the network in real time and complete the broadband root cause analysis process.

[0083] S3, through the Pearson correlation coefficient, the network device characteristics and network service characteristics of the pre-processed historical alarm data are analyzed for correlation.

[0084] Among them, the Pearson correlation coefficient is a statistical indicator used to measure the linear correlation between two variables. The network device characteristics refer to the operating status characteristics of the network device itself, such as the number of times the physical machine network port is down. The "down" of the physical machine network port usually refers to the physical state of the network interface being "closed" or "inactivated". In short, "down" is a description of the interface status, indicating that the network port is currently in a non-working state and needs to be checked and repaired from aspects such as physical connection, device configuration or hardware. Network service characteristics refer to the characteristics reflected in the actual network service, such as the number of times the network element link destination is unreachable.

[0085] It should be noted that by analyzing the correlation between network equipment characteristics and business characteristics through the Pearson correlation coefficient, it is possible to effectively screen out the characteristics that have a direct impact on the fault, providing an important basis for fault root cause analysis. This can help the system focus on factors with greater impact and reduce interference from irrelevant variables, thereby improving the efficiency and accuracy of the analysis.

[0086] In a possible implementation manner, the network device characteristic is specifically the number of times a physical machine network port is down, and the network service characteristic is specifically the number of times a network element link destination is unreachable.

[0087] In a possible implementation, the Pearson correlation coefficient is specifically:

[0088]

[0089] Where X represents the number of physical machine network port down failures, Y represents the number of network element link destination unreachable failures, and ρ X,Y represents the Pearson correlation coefficient between the number of physical machine network port down failures and the number of network element link destination unreachable failures, cov(X,Y) represents the covariance between the number of physical machine network port down failures and the number of network element link destination unreachable failures, σ XIndicates the standard deviation of the number of physical machine network port down failures, σ Y represents the standard deviation of the number of network element link destination unreachable failures, E represents the expected value symbol, μ X Indicates the average number of physical machine network port down failures, μ Y Indicates the average number of NE link destination unreachable failures.

[0090] It should be noted that the role of the Pearson correlation coefficient in RCA rule generation is to help identify and screen strongly correlated features, and provide data support for subsequent causal analysis and rule generation. Through this feature extraction, the system can accurately locate the key factors in the alarm data, thereby improving the accuracy and practicality of the RCA rules, and ultimately achieving rapid location and analysis of the root cause of the fault.

[0091] S4, generating root cause analysis rules for home broadband services through a Markov chain algorithm based on the correlation analysis results.

[0092] Among them, the Markov chain algorithm is a mathematical model used to describe the transition process of the system state. In the root cause analysis, the Markov chain uses the time series model to calculate the transition probability between different alarm states, thereby generating a regular pattern of specific faults. Home broadband service refers to broadband network services for home users, including access, transmission and connection functions. The root cause analysis rules are a series of rules generated based on historical alarm data and state transition patterns, which are used to guide the analysis of real-time alarms.

[0093] It should be noted that by combining the correlation analysis results with the Markov chain algorithm, root cause analysis rules for home broadband failures are generated. These rules help the system identify the transfer patterns between alarm events. When a broadband failure occurs, the root cause can be quickly located by matching the alarm pattern, thus achieving accurate and efficient fault diagnosis.

[0094] In a possible implementation, S4 specifically includes:

[0095] S401, constructing the historical alarm data into a time series model, in which different alarm types are regarded as different states of the system, and the change of the alarm state indicates the transfer of the state.

[0096] Among them, the time series model is a data model that arranges historical alarm data in chronological order, so that the sequence and frequency of alarm events can be identified and analyzed. Different alarm types refer to the types of failures, such as equipment failure and link unreachability. The change of alarm status refers to the occurrence or disappearance of an alarm event, indicating the transition of status.

[0097] S402, simulating the state transition process between various alarm types in the time series model through the Markov chain algorithm.

[0098] S403: Calculate the state transition probability between various alarm types according to the historical alarm data.

[0099] The state transition probability refers to the possibility that the system transitions from one alarm state (alarm type) to another alarm state.

[0100] S404: Generate a root cause analysis rule according to the state transition frequency.

[0101] Among them, the root cause analysis rules are a set of logical rules or criteria used to identify and locate the root causes of faults in the system. In home broadband fault analysis, the root cause analysis rules are based on the correlation between historical alarm data and alarm events to determine which alarm states or alarm modes are most likely to cause faults.

[0102] It should be noted that the generated root cause analysis rules enable the system to quickly locate potential fault causes when similar alarm conditions are detected in real time. Compared with single alarm analysis, the model based on time series and transition probability has higher accuracy and timeliness, which helps to detect and prevent complex chain failures in advance.

[0103] S5, obtain real-time alarm data.

[0104] Among them, real-time alarm data refers to the fault or abnormal information monitored by the system at any time during the actual operation of the broadband network.

[0105] S6, matches the real-time alarm data with the root cause analysis rules through a matching algorithm.

[0106] Among them, the matching algorithm refers to an algorithm used to compare real-time alarm data with preset root cause analysis rules. Through the matching algorithm, the system can quickly compare the current real-time alarm with the established root cause analysis rules, thereby achieving accurate fault location and real-time analysis. This process significantly reduces the complexity and time of manual troubleshooting, allowing faults to be immediately processed and fed back when they occur.

[0107] Reference Figure 4 , showing a schematic diagram of a root cause alarm presentation method provided by an embodiment of the present invention.

[0108] Figure 4In the alarm monitoring window, the main alarm (also called the main alarm) usually refers to the core problem that affects the normal operation of the system or service. This type of alarm reflects a serious failure of the system and usually needs to be handled as a priority. The secondary alarm (also called the minor alarm) usually refers to a minor or intermittent problem in the system. These problems themselves may not necessarily lead to a complete service interruption, but may affect performance or increase the risk of failure. For the main and secondary alarms, the associated main alarm and secondary alarm are presented in the alarm monitoring flow window. As long as there are active alarms in the main and secondary alarms, the associated presentation continues to exist. Cleared alarms are presented in a gray background color, and uncleared alarms are presented in a normal background color.

[0109] Specifically, there are two main categories of root cause analysis rules for home broadband services, including:

[0110] Inter-network element alarm compression: Based on time series analysis, network element alarms in the same area (with topological relationships) are aggregated to find high-frequency alarms of the same type and form an aggregated alarm. This achieves the purpose of compressing alarms, such as "multiple FTTH offline alarms appearing under the same PON port."

[0111] Root cause analysis across network elements: Based on a Markov chain, the alarm data that occurs successively between network elements is modeled, and the probability of primary and secondary alarm conversion is calculated through the frequency statistics of the main alarm and the frequency statistics of the primary and secondary alarms. High-probability primary and secondary alarms can be used as the basis for root cause analysis, such as OLT disconnection and related dynamic ring alarms, related hardware alarms, and related transmission alarms.

[0112] Specifically, multiple PON port alarms, multiple ONU alarms, and multiple PON port alarms belong to the secondary alarm category, and OLT outage alarms and multiple OLT interruption alarms under the same BRAS belong to the secondary alarm category.

[0113] It should be noted that primary and secondary alarms are judged by RCA rules, among which the main RCA rules are composed of multi-ONU alarm derivation, multi-PON port alarm derivation, multi-OLT interruption alarm under the same BRAS and OLT decommissioning alarm fault cause judgment.

[0114] Reference Figure 5 , showing a schematic diagram of a derivative alarm presentation method provided by an embodiment of the present invention.

[0115] Figure 5 In the context of a network failure or anomaly, a derivative alarm refers to a subsequent alarm triggered by an initial failure. These alarms usually do not occur independently, but are chain reactions caused by a fundamental failure or problem. For derivative alarms, the associated derivative alarms and sub-alarms are presented in the alarm monitoring flow window. As long as there are active alarms in the derivative alarms, the associated presentation continues to exist. Cleared alarms are presented in a gray background color, while uncleared alarms are presented in a normal background color.

[0116] Specifically, the derivation of multi-ONU alarms refers to the simultaneous occurrence of multiple FTTH offline alarms under the same PON port, which satisfies the condition of 'multiple FTTH offline alarms under the same PON port'. The derivation of multi-PON port alarms refers to the occurrence of multiple PON port interruption alarms under the same OLT, which satisfies the condition of 'multiple PON port interruption alarms under the same OLT'. The multiple OLT interruption alarms under the same BRAS refer to the occurrence of multiple OLT upstream interruptions, which automatically determines whether there is a common upstream BRAS. The cause of the OLT out-of-service alarm fault determination refers to the occurrence of the derived alarm 'OLT out-of-service', which determines whether it is due to power failure, and matches the 'total voltage , one-time power-off, first-level low-voltage disconnection alarm, power outage alarm, presence or absence of AC power, output voltage too low, node power outage, AC power outage, low output voltage, first-level low-voltage disconnection alarm, second-level low-voltage disconnection alarm, battery pack first-level disconnection alarm, AC input power outage alarm, AC input phase loss alarm, AC input voltage too low alarm, battery power supply alarm, base station power outage alarm, base station DC power supply abnormal alarm, power supply abnormal power off" and other alarms, or determine whether it is the cause of the OLT device itself, match "board offline, single board communication failure, board software not running, single board temperature is too high", or determine whether it is a transmission problem.

[0117] S7, based on the matching result, outputs the root cause of the home broadband failure.

[0118] Among them, the matching result refers to the output after comparing the real-time alarm data and the pre-generated root cause analysis rules through the matching algorithm. The root cause of home broadband failure refers to the main reason for the abnormal user broadband connection, which may include equipment failure, network line interruption, environmental factors, etc. By outputting the root cause of home broadband failure through matching results, the system can intuitively feedback the cause of the failure to the maintenance personnel or the automation system in the shortest time, realizing real-time and efficient fault location.

[0119] The beneficial effects of the present invention are as follows:

[0120] In an embodiment of the present invention, the acquired historical alarm data is preprocessed, and then the Pearson correlation coefficient is used to perform correlation analysis on the network device characteristics and network service characteristics of the preprocessed historical alarm data. Based on the correlation analysis results, the root cause analysis rules of the home broadband service are generated through the Markov chain algorithm. Finally, the real-time alarm data is matched with the root cause analysis rules through the matching algorithm to output the root cause failure of the home broadband. The method automatically learns from the massive alarm data of the home broadband by using the Markov chain machine learning algorithm, continuously refines and summarizes the RCA rules related to the home broadband, performs alarm compression across network elements through the RCA rules, and performs root cause analysis across network elements, accurately analyzes the alarm causes of the home broadband, improves the work order compression ratio, improves maintenance efficiency, reduces operation and maintenance costs, reduces maintenance workload, and enhances user experience.

[0121] Device Embodiment

[0122] Reference Manual Attached Figure 6 , showing a schematic structural diagram of a broadband fault root cause analysis device provided by an embodiment of the present invention.

[0123] Another specific embodiment of the present invention discloses a broadband fault root cause analysis device, which is used to implement any broadband fault root cause analysis method based on a Markov chain algorithm in any method embodiment. The broadband fault root cause analysis device 20 includes:

[0124] The first acquisition module 201 is used to acquire historical alarm data.

[0125] The processing module 202 pre-processes the historical alarm data.

[0126] The analysis module 203 performs correlation analysis on the network device characteristics and network service characteristics of the pre-processed historical alarm data through the Pearson correlation coefficient.

[0127] The generating module 204 is used to generate root cause analysis rules of the home broadband service according to the correlation analysis result by using a Markov chain algorithm.

[0128] The second acquisition module 205 is used to acquire real-time alarm data.

[0129] The matching module 206 matches the real-time alarm data with the root cause analysis rules through a matching algorithm.

[0130] The output module 207 is used to output the root cause of the home broadband failure according to the matching result.

[0131] In a possible implementation manner, the historical alarm data includes: ONU alarm data, PON port alarm data, OLT uplink interruption alarm data, power environment alarm data and BRAS alarm data.

[0132] In a possible implementation, preprocessing includes data cleaning, multi-dimensional data supplementation, and data time series analysis.

[0133] In a possible implementation manner, the network device characteristic is specifically the number of times a physical machine network port is down, and the network service characteristic is specifically the number of times a network element link destination is unreachable.

[0134] In a possible implementation, the Pearson correlation coefficient is specifically:

[0135]

[0136] Where X represents the number of physical machine network port down failures, Y represents the number of network element link destination unreachable failures, and ρ X,Y represents the Pearson correlation coefficient between the number of physical machine network port down failures and the number of network element link destination unreachable failures, cov(X,Y) represents the covariance between the number of physical machine network port down failures and the number of network element link destination unreachable failures, σ X Indicates the standard deviation of the number of physical machine network port down failures, σ Y represents the standard deviation of the number of network element link destination unreachable failures, E represents the expected value symbol, μ X Indicates the average number of physical machine network port down failures, μ Y Indicates the average number of NE link destination unreachable failures.

[0137] In a possible implementation, the generation module is used to:

[0138] The historical alarm data is constructed as a time series model. Different alarm types in the time series model are regarded as different states of the system, and the change of alarm state indicates the transition of state.

[0139] The Markov chain algorithm is used to simulate the state transition process between various alarm types in the time series model.

[0140] Based on historical alarm data, calculate the state transition probability between various alarm types.

[0141] Generate root cause analysis rules based on the state transition frequency.

[0142] The device provided in the embodiment of the present invention can implement each process and step in the above method embodiment, and can obtain the same or similar technical effects. To avoid repetition, the present invention will not go into details.

[0143] The beneficial effects of the present invention are as follows:

[0144] In an embodiment of the present invention, the acquired historical alarm data is preprocessed, and then the Pearson correlation coefficient is used to perform correlation analysis on the network device characteristics and network service characteristics of the preprocessed historical alarm data. Based on the correlation analysis results, the root cause analysis rules of the home broadband service are generated through the Markov chain algorithm. Finally, the real-time alarm data is matched with the root cause analysis rules through the matching algorithm to output the root cause failure of the home broadband. The method automatically learns from the massive alarm data of the home broadband by using the Markov chain machine learning algorithm, continuously refines and summarizes the RCA rules related to the home broadband, performs alarm compression across network elements through the RCA rules, and performs root cause analysis across network elements, accurately analyzes the alarm causes of the home broadband, improves the work order compression ratio, improves maintenance efficiency, reduces operation and maintenance costs, reduces maintenance workload, and enhances user experience.

[0145] Computer Readable Storage Medium Embodiments

[0146] The present invention provides a computer-readable storage medium on which a computer program is stored. The computer program can be loaded and executed by a processor to provide a broadband fault root cause analysis method based on a Markov chain algorithm according to the first aspect.

[0147] The applicant of the present invention has made a detailed explanation and description of the implementation examples of the present invention in conjunction with the drawings in the specification. However, those skilled in the art should understand that the above implementation examples are only preferred implementation schemes of the present invention, and the detailed description is only to help readers better understand the spirit of the present invention, but not to limit the scope of protection of the present invention. On the contrary, any improvements or modifications based on the inventive spirit of the present invention should fall within the scope of protection of the present invention.

[0148] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the embodiments of the present invention, rather than to limit them. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the above embodiments, or replace some of the technical features therein by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention. Any changes or replacements that can be easily thought of by those skilled in the art within the technical scope disclosed in the present invention should be covered within the protection scope of the present invention.

Claims

1. A broadband fault root cause analysis method based on Markov chain algorithm, characterized in that: include: S1, obtain historical alarm data; S2, preprocessing the historical alarm data; S3, performing correlation analysis on the network device characteristics and network service characteristics of the pre-processed historical alarm data through the Pearson correlation coefficient; S4, generating root cause analysis rules for home broadband services through a Markov chain algorithm based on the correlation analysis results; S5, obtain real-time alarm data; S6, matching the real-time alarm data with the root cause analysis rule through a matching algorithm; S7, based on the matching result, outputs the root cause of the home broadband failure.

2. The broadband fault root cause analysis method based on Markov chain algorithm according to claim 1 is characterized in that: The historical alarm data includes: ONU alarm data, PON port alarm data, OLT uplink interruption alarm data, power environment alarm data and BRAS alarm data.

3. The broadband fault root cause analysis method based on Markov chain algorithm according to claim 1 is characterized in that: The preprocessing includes data cleaning, multi-dimensional data supplementation and data time series analysis.

4. The method for analyzing the root cause of broadband faults based on a Markov chain algorithm according to claim 1, characterized in that: The network device characteristic is specifically the number of times the physical machine network port is down, and the network service characteristic is specifically the number of times the network element link destination is unreachable.

5. The method for analyzing broadband fault root causes based on Markov chain algorithm according to claim 4, characterized in that: The Pearson correlation coefficient is specifically: Where X represents the number of physical machine network port down failures, Y represents the number of network element link destination unreachable failures, and ρ X,Y represents the Pearson correlation coefficient between the number of physical machine network port down failures and the number of network element link destination unreachable failures, cov(X,Y) represents the covariance between the number of physical machine network port down failures and the number of network element link destination unreachable failures, σ X Indicates the standard deviation of the number of physical machine network port down failures, σ Y represents the standard deviation of the number of network element link destination unreachable failures, E represents the expected value symbol, μ X Indicates the average number of physical machine network port down failures, μ Y Indicates the average number of NE link destination unreachable failures.

6. The broadband fault root cause analysis method based on Markov chain algorithm according to claim 1, characterized in that: The S4 specifically includes: S401, constructing the historical alarm data into a time series model, wherein different alarm types in the time series model are regarded as different states of the system, and changes in the alarm state represent state transitions; S402, simulating the state transition process between various alarm types in the time series model through the Markov chain algorithm; S403, calculating the state transition probability between various alarm types according to the historical alarm data; S404: Generate the root cause analysis rule according to the state transition frequency.

7. A broadband fault root cause analysis device based on Markov chain algorithm, characterized in that: For executing the method according to any one of claims 1 to 6, the device comprises: The first acquisition module is used to acquire historical alarm data; A processing module, which pre-processes the historical alarm data; The analysis module uses the Pearson correlation coefficient to perform correlation analysis on the network device characteristics and network service characteristics of the pre-processed historical alarm data; A generation module, used to generate root cause analysis rules of home broadband services through a Markov chain algorithm based on the correlation analysis results; The second acquisition module is used to obtain real-time alarm data; A matching module, which matches the real-time alarm data with the root cause analysis rule through a matching algorithm; The output module is used to output the root cause fault of the home broadband according to the matching result.

8. The broadband fault root cause analysis device based on Markov chain algorithm according to claim 7, characterized in that: The historical alarm data includes: ONU alarm data, PON port alarm data, OLT uplink interruption alarm data, power environment alarm data and BRAS alarm data.

9. The broadband fault root cause analysis device based on Markov chain algorithm according to claim 7, characterized in that: The preprocessing includes data cleaning, multi-dimensional data supplementation and data time series analysis.

10. The broadband fault root cause analysis device based on Markov chain algorithm according to claim 7, characterized in that: The network device characteristic is specifically the number of times the physical machine network port is down, and the network service characteristic is specifically the number of times the network element link destination is unreachable.

Images