Software intelligent operation and maintenance method and system

Abstract

The invention discloses an intelligent software operation and maintenance method and system, relates to the technical field of software operation and maintenance, and aims to solve the technical problem that a current operation and maintenance mode is difficult to meet efficient and accurate requirements. Software data are collected from all components, a database and log files of a software system regularly or in real time through a data collection module, the software data comprise performance indexes, system logs and user behavior data, data preprocessing is carried out, preprocessing operation is carried out on the collected original data, the quality and consistency of the data are ensured, state monitoring is carried out, and the data processing efficiency is improved. Time sequence analysis is carried out through an ARIMA autoregressive moving average model, an automatic operation and maintenance process is adopted, multiple advanced algorithms are comprehensively applied, all-around operation and maintenance management from state monitoring to fault report generation is achieved, stable operation of the system is ensured, and the system reliability is improved. And the requirements of the software system on high efficiency and accuracy of operation and maintenance are guaranteed through all-around operation and maintenance management.

Description

Technical Field

[0001] The present invention relates to the field of software operation and maintenance technology, and more specifically, to a software intelligent operation and maintenance method and system. Background Art

[0002] Operation and maintenance, here refers to Internet operation and maintenance, which usually belongs to the technical department. Along with R&D, testing, and system management, they are the four major departments that provide technical support for Internet products. This division is somewhat different between domestic and foreign companies and between large and small companies. The process of generating an Internet product generally goes through: project establishment, demand analysis, development by the R&D department, testing by the testing department, deployment and release by the operation and maintenance department, and long-term operation and maintenance. Operation and maintenance is essentially the operation and maintenance of networks, servers, and services at all stages of their life cycle, to reach a consensus on acceptable conditions in terms of cost, stability, and efficiency.

[0003] At present, with the rapid development of information technology, the scale and complexity of software systems are increasing, and traditional operation and maintenance methods can no longer meet the needs of efficiency and accuracy. Therefore, it is particularly important to develop an intelligent operation and maintenance system that can automatically monitor system status, quickly locate and track the root cause of faults, and automatically generate operation and maintenance reports. In view of this, we propose a software intelligent operation and maintenance method and system. Summary of the invention

[0004] The purpose of the present invention is to provide a software intelligent operation and maintenance method and system to solve the technical problem that the current operation and maintenance methods are difficult to meet the requirements of high efficiency and accuracy.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions: a software intelligent operation and maintenance method, comprising the following steps:

[0006] S1. Condition monitoring process;

[0007] S101, data collection;

[0008] Use the data collection module to collect software data from various components, databases, and log files of the software system on a regular or real-time basis, where the software data includes performance indicators, system logs, and user behavior data;

[0009] S102, data preprocessing;

[0010] Pre-process the collected raw data to ensure data quality and consistency;

[0011] S103, status monitoring;

[0012] The ARIMA autoregressive moving average model is used to analyze time series, and then new observations are predicted by fitting historical data, and the prediction error is calculated. When the prediction error exceeds the preset threshold, an abnormal alarm is triggered;

[0013] S2, Fault tracing and location process;

[0014] S201, cluster analysis;

[0015] Use K-means to group abnormal data, and then iteratively optimize the cluster centers to minimize the objective function to analyze the clustering results and identify possible fault areas or fault modes;

[0016] S202, root cause analysis;

[0017] Combine Bayesian networks and causal reasoning to trace faults, calculate the posterior probability of each potential fault cause under given observation data, and then use the observed fault phenomenon to update the probability of each node in the network, so as to infer the most likely fault cause based on the updated probability distribution;

[0018] S3, fault tracing process;

[0019] Construct or update the fault propagation graph, and then use the graph traversal algorithm to simulate the fault propagation process in the system to calculate the fault propagation probability on each propagation path;

[0020] S4, automated report generation process;

[0021] Based on the results of fault tracing and tracking, an operation and maintenance report is automatically generated, and then natural language processing technology is used to generate the report text, and then visualization tools are used to display the fault analysis results and system status.

[0022] The present invention significantly improves operation and maintenance efficiency and reduces human intervention and errors by adopting automated operation and maintenance processes, and comprehensively uses a variety of advanced algorithms to improve the accuracy of fault detection and tracing, and realizes all-round operation and maintenance management from status monitoring to fault report generation, ensuring the stable operation of the system, and guaranteeing the software system's efficient and accurate operation and maintenance requirements through all-round operation and maintenance management.

[0023] Preferably, the performance indicators include CPU usage, memory occupancy, and response time.

[0024] Preferably, the preprocessing operations include cleaning, denoising, and formatting.

[0025] Preferably, the S103 status monitoring further includes the following steps:

[0026] S1031. Apply the ARIMA autoregressive moving average model to perform time series analysis to detect abnormal fluctuations in performance indicators. The ARIMA autoregressive moving average model is:

[0027]

[0028] Where, L is the lag operator;

[0029] p is the number of autoregressive terms;

[0030] q is the number of moving average items;

[0031] μ is the mean of the series (usually eliminated by differencing in ARIMA);

[0032] Z t is a white noise sequence;

[0033] S1032. Estimate the model parameter φ by fitting the historical data i ,θ i and mean μ;

[0034] S1033, using the estimated model to predict new observations and calculate the prediction error;

[0035] S1034. When the prediction error exceeds a preset threshold, an abnormal alarm is triggered.

[0036] Preferably, the cluster analysis in step S201 further includes the following steps:

[0037] S2011. K-means is used to group abnormal data to identify failure modes with similar characteristics. The objective function of K-means is:

[0038] minimize

[0039] Where k is the number of clusters;

[0040] C i is the i-th cluster;

[0041] μ i It is C i The mean of

[0042] x is the data point in the cluster;

[0043] S2012, minimize the objective function by iteratively optimizing the cluster centers;

[0044] S2013. Analyze the clustering results and identify possible fault areas or fault modes.

[0045] Preferably, the root cause analysis in step S202 further includes the following steps:

[0046] S2021. Combining Bayesian network and causal reasoning for fault tracing, the conditional probability formula in the Bayesian network is used to calculate the posterior probability of each potential fault cause under given observation data:

[0047]

[0048] Among them, A is the cause of the fault;

[0049] B is the observed fault phenomenon;

[0050] A' is a possible cause other than A;

[0051] S2022. Build or learn the structure and parameters of Bayesian networks;

[0052] S2023. Use the observed fault phenomenon to update the probability of each node in the network;

[0053] S2024. Infer the most likely cause of the failure based on the updated probability distribution.

[0054] Preferably, the S3 fault tracing process further includes the following steps:

[0055] S301, construct or update a fault propagation graph to reflect dependencies and fault propagation paths between system components;

[0056] S302, using a graph traversal algorithm to simulate the propagation process of the fault in the system;

[0057] S303, calculating the fault propagation probability on each propagation path, which usually involves calculating the product of the propagation probabilities of each node on the path:

[0058]

[0059] Among them, e is the edge on the path;

[0060] p(e) is the propagation probability of edge e.

[0061] Preferably, the S4 automated report generation process further includes the following steps:

[0062] S401, report content generation;

[0063] Automatically generate an operation and maintenance report based on the results of fault tracing and tracking, and then use natural language processing technology to generate the report text;

[0064] S402, visual display;

[0065] Use visualization tools such as charts, heat maps, and fault trees to display fault analysis results and system status to improve the readability and intuitiveness of reports.

[0066] Preferably, the S401 report content generation further comprises the following steps:

[0067] S4011. Automatically generate an operation and maintenance report containing fault description, impact scope, possible causes, and recommended repair contents based on the results of fault tracing and tracking;

[0068] S4012. Use natural language processing technology to generate easy-to-understand report text, where natural language processing technology includes template filling and text generation model.

[0069] A software intelligent operation and maintenance system includes a data acquisition module for periodically or in real time acquiring various types of data, a data processing module for processing the acquired data, an analysis and diagnosis module for comparing the acquired data with a preset threshold, an automatic implementation module for implementing measures after a fault, and a visual reporting module for generating reports, wherein the data acquisition module is electrically connected to the data processing module, the analysis and diagnosis module is electrically connected to the data processing module, the analysis and diagnosis module is electrically connected to the automatic implementation module, and the automatic implementation module is electrically connected to the visual reporting module.

[0070] Compared with the prior art, the present invention has the following beneficial effects:

[0071] The present invention significantly improves operation and maintenance efficiency and reduces human intervention and errors by adopting automated operation and maintenance processes, and comprehensively uses a variety of advanced algorithms to improve the accuracy of fault detection and tracing, and realizes all-round operation and maintenance management from status monitoring to fault report generation, ensuring the stable operation of the system, and guaranteeing the software system's efficient and accurate operation and maintenance requirements through all-round operation and maintenance management. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] Figure 1 It is a schematic block diagram of the software intelligent operation and maintenance system in the present invention. DETAILED DESCRIPTION

[0073] In order to facilitate those skilled in the art to understand the technical solution of the present invention, the technical solution of the present invention is further described in conjunction with the accompanying drawings of the specification.

[0074] Embodiment 1:

[0075] The present invention provides a software intelligent operation and maintenance method, comprising the following steps:

[0076] S1. Condition monitoring process;

[0077] S101, data collection;

[0078] Use the data collection module to collect software data from various components, databases, and log files of the software system on a regular or real-time basis, where the software data includes performance indicators, system logs, and user behavior data;

[0079] S102, data preprocessing;

[0080] Pre-process the collected raw data to ensure data quality and consistency;

[0081] S103, status monitoring;

[0082] The ARIMA autoregressive moving average model is used to analyze time series, and then new observations are predicted by fitting historical data, and the prediction error is calculated. When the prediction error exceeds the preset threshold, an abnormal alarm is triggered;

[0083] S2, Fault tracing and location process;

[0084] S201, cluster analysis;

[0085] Use K-means to group abnormal data, and then iteratively optimize the cluster center to minimize the objective function to analyze the clustering results and identify possible fault areas or fault modes;

[0086] S202, root cause analysis;

[0087] Combine Bayesian networks and causal reasoning to trace faults, calculate the posterior probability of each potential fault cause under given observation data, and then use the observed fault phenomenon to update the probability of each node in the network, so as to infer the most likely fault cause based on the updated probability distribution;

[0088] S3, fault tracing process;

[0089] Construct or update the fault propagation graph, and then use the graph traversal algorithm to simulate the fault propagation process in the system to calculate the fault propagation probability on each propagation path;

[0090] S4, automated report generation process;

[0091] Based on the results of fault tracing and tracking, an operation and maintenance report is automatically generated, and then natural language processing technology is used to generate the report text, and then visualization tools are used to display the fault analysis results and system status.

[0092] The present invention significantly improves operation and maintenance efficiency and reduces human intervention and errors by adopting automated operation and maintenance processes, and comprehensively uses a variety of advanced algorithms to improve the accuracy of fault detection and tracing, and realizes all-round operation and maintenance management from status monitoring to fault report generation, ensuring the stable operation of the system, and guaranteeing the software system's efficient and accurate operation and maintenance requirements through all-round operation and maintenance management.

[0093] In an embodiment of the present invention, the performance indicators include CPU usage, memory occupancy, and response time.

[0094] In an embodiment of the present invention, the pre-processing operation includes cleaning, denoising, and formatting.

[0095] In an embodiment of the present invention, the S103 status monitoring further includes the following steps:

[0096] S1031. Apply the ARIMA autoregressive moving average model to perform time series analysis to detect abnormal fluctuations in performance indicators. The ARIMA autoregressive moving average model is:

[0097]

[0098] Where, L is the lag operator;

[0099] p is the number of autoregressive terms;

[0100] q is the number of moving average items;

[0101] μ is the mean of the series (usually eliminated by differencing in ARIMA);

[0102] Z t is a white noise sequence;

[0103] S1032. Estimate the model parameter φ by fitting the historical data i ,θ i and mean μ;

[0104] S1033, using the estimated model to predict new observations and calculate the prediction error;

[0105] S1034. When the prediction error exceeds a preset threshold, an abnormal alarm is triggered.

[0106] In an embodiment of the present invention, the cluster analysis in step S201 further includes the following steps:

[0107] S2011. K-means is used to group abnormal data to identify failure modes with similar characteristics. The objective function of K-means is:

[0108] minimize

[0109] Where k is the number of clusters;

[0110] C i is the i-th cluster;

[0111] μ i It is C i The mean of

[0112] x is the data point in the cluster;

[0113] S2012, minimize the objective function by iteratively optimizing the cluster centers;

[0114] S2013. Analyze the clustering results and identify possible fault areas or fault modes.

[0115] In an embodiment of the present invention, the root cause analysis in step S202 further includes the following steps:

[0116] S2021. Combining Bayesian network and causal reasoning for fault tracing, the conditional probability formula in the Bayesian network is used to calculate the posterior probability of each potential fault cause under given observation data:

[0117]

[0118] Among them, A is the cause of the fault;

[0119] B is the observed fault phenomenon;

[0120] A' is a possible cause other than A;

[0121] S2022. Build or learn the structure and parameters of Bayesian networks;

[0122] S2023. Use the observed fault phenomenon to update the probability of each node in the network;

[0123] S2024. Infer the most likely cause of the failure based on the updated probability distribution.

[0124] In an embodiment of the present invention, the S3 fault tracing process further includes the following steps:

[0125] S301, construct or update a fault propagation graph to reflect dependencies and fault propagation paths between system components;

[0126] S302, using a graph traversal algorithm to simulate the propagation process of the fault in the system;

[0127] S303, calculating the fault propagation probability on each propagation path, which usually involves calculating the product of the propagation probabilities of each node on the path:

[0128]

[0129] Among them, e is the edge on the path;

[0130] p(e) is the propagation probability of edge e.

[0131] In an embodiment of the present invention, the S4 automated report generation process further includes the following steps:

[0132] S401, report content generation;

[0133] Automatically generate an operation and maintenance report based on the results of fault tracing and tracking, and then use natural language processing technology to generate the report text;

[0134] S402, visual display;

[0135] Use visualization tools such as charts, heat maps, and fault trees to display fault analysis results and system status to improve the readability and intuitiveness of reports.

[0136] In an embodiment of the present invention, the S401 report content generation further comprises the following steps:

[0137] S4011. Automatically generate an operation and maintenance report containing fault description, impact scope, possible causes, and recommended repair contents based on the results of fault tracing and tracking;

[0138] S4012. Use natural language processing technology to generate easy-to-understand report text, where natural language processing technology includes template filling and text generation model.

[0139] Embodiment 2:

[0140] like Figure 1 As shown, a software intelligent operation and maintenance system includes a data acquisition module for periodically or in real time acquiring various types of data, a data processing module for processing the acquired data, an analysis and diagnosis module for comparing the acquired data with a preset threshold, an automatic implementation module for implementing measures after a fault, and a visual reporting module for generating reports, wherein the data acquisition module is electrically connected to the data processing module, the analysis and diagnosis module is electrically connected to the data processing module, the analysis and diagnosis module is electrically connected to the automatic implementation module, and the automatic implementation module is electrically connected to the visual reporting module.

[0141] The embodiments of the present invention disclose preferred embodiments, but are not limited thereto. A person skilled in the art can easily understand the spirit of the present invention based on the above embodiments and make different extensions and changes. However, as long as they do not deviate from the spirit of the present invention, they are all within the protection scope of the present invention.

Claims

1. A software intelligent operation and maintenance method, characterized in that: The following steps are involved: S1. Condition monitoring process; S101, data collection; Use the data collection module to collect software data from various components, databases, and log files of the software system on a regular or real-time basis, where the software data includes performance indicators, system logs, and user behavior data; S102, data preprocessing; Pre-process the collected raw data to ensure data quality and consistency; S103, status monitoring; The ARIMA autoregressive moving average model is used to analyze time series, and then new observations are predicted by fitting historical data, and the prediction error is calculated. When the prediction error exceeds the preset threshold, an abnormal alarm is triggered; S2, Fault tracing and location process; S201, cluster analysis; Use K-means to group abnormal data, and then iteratively optimize the cluster centers to minimize the objective function to analyze the clustering results and identify possible fault areas or fault modes; S202, root cause analysis; Combine Bayesian networks and causal reasoning to trace faults, calculate the posterior probability of each potential fault cause under given observation data, and then use the observed fault phenomenon to update the probability of each node in the network, so as to infer the most likely fault cause based on the updated probability distribution; S3, fault tracing process; Construct or update the fault propagation graph, and then use the graph traversal algorithm to simulate the fault propagation process in the system to calculate the fault propagation probability on each propagation path; S4, automated report generation process; Based on the results of fault tracing and tracking, an operation and maintenance report is automatically generated, and then natural language processing technology is used to generate the report text, and then visualization tools are used to display the fault analysis results and system status.

2. A software intelligent operation and maintenance method according to claim 1, characterized in that: The performance indicators include CPU usage, memory usage, and response time.

3. A software intelligent operation and maintenance method according to claim 1, characterized in that: The pre-processing operations include cleaning, denoising, and formatting.

4. A software intelligent operation and maintenance method according to claim 1, characterized in that: The S103 status monitoring further includes the following steps: S1031. Apply the ARIMA autoregressive moving average model to perform time series analysis to detect abnormal fluctuations in performance indicators. The ARIMA autoregressive moving average model is: Where, L is the lag operator; p is the number of autoregressive terms; q is the number of moving average items; μ is the mean of the series (usually eliminated by differencing in ARIMA); Z t is a white noise sequence; S1032. Estimate the model parameter φ by fitting the historical data i ,θ i and mean μ; S1033, using the estimated model to predict new observations and calculate the prediction error; S1034: When the prediction error exceeds a preset threshold, an abnormal alarm is triggered.

5. A software intelligent operation and maintenance method according to claim 1, characterized in that: The S201 cluster analysis further comprises the following steps: S2011. K-means is used to group abnormal data to identify failure modes with similar characteristics. The objective function of K-means is: Minimize Where k is the number of clusters; C i is the i-th cluster; μ i It is C i The mean of x is the data point in the cluster; S2012, minimize the objective function by iteratively optimizing the cluster centers; S2013. Analyze the clustering results and identify possible fault areas or fault modes.

6. A software intelligent operation and maintenance method according to claim 1, characterized in that: The root cause analysis in step S202 further includes the following steps: S2021. Combining Bayesian network and causal reasoning for fault tracing, the conditional probability formula in the Bayesian network is used to calculate the posterior probability of each potential fault cause under given observation data: Among them, A is the cause of the fault; B is the observed fault phenomenon; A' is a possible cause other than A; S2022. Build or learn the structure and parameters of Bayesian networks; S2023. Use the observed fault phenomenon to update the probability of each node in the network; S2024. Infer the most likely cause of the failure based on the updated probability distribution.

7. A software intelligent operation and maintenance method according to claim 1, characterized in that: The S3 fault tracing process also includes the following steps: S301, construct or update a fault propagation graph to reflect dependencies and fault propagation paths between system components; S302, using a graph traversal algorithm to simulate the propagation process of the fault in the system; S303, calculating the fault propagation probability on each propagation path, which usually involves calculating the product of the propagation probabilities of each node on the path: Among them, e is the edge on the path; p(e) is the propagation probability of edge e.

8. A software intelligent operation and maintenance method according to claim 1, characterized in that: The S4 automated report generation process also includes the following steps: S401, report content generation; Automatically generate an operation and maintenance report based on the results of fault tracing and tracking, and then use natural language processing technology to generate the report text; S402, visual display; Use visualization tools such as charts, heat maps, and fault trees to display fault analysis results and system status to improve the readability and intuitiveness of reports.

9. A software intelligent operation and maintenance method according to claim 8, characterized in that: The S401 report content generation also includes the following steps: S4011. Automatically generate an operation and maintenance report containing fault description, impact scope, possible causes, and recommended repair contents based on the results of fault tracing and tracking; S4012. Use natural language processing technology to generate easy-to-understand report text, where natural language processing technology includes template filling and text generation model.

10. A software intelligent operation and maintenance system, used in a software intelligent operation and maintenance method according to claims 1 to 9, characterized in that: It includes a data acquisition module for collecting various types of data in a scheduled or real-time manner, a data processing module for processing the collected data, an analysis and diagnosis module for comparing the collected data with a preset threshold, an automatic implementation module for implementing measures after a fault, and a visual reporting module for generating reports. The data acquisition module is electrically connected to the data processing module, the analysis and diagnosis module is electrically connected to the data processing module, the analysis and diagnosis module is electrically connected to the automatic implementation module, and the automatic implementation module is electrically connected to the visual reporting module.

Images