System and method for monitoring alarm in real time based on OLT (Optical Line Terminal) equipment

Abstract

The invention relates to the field of communication warning, in particular to a real-time monitoring system and method based on OLT equipment warning, and the method comprises the following steps: S1, collecting equipment response time, packet loss rate, state information of OLT equipment and disconnection data of a user, and carrying out the preprocessing of the collected data; s2, analyzing and processing the collected data, and identifying a complex fault mode and a multi-factor problem; s3, alarm information is generated for the Pi ng measurement alarm, the port abnormity alarm, the optical modem offline alarm and the comprehensive alarm, operation and maintenance personnel are informed of the generated alarm information through automatic processing, and processing suggestions are provided; and S4, real-time state display is performed through the data acquisition module, and comprehensive data display, historical data query and data interaction are performed through the data processing module. The fault response speed can be improved, the network stability is enhanced, the operation and maintenance efficiency is improved, and the user experience is improved.

Description

Technical Field

[0001] The present invention relates to the field of communication alarms, and in particular to a system and method based on real-time monitoring of OLT equipment alarms. Background Art

[0002] In today's communications field, optical line terminal (OLT) equipment, as one of the core network devices, undertakes the important task of connecting multiple user terminals and upper-layer networks. With the continuous development of communication technology and the increasing requirements of users for network service quality, higher requirements are placed on the operating stability and reliability of OLT equipment.

[0003] Traditional OLT equipment monitoring methods mainly rely on manual inspections and simple threshold alarms. This method has the disadvantages of long response time, untimely fault detection, and difficulty in accurately identifying complex fault modes and multi-factor problems. When a device fails, it often takes a long time to be discovered and processed, which has a great impact on the user's network experience and also increases the operation and maintenance costs of network operators and enterprises. There are some existing alarm systems that can monitor the alarm information of OLT equipment, but these systems usually run independently and cannot be linked with other business systems, resulting in cumbersome and inefficient alarm processing. Many systems rely on a passive response mode, that is, they only process after the alarm occurs, lack an early warning mechanism, and cannot detect potential problems in advance. There are compatibility issues in the monitoring systems provided by different equipment manufacturers, which makes it difficult to integrate data and form a unified monitoring and management view, affecting the overall operation and maintenance efficiency. Summary of the invention

[0004] To solve the above problems, the present invention provides a system and method based on real-time monitoring of OLT device alarms, which can improve fault response speed, enhance network stability, improve operation and maintenance efficiency, and enhance user experience.

[0005] To achieve the above object, the technical solution adopted by the present invention is:

[0006] A system based on real-time monitoring of OLT equipment alarms includes: a data acquisition module, a data processing module, an alarm module and a monitoring interface module connected in communication, wherein:

[0007] The data acquisition module is used to collect device response time, packet loss rate, OLT device status information and user disconnection data, and pre-process the collected data;

[0008] The data processing module is used to analyze and process the pre-processed data to identify complex failure modes and multi-factor problems;

[0009] The alarm module includes a generating unit and a processing unit, wherein:

[0010] The generation unit is used to generate alarm information for Ping test alarm, port abnormality alarm, optical modem offline alarm and comprehensive alarm;

[0011] The processing unit is used to notify the operation and maintenance personnel of the generated alarm information through automatic processing and provide processing suggestions;

[0012] The monitoring interface module is used for real-time status display, comprehensive data display, historical data query and data interaction.

[0013] Furthermore, the collecting device response time, packet loss rate, OLT device status information and user disconnection data, and pre-processing the collected data include:

[0014] The response time is calculated by recording the sending time and the receiving time of each request through Ping measurement; the packet loss rate is calculated based on the number of data packets sent and the number of responses received;

[0015] Obtain detailed status information of the OLT device through the SNMP protocol or the NetConf protocol, wherein the detailed status of the OLT device includes the board status, the port status and the optical modem status;

[0016] Obtain user disconnection data through the business system, including information about the user's connection status, disconnection time, and disconnection reason;

[0017] The preprocessing includes data cleaning, data standardization and data aggregation.

[0018] Furthermore, the analysis and processing of the pre-processed data includes the following steps:

[0019] A1: Through the data acquisition module, in the DMI technical architecture environment, based on the prior knowledge of OLT equipment failure modes and multi-factor problems, relevant features are selected from the data and the data set is divided into a training set and a test set;

[0020] A2: Train the Naive Bayes model with the training set, evaluate the trained Naive Bayes model with the test set, and use the trained Naive Bayes model to predict new data and identify possible fault modes.

[0021] A3: Determine the input variables and output variables based on the multivariate data processed by partial least squares method. The input variables include the output results and related features of the naive Bayes model. Perform feature extraction on the input variables and divide the data set into a training set and a test set.

[0022] A4: Train the PLS model with the training set, evaluate the performance of the PLS model with the test set, analyze multi-factor problems based on the trained PLS model, and determine the influence of each input variable on the output variable by looking at the regression coefficient of the model.

[0023] Furthermore, the method of training the naive Bayesian model by the training set and evaluating the trained naive Bayesian model by the test set comprises the following steps:

[0024] A21: Calculate the prior probability of each fault category in the training set;

[0025] A22: For each fault feature and fault category, calculate the conditional probability of the fault feature under the fault category;

[0026] A23: When the fault feature is a discrete variable, the frequency of fault feature values ​​under the fault category is calculated as the first conditional probability; when the fault feature is a continuous variable, it is assumed that the fault feature follows the probability distribution of the normal distribution, the mean and standard deviation of the normal distribution are estimated through the training set data, and then the second conditional probability is calculated;

[0027] A24: For each test data point, the posterior probability of belonging to each fault category is calculated according to the naive Bayes model. The test data point is assigned to the category with the highest probability according to the posterior probability, and compared with the true category label of the test set to calculate the evaluation index of the model.

[0028] Furthermore, the calculation formula of the prior probability is: Where: P(C j ) is the prior probability, C j is the fault category, is the fault category C in the training set j The number of data points, N train is the total number of data points in the training set;

[0029] The calculation formula of the first conditional probability is: Where: X i is the fault characteristic, is the fault category C in the training set j And the fault feature X i The number of data points that take a specific value, is the fault category C in the training set j The number of data points;

[0030] The calculation formula of the second conditional probability is: Where: μx i , C j In fault category C j Fault feature Xi The mean of In fault category C j Fault feature X i The variance of

[0031] The posterior probability formula is: Where: n is the number of fault features, m is the number of fault categories.

[0032] Furthermore, the method of training the partial least squares model by using the training set data and evaluating the performance of the partial least squares model by using the test set data comprises the following steps:

[0033] A41: Initialize the first component as a linear combination of the input variable matrix, determine the weight vector based on the maximum covariance between the first component and the output variable vector; calculate the regression coefficient of the first component on the output variable vector, and remove the influence of the first component from the input variable matrix to obtain a new matrix;

[0034] A42: repeating step A41 until a preset number of components are extracted, and determining the number of extracted components according to cross-validation;

[0035] A43: Use the test set data to calculate the root mean square error and determination coefficient to evaluate the performance of the partial least squares model;

[0036] In step A42, the final regression model is Where y is the output variable, k is the number of components extracted, and t i It is an ingredient, c i Yes i The regression coefficient for y, e is the residual, and i is the index variable.

[0037] Furthermore, the generating of alarm information for Ping test alarm, port abnormality alarm, optical modem offline alarm and comprehensive alarm comprises the following steps:

[0038] B1: Continuously monitor ping data, port status data, optical modem connection status data and comprehensive data, and compare the real-time data with the set alarm trigger conditions;

[0039] B2: Based on the comparison results, determine whether the alarm triggering conditions are met. When the conditions are met, the corresponding alarm is triggered. When multiple alarms occur at the same time or specific combination conditions are met, a comprehensive alarm is triggered;

[0040] B3: When an alarm is triggered, alarm information is generated, including: alarm type, alarm time, device information and processing suggestions, and the alarm information is stored in the database and log file.

[0041] Furthermore, the automatic processing notifies the operation and maintenance personnel of the generated alarm information and provides processing suggestions, including:

[0042] Determine the notification method based on the needs of the operation and maintenance personnel. When the alarm information is generated, send the alarm information to the operation and maintenance personnel according to the configured notification method;

[0043] Through association rule mining technology, the association relationship between alarms and processing methods is determined from historical alarm data and processing records. The frequent item sets in historical alarm data are mined through the Apriori algorithm to determine the alarm types and processing methods that appear together. The detailed information and processing results of each alarm are recorded to form a historical alarm database. When a new alarm occurs, the alarm case is determined by querying the historical database and corresponding processing suggestions are provided.

[0044] Furthermore, the real-time status display, comprehensive data display, historical data query and data interaction include:

[0045] Through the front-end framework and data visualization library, the real-time status of the OLT device data continuously collected by the data collection module is displayed in real time. Different types of data are displayed through different charts. The front-end page is updated in real time through WebSocket technology. When the back-end collects new data, the data is immediately pushed to the front-end page.

[0046] Integrate and calculate the various collected data to generate comprehensive data indicators, process and calculate the data through the Pandas data analysis library, display the comprehensive data through different charts, and realize the interactive functions of data screening, time range selection and indicator switching through the front-end interactive technology of JavaScript event processing. Store the collected historical data in a relational database, design the back-end data query interface through RESTful API technology, and display the historical data in a visual way according to the user's query results.

[0047] A method based on real-time monitoring of OLT device alarms, applied to a system based on real-time monitoring of OLT device alarms as described in any one of the above contents, comprises the following steps:

[0048] S1: Collect device response time, packet loss rate, OLT device status information and user disconnection data, and pre-process the collected data;

[0049] S2: Analyze and process the collected data to identify complex failure modes and multi-factor problems;

[0050] S3: Generates alarm information for ping test alarms, port abnormality alarms, optical modem offline alarms, and comprehensive alarms, notifies the operation and maintenance personnel of the generated alarm information through automatic processing, and provides handling suggestions;

[0051] S4: Real-time status display through data acquisition equipment, comprehensive data display, historical data query and data interaction through data analysis results.

[0052] The beneficial effects of the present invention are:

[0053] 1. Through continuous data collection and real-time status display, the system can detect abnormal conditions as soon as the OLT equipment fails. For example, when the equipment response time is too long, the packet loss rate is too high, the port is abnormal, or the optical modem is offline, the system can immediately trigger an alarm and notify the operation and maintenance personnel in a variety of ways, greatly shortening the time to find the fault, allowing the operation and maintenance personnel to take quick action to reduce the impact of the fault on network services. The system accurately identifies complex fault modes and multi-factor problems by analyzing and processing the collected data using the naive Bayes algorithm and partial least squares method. Through comprehensive analysis of different types of alarms, such as Ping alarms, port abnormality alarms, optical modem offline alarms, and comprehensive alarms, the operation and maintenance personnel can quickly locate the fault point and improve the efficiency of troubleshooting.

[0054] 2. Through the analysis of historical data and the application of machine learning algorithms, the system can discover potential failure modes and trends and take preventive measures in advance. For example, based on the occurrence patterns of similar failures in historical data, the system predicts possible problems and reminds operation and maintenance personnel to perform preventive maintenance, such as checking equipment status and optimizing network configuration. This can avoid failures and improve network stability. The system monitors and comprehensively analyzes various performance indicators of OLT equipment in real time, such as equipment response time, packet loss rate, port status, etc. Based on the analysis results, operation and maintenance personnel can adjust network parameters and optimize network structure in a timely manner to improve network performance and reliability.

[0055] 3. After generating alarm information, the system can automatically provide processing suggestions, reducing the decision-making time and workload of operation and maintenance personnel. At the same time, the automated alarm processing and notification functions reduce the workload of manual inspections and improve operation and maintenance efficiency. The real-time status display, comprehensive data display and historical data query functions provided by the system enable operation and maintenance personnel to intuitively understand the operation status and historical trends of the equipment. Through the visual interface and interactive functions, operation and maintenance personnel can easily analyze and explore data and quickly obtain the required information. In addition, the permission management function can provide personalized interfaces and functions according to the roles and permissions of different users, improving team collaboration efficiency. Since the system can respond to faults quickly and notify operation and maintenance personnel in time for processing, the network interruption time is greatly shortened. For users, this means more stable and reliable network services, reducing the inconvenience and losses caused by network failures. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 It is a schematic diagram of a system based on real-time monitoring of OLT equipment alarms.

[0057] Figure 2 The invention is a flow chart of a method for real-time monitoring of an OLT device alarm.

[0058] Figure 3 It is the flow chart of the data processing module. DETAILED DESCRIPTION

[0059] See also Figure 1 , Figure 2 and Figure 3 As shown, the present invention relates to a system for real-time monitoring of OLT device alarms, comprising: a data acquisition module, a data processing module, an alarm module and a monitoring interface module connected in communication, wherein:

[0060] The data acquisition module is used to collect device response time, packet loss rate, OLT device status information and user disconnection data, and pre-process the collected data;

[0061] The data processing module is used to analyze and process the pre-processed data to identify complex failure modes and multi-factor problems;

[0062] The alarm module includes a generating unit and a processing unit, wherein:

[0063] The generation unit is used to generate alarm information for Ping test alarm, port abnormality alarm, optical modem offline alarm and comprehensive alarm;

[0064] The processing unit is used to notify the operation and maintenance personnel of the generated alarm information through automatic processing and provide processing suggestions;

[0065] The monitoring interface module is used for real-time status display, comprehensive data display, historical data query and data interaction.

[0066] Furthermore, the collecting device response time, packet loss rate, OLT device status information and user disconnection data, and pre-processing the collected data includes:

[0067] The response time is calculated by recording the sending time and the receiving time of each request through Ping measurement; the packet loss rate is calculated based on the number of data packets sent and the number of responses received;

[0068] Obtain detailed status information of the OLT device through the SNMP protocol or the NetConf protocol, wherein the detailed status of the OLT device includes the board status, the port status and the optical modem status;

[0069] Obtain user disconnection data through the business system, including information about the user's connection status, disconnection time, and disconnection reason;

[0070] The preprocessing includes data cleaning, data standardization and data aggregation.

[0071] Specifically, send a request through the SNMP or NetConf client library to obtain the status information of the OLT device, such as device temperature, voltage, optical power, port status, traffic statistics, etc. For SNMP, specify the information to be obtained through OID (Object Identifier). For NetConf, use get-config to obtain device configuration and status data. Parse the response data returned from the device and extract the required status information.

[0072] Connect with the business system to determine the storage location and acquisition method of user disconnection data. The acquisition method is to obtain it from a specific table in the database, log file or through the API interface. According to the determined data source, use the corresponding database query statement, log analysis tool or API call method to extract user disconnection data. Perform necessary conversion and formatting on the extracted user disconnection data to make it meet the requirements of the data acquisition module. Store the converted disconnection data in the database or data structure for subsequent analysis and preprocessing.

[0073] The data preprocessing includes checking whether the collected data contains abnormal values ​​or missing values. For abnormal values, delete, replace or mark them according to the specific situation. For missing values, interpolation, mean filling and other methods are used to fill them. Data from different sources are standardized so that they have a unified format and unit. As needed, the collected data are aggregated to reduce the amount of data and improve analysis efficiency. The preprocessed data is stored in a database or file for subsequent data analysis and monitoring. At the same time, the data is output to the monitoring interface or other systems as needed for operation and maintenance personnel to view and analyze.

[0074] Furthermore, the analysis and processing of the pre-processed data includes the following steps:

[0075] A1: Through the data acquisition module, in the DMI technical architecture environment, based on the prior knowledge of OLT equipment failure modes and multi-factor problems, relevant features are selected from the data and the data set is divided into a training set and a test set;

[0076] A2: Train the Naive Bayes model with the training set, evaluate the trained Naive Bayes model with the test set, and use the trained Naive Bayes model to predict new data and identify possible fault modes.

[0077] A3: Determine the input variables and output variables based on the multivariate data processed by partial least squares method. The input variables include the output results and related features of the naive Bayes model. Perform feature extraction on the input variables and divide the data set into a training set and a test set.

[0078] A4: Train the PLS model with the training set, evaluate the performance of the PLS model with the test set, analyze multi-factor problems based on the trained PLS model, and determine the influence of each input variable on the output variable by looking at the regression coefficient of the model.

[0079] Specifically, in step A1, the DMI technical architecture environment includes a data collection layer, a data storage layer, a data analysis layer, and an application display layer. The data collection layer is responsible for collecting data from various data sources and performing preliminary preprocessing; the data storage layer is used to store the collected data, using relational databases, non-relational databases, or distributed file systems; the data analysis layer is the core part of DMI, including data analysis algorithms such as naive Bayes and partial least squares, as well as data mining, machine learning, etc., for in-depth analysis of data, identification of fault modes and multi-factor problems; the application display layer displays the analysis results to operation and maintenance personnel in an intuitive manner, such as dashboards, reports, alarms, etc. The data analysis platform is built through Python's data analysis library. The application display layer is realized by developing dashboards and reports through Web technology. The DMI system is regularly evaluated and optimized for performance, and the system's functions and performance are continuously improved based on the feedback from operation and maintenance personnel and actual application conditions, so as to improve the accuracy and efficiency of fault diagnosis and multi-factor problem analysis.

[0080] The selection of relevant features, such as the change of optical power, the stability of port status, the level of packet loss rate, and other features that may be closely related to the fault mode. At the same time, the feature importance evaluation method is used to determine which features contribute more to fault classification through information gain. The data set is randomly divided into a training set and a test set, which can usually be divided in a ratio of 7:3 or 8:2.

[0081] In step A3, since partial least squares is mainly used to process multivariate data, the input variables, i.e., the output results of the naive Bayes model and other relevant features, and the output variables (such as the performance indicators of the equipment or the severity of the fault, etc.) are first determined, and the input variables are feature extracted to reduce the correlation between the variables and extract the main components. The partial least squares method finds a set of new components that can both explain the variation of the input variables to the greatest extent and be most related to the output variables by considering the relationship between the input variables and the output variables at the same time. The data is divided into a training set and a test set to ensure that the division of the training set and the test set is random and can represent the characteristics of the entire data set.

[0082] Furthermore, the method of training the naive Bayesian model by the training set and evaluating the trained naive Bayesian model by the test set comprises the following steps:

[0083] A21: Calculate the prior probability of each fault category in the training set;

[0084] A22: For each fault feature and fault category, calculate the conditional probability of the fault feature under the fault category;

[0085] A23: When the fault feature is a discrete variable, the frequency of fault feature values ​​under the fault category is calculated as the first conditional probability; when the fault feature is a continuous variable, it is assumed that the fault feature follows the probability distribution of the normal distribution, the mean and standard deviation of the normal distribution are estimated through the training set data, and then the second conditional probability is calculated;

[0086] A24: For each test data point, the posterior probability of belonging to each fault category is calculated according to the naive Bayes model. The test data point is assigned to the category with the highest probability according to the posterior probability, and compared with the true category label of the test set to calculate the evaluation index of the model.

[0087] Specifically, the fault feature is a discrete variable, such as the port status fault or normal. The fault feature is a continuous variable, such as the device response time or packet loss rate.

[0088] The trained Naive Bayes model is evaluated using the test set data. For each test data point, the posterior probability of belonging to each fault category is calculated based on the model. The trained Naive Bayes model is used to predict new OLT equipment data and identify possible fault modes. Based on the probability distribution of fault categories output by the model, suggestions for fault diagnosis and handling are provided to operation and maintenance personnel.

[0089] Furthermore, the calculation formula of the prior probability is: Where: P(C j ) is the prior probability, C j is the fault category, is the fault category C in the training set j The number of data points, N train is the total number of data points in the training set;

[0090] The calculation formula of the first conditional probability is: Where: X i is the fault characteristic, is the fault category C in the training set j And the fault feature X i The number of data points that take a specific value, is the fault category C in the training setj The number of data points;

[0091] The calculation formula of the second conditional probability is: Where: μx i , C j In fault category C j Fault feature X i The mean of In fault category C j Fault feature X i The variance of

[0092] The posterior probability formula is: Where: n is the number of fault features, m is the number of fault categories.

[0093] Furthermore, the method of training the partial least squares model by using the training set data and evaluating the performance of the partial least squares model by using the test set data comprises the following steps:

[0094] A41: Initialize the first component as a linear combination of the input variable matrix, determine the weight vector based on the maximum covariance between the first component and the output variable vector; calculate the regression coefficient of the first component on the output variable vector, and remove the influence of the first component from the input variable matrix to obtain a new matrix;

[0095] A42: repeating step A41 until a preset number of components are extracted, and determining the number of extracted components according to cross-validation;

[0096] A43: Use the test set data to calculate the root mean square error and determination coefficient to evaluate the performance of the partial least squares model;

[0097] In step A42, the final regression model is Where y is the output variable, k is the number of components extracted, and t i It is an ingredient, c i Yes i The regression coefficient for y, e is the residual, and i is the index variable.

[0098] Specifically, the first component t1 is initialized as a linear combination of the input variable matrix X: t1 = Xw1, where: the input variable matrix X is an n×p matrix, where n is the number of samples, p is the input variable, and w1 is the weight vector. w1 is determined based on the maximum covariance between t1 and the output variable vector y, where y is an n×1 vector;

[0099] The regression coefficient of t1 on y is calculated according to the formula y=t1c1+e1, where e1 is the residual. Remove the influence of t1 from X to obtain a new X matrix, where p1 is the projection vector of t1 on X; repeat the above steps until the preset number of components is extracted, and the number of extracted components is determined based on cross-validation. The final regression model is: Where k is the number of extracted components, i is the index variable, representing the ordinal number of the component; the performance of the partial least squares model is evaluated by calculating the root mean square error and the coefficient of determination through the test set data.

[0100] The final regression model is a regression model that can effectively realize the functions of equipment fault diagnosis, processing suggestions, etc. through various data collection and analysis technologies and different algorithms in the process of real-time monitoring of OLT equipment alarms. This model may take various factors into consideration, has high accuracy and reliability, and can provide strong support for operation and maintenance personnel.

[0101] In step A43, the prediction accuracy of the model is evaluated, and the root mean square error is calculated as: where y i is the actual value, is the predicted value, n is the number of samples, and the smaller the root mean square error value is, the smaller the prediction error of the model is.

[0102] The formula for calculating the coefficient of determination is: in is the mean of the actual values, R 2 The closer the value is to 1, the better the model fit is.

[0103] The multi-factor problem of OLT equipment is analyzed through the trained partial least squares model. By looking at the regression coefficient of the model, the degree of influence of each input variable on the output variable can be determined. For example, if the regression coefficient of an input variable is large, it means that the variable has a greater impact on the performance or fault severity of the equipment. Based on this information, operation and maintenance personnel can focus on those factors that have a greater impact on equipment operation and take corresponding measures for optimization and maintenance.

[0104] Further, the generating of alarm information for the Ping test alarm, the port abnormality alarm, the optical modem offline alarm and the comprehensive alarm comprises the following steps:

[0105] B1: Continuously monitor the Ping measurement data, port status data, optical modem connection status data and comprehensive data, and compare the real-time data with the set alarm trigger conditions;

[0106] B2: Based on the comparison results, determine whether the alarm triggering conditions are met. When the conditions are met, the corresponding alarm is triggered. When multiple alarms occur at the same time or specific combination conditions are met, a comprehensive alarm is triggered;

[0107] B3: When an alarm is triggered, alarm information is generated, including: alarm type, alarm time, device information and processing suggestions, and the alarm information is stored in the database and log file.

[0108] Specifically, the triggering condition of the Ping measurement alarm: When the response time of the Ping measurement exceeds the set threshold or the packet loss rate is higher than the set percentage, the Ping measurement alarm is triggered. For example, when the response time is set to exceed 100ms or the packet loss rate is higher than 5%, it is regarded as a Ping measurement abnormality and an alarm is triggered. The triggering condition of the port abnormality alarm: When the port status of the OLT device is abnormal, such as port closure, port failure or port traffic abnormality, the port abnormality alarm is triggered. For port traffic abnormality, it is set that when the port traffic exceeds a certain proportion of its maximum capacity (such as 80%), it is regarded as abnormal. The triggering condition of the optical modem offline alarm: When the connection between the optical modem and the OLT device is interrupted, that is, the optical modem is in an offline state, the optical modem offline alarm is triggered. It is determined whether the optical modem is offline by monitoring the online status flag of the optical modem or regularly sending a detection signal to the optical modem. Comprehensive alarm triggering condition: Taking into account multiple factors, a comprehensive alarm is triggered when multiple alarms occur at the same time or specific combination conditions are met. For example, when the Ping test alarm and the port abnormality alarm appear at the same time, or when the optical modem is offline and accompanied by a high packet loss rate, a comprehensive alarm is triggered.

[0109] After training and evaluation, the Naive Bayes model will classify and predict the fault mode of new data. Analyze its output results to determine whether there is a fault mode related to a specific alarm type. For example, if the model predicts a certain network fault mode, and this mode is related to the Ping measurement alarm, it may be necessary to further check whether the Ping measurement data meets the alarm triggering conditions. The partial least squares model is mainly used to analyze multi-factor problems. Check its output results to determine the degree of influence of each input variable on the output variable (such as equipment performance indicators or fault severity). If an input variable related to the alarm (such as port status parameters, optical modem connection status, etc.) has a greater impact on the output variable, the corresponding alarm needs to be triggered.

[0110] In data monitoring and comparison, for example, regularly obtain the response time and packet loss rate data of Ping test and compare them with the set threshold; monitor the port status in real time to determine whether there is an abnormality; check the online status of the optical modem. In alarm judgment and triggering, if the response time of Ping test exceeds the set threshold, the Ping test alarm is triggered; if the port status is abnormal, the port abnormality alarm is triggered; if the optical modem is offline, the optical modem offline alarm is triggered. In comprehensive alarm judgment, for example, when the Ping test alarm and the port abnormality alarm appear at the same time, the comprehensive alarm is triggered. Use logical judgment statements to implement the judgment of comprehensive alarms, such as "If the Ping test alarm is true and the port abnormality alarm is true, then trigger the comprehensive alarm." In alarm information generation, for example, for Ping test alarms, the alarm information can include "Ping test alarm, time: [specific time], device IP: [OLT device IP], response time: [actual response time], packet loss rate: [actual packet loss rate]".

[0111] Furthermore, the automatic processing notifies the operation and maintenance personnel of the generated alarm information and provides processing suggestions, including:

[0112] Determine the notification method based on the needs of the operation and maintenance personnel. When the alarm information is generated, send the alarm information to the operation and maintenance personnel according to the configured notification method;

[0113] Through association rule mining technology, the association relationship between alarms and processing methods is determined from historical alarm data and processing records. The frequent item sets in historical alarm data are mined through the Apriori algorithm to determine the alarm types and processing methods that appear together. The detailed information and processing results of each alarm are recorded to form a historical alarm database. When a new alarm occurs, the alarm case is determined by querying the historical database and corresponding processing suggestions are provided.

[0114] Specifically, use a variety of methods to notify the operation and maintenance personnel, such as email, SMS, instant messaging tools (such as WeChat for Enterprise, DingTalk, etc.) or special operation and maintenance management system notifications. According to the work habits and actual needs of the operation and maintenance team, select one or more notification methods to ensure that the alarm information can be conveyed to relevant personnel in a timely manner. When the alarm information is generated, send the alarm information to the operation and maintenance personnel according to the configured notification method. Ensure that the alarm information contains key information, such as alarm type, alarm time, device information, alarm details and handling suggestions.

[0115] Analyze historical alarm data and processing records to find out common processing methods for similar alarms. When new alarm information is generated, refer to the processing experience in historical data and provide processing suggestions. Use data mining techniques, such as association rule mining and cluster analysis, to find the correlation between alarms and processing methods from historical data. Use the Apriori algorithm to mine frequent item sets in historical alarm data to find out the alarm types and processing methods that often appear together, and provide references for new alarms. Establish a historical alarm database to record the detailed information and processing results of each alarm. When a new alarm occurs, query the historical database to find similar alarm cases and provide corresponding processing suggestions. In this way, in a system based on real-time monitoring of OLT equipment alarms, the function of notifying operation and maintenance personnel of alarm information and automatically providing processing suggestions can be realized, thereby improving operation and maintenance efficiency and reducing the impact of equipment failures on business.

[0116] Furthermore, the real-time status display, comprehensive data display, historical data query and data interaction include:

[0117] Through the front-end framework and data visualization library, the real-time status of the OLT device data continuously collected by the data collection module is displayed in real time. Different types of data are displayed through different charts. The front-end page is updated in real time through WebSocket technology. When the back-end collects new data, the data is immediately pushed to the front-end page.

[0118] Integrate and calculate the various collected data to generate comprehensive data indicators, process and calculate the data through the Pandas data analysis library, display the comprehensive data through different charts, and realize the interactive functions of data screening, time range selection and indicator switching through the front-end interactive technology of JavaScript event processing. Store the collected historical data in a relational database, design the back-end data query interface through RESTful API technology, and display the historical data in a visual way according to the user's query results.

[0119] Specifically, the real-time data of the OLT device is continuously collected through the data acquisition module, and the frequency of data collection is set to ensure the timeliness of the real-time status display. The front-end framework includes Vue.js or React.js; the data visualization library includes Echarts or D3.js. Different visualization methods are used for different types of data. For example, the device response time and packet loss rate are displayed with a line graph to show the change trend over time; the port status is represented by an icon or color (normal is green, abnormal is red); the optical modem connection status can be displayed in a list or chart to show the number of online and offline optical modems. Ensure the smoothness and stability of the real-time status display to avoid freezes or delays during data updates.

[0120] The generation of comprehensive data indicators, such as calculating comprehensive indicators such as average response time, average packet loss rate, port abnormality rate, optical modem offline rate, etc. The Pandas data analysis library ensures the accuracy and consistency of the data. Design an interface for comprehensive data display, present comprehensive indicators in a clear and concise manner, and use dashboards, bar charts, pie charts and other visual methods to display comprehensive data. For example, display the current value and status of key indicators through dashboards; use bar charts to compare the changes in comprehensive indicators in different time periods; use pie charts to display the proportion of port abnormality types. Add interactive functions to the comprehensive data display interface to facilitate users to analyze and explore data. The interactive functions include: realizing data filtering, time range selection, indicator switching and other functions. Use front-end interactive technology (such as JavaScript event processing or Ajax requests) to implement interactive functions, interact with back-end data services, and obtain user-selected data for display.

[0121] Store the collected historical data in the database for query and analysis. You can choose a relational database or a non-relational database for data storage. Index according to fields such as timestamp, device ID, and alarm type to increase query speed. Design a query interface through RESTful API or GraphQL technology. The interface supports multiple query conditions, such as time range, device ID, alarm type, etc., so that users can flexibly query the required historical data. Display the historical data in a visual way based on the user's query results. Use visualization tools and techniques similar to real-time status display and comprehensive data display. For example, for historical response time data, use a line chart to show the trend of changes in different time periods; for historical alarm data, use a list or chart to show the time, type, and device information of the alarm.

[0122] User interface design: The interface should include entrances to functional modules such as real-time status display, comprehensive data display, historical data query, and various interactive elements such as buttons, drop-down menus, input boxes, etc. Use front-end design principles and techniques to ensure the usability and aesthetics of the user interface. Implement the interactive logic through JavaScript front-end programming language to ensure the correctness and stability of the interactive logic and avoid errors or abnormal situations, such as validating the legitimacy of user input, processing error messages returned by the back-end, and providing clear feedback to users. Implement permission management through OAuth2 user authentication and authorization technology, and implement permission management functions based on the roles and permissions of different users. For example, administrator users can view the data of all devices and make system settings, while ordinary users can only view the device data they are responsible for, ensuring the security of the system and the confidentiality of data.

[0123] A method based on real-time monitoring of OLT device alarms, applied to a system based on real-time monitoring of OLT device alarms as described in any one of the above contents, comprises the following steps:

[0124] S1: Collect device response time, packet loss rate, OLT device status information and user disconnection data, and pre-process the collected data;

[0125] S2: Analyze and process the collected data to identify complex failure modes and multi-factor problems;

[0126] S3: Generates alarm information for ping test alarms, port abnormality alarms, optical modem offline alarms, and comprehensive alarms, notifies the operation and maintenance personnel of the generated alarm information through automatic processing, and provides handling suggestions;

[0127] S4: Real-time status display through data acquisition equipment, comprehensive data display, historical data query and data interaction through data analysis results.

[0128] The above implementation modes are merely descriptions of the preferred implementation modes of the present invention, and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary engineering and technical personnel in the field shall fall within the protection scope determined by the claims of the present invention.

Claims

1. A system based on real-time monitoring of OLT equipment alarms, characterized in that: include: The data acquisition module, data processing module, alarm module and monitoring interface module are connected by communication, wherein: The data acquisition module is used to collect device response time, packet loss rate, OLT device status information and user disconnection data, and pre-process the collected data; The data processing module is used to analyze and process the pre-processed data to identify complex failure modes and multi-factor problems; The alarm module includes a generating unit and a processing unit, wherein: The generation unit is used to generate alarm information for Ping test alarm, port abnormality alarm, optical modem offline alarm and comprehensive alarm; The processing unit is used to notify the operation and maintenance personnel of the generated alarm information through automatic processing and provide processing suggestions; The monitoring interface module is used for real-time status display, comprehensive data display, historical data query and data interaction.

2. A system based on real-time monitoring of OLT equipment alarms according to claim 1, characterized in that: The collecting device response time, packet loss rate, OLT device status information and user disconnection data, and pre-processing the collected data include: Ping records the time each request is sent and the time a response is received, and calculates the response time. The packet loss rate is calculated based on the number of packets sent and the number of responses received. Obtain detailed status information of the OLT device through the SNMP protocol or the NetConf protocol, wherein the detailed status of the OLT device includes the board status, the port status and the optical modem status; Obtain user disconnection data through the business system, including information about the user's connection status, disconnection time, and disconnection reason; The preprocessing includes data cleaning, data standardization and data aggregation.

3. A system based on real-time monitoring of OLT equipment alarms according to claim 1, characterized in that: The analysis and processing of the pre-processed data comprises the following steps: A1: Through the data acquisition module, in the DMI technical architecture environment, based on the prior knowledge of OLT equipment failure modes and multi-factor problems, relevant features are selected from the data and the data set is divided into a training set and a test set; A2: Train the Naive Bayes model with the training set, evaluate the trained Naive Bayes model with the test set, and use the trained Naive Bayes model to predict new data and identify possible fault modes. A3: Determine the input variables and output variables based on the multivariate data processed by partial least squares method. The input variables include the output results and related features of the naive Bayes model. Perform feature extraction on the input variables and divide the data set into a training set and a test set. A4: Train the PLS model with the training set, evaluate the performance of the PLS model with the test set, analyze multi-factor problems based on the trained PLS model, and determine the influence of each input variable on the output variable by looking at the regression coefficient of the model.

4. A system based on real-time monitoring of OLT equipment alarms according to claim 3, characterized in that: The method of training the naive Bayesian model by using the training set and evaluating the trained naive Bayesian model by using the test set comprises the following steps: A21: Calculate the prior probability of each fault category in the training set; A22: For each fault feature and fault category, calculate the conditional probability of the fault feature under the fault category; A23: When the fault feature is a discrete variable, the frequency of fault feature values ​​under the fault category is calculated as the first conditional probability; when the fault feature is a continuous variable, it is assumed that the fault feature follows the probability distribution of the normal distribution, the mean and standard deviation of the normal distribution are estimated through the training set data, and then the second conditional probability is calculated; A24: For each test data point, the posterior probability of belonging to each fault category is calculated according to the naive Bayes model. The test data point is assigned to the category with the highest probability according to the posterior probability, and compared with the true category label of the test set to calculate the evaluation index of the model.

5. A system based on real-time monitoring of OLT equipment alarms according to claim 4, characterized in that: The calculation formula of the prior probability is: Where: P(C j ) is the prior probability, C j is the fault category, is the fault category C in the training set j The number of data points, N train is the total number of data points in the training set; The calculation formula of the first conditional probability is: Where: X i is the fault characteristic, is the fault category C in the training set j And the fault feature X i The number of data points that take a specific value, is the fault category C in the training set j The number of data points; The calculation formula of the second conditional probability is: in: In fault category C j Fault feature X i The mean of In fault category C j Fault feature X i The variance of The posterior probability formula is: Where: n is the number of fault features, m is the number of fault categories.

6. A system based on real-time monitoring of OLT equipment alarms according to claim 3, characterized in that: The method of training the partial least squares model by using the training set data and evaluating the performance of the partial least squares model by using the test set data comprises the following steps: A41: Initialize the first component as a linear combination of the input variable matrix, determine the weight vector based on the maximum covariance between the first component and the output variable vector; calculate the regression coefficient of the first component on the output variable vector, and remove the influence of the first component from the input variable matrix to obtain a new matrix; A42: repeating step A41 until a preset number of components are extracted, and determining the number of extracted components according to cross-validation; A43: Use the test set data to calculate the root mean square error and determination coefficient to evaluate the performance of the partial least squares model; In step A42, the final regression model is Where y is the output variable, k is the number of components extracted, and t i It is an ingredient, c i Yes i The regression coefficient for y, e is the residual, and i is the index variable.

7. A system based on real-time monitoring of OLT equipment alarms according to claim 1, characterized in that: The generating of alarm information for Ping test alarm, port abnormality alarm, optical modem offline alarm and comprehensive alarm comprises the following steps: B1: Continuously monitor ping data, port status data, optical modem connection status data and comprehensive data, and compare the real-time data with the set alarm trigger conditions; B2: Based on the comparison results, determine whether the alarm triggering conditions are met. When the conditions are met, the corresponding alarm is triggered. When multiple alarms occur at the same time or specific combination conditions are met, a comprehensive alarm is triggered; B3: When an alarm is triggered, alarm information is generated, including: alarm type, alarm time, device information and processing suggestions, and the alarm information is stored in the database and log file.

8. The system based on real-time monitoring of OLT equipment alarm according to claim 1, characterized in that: The automatic processing notifies the operation and maintenance personnel of the generated alarm information and provides processing suggestions, including: Determine the notification method based on the needs of the operation and maintenance personnel. When the alarm information is generated, send the alarm information to the operation and maintenance personnel according to the configured notification method; Through association rule mining technology, the association relationship between alarms and processing methods is determined from historical alarm data and processing records. The frequent item sets in historical alarm data are mined through the Apriori algorithm to determine the alarm types and processing methods that appear together. The detailed information and processing results of each alarm are recorded to form a historical alarm database. When a new alarm occurs, the alarm case is determined by querying the historical database and corresponding processing suggestions are provided.

9. The system based on real-time monitoring of OLT equipment alarm according to claim 1, characterized in that: The real-time status display, comprehensive data display, historical data query and data interaction include: Through the front-end framework and data visualization library, the real-time status of the OLT device data continuously collected by the data collection module is displayed in real time. Different types of data are displayed through different charts. The front-end page is updated in real time through WebSocket technology. When the back-end collects new data, the data is immediately pushed to the front-end page. Integrate and calculate the various collected data to generate comprehensive data indicators, process and calculate the data through the Pandas data analysis library, display the comprehensive data through different charts, and realize the interactive functions of data screening, time range selection and indicator switching through the front-end interactive technology of JavaScript event processing. Store the collected historical data in a relational database, design the back-end data query interface through RESTful API technology, and display the historical data in a visual way according to the user's query results.

10. A method for real-time monitoring of OLT device alarms, applied to a system for real-time monitoring of OLT device alarms as claimed in any one of claims 1 to 9, characterized in that: The following steps are involved: S1: Collect device response time, packet loss rate, OLT device status information and user disconnection data, and pre-process the collected data; S2: Analyze and process the collected data to identify failure modes and multi-factor problems; S3: Generates alarm information for ping test alarms, port abnormality alarms, optical modem offline alarms, and comprehensive alarms, notifies the operation and maintenance personnel of the generated alarm information through automatic processing, and provides handling suggestions; S4: Real-time status display is performed through the data acquisition module, and comprehensive data display, historical data query and data interaction are performed through the data processing module.

Images