Network transmission intelligent operation and maintenance inspection system

Abstract

This invention relates to the field of network operation and maintenance inspection technology, and in particular to an intelligent network transmission operation and maintenance inspection system, comprising: a data acquisition module for collecting data on the network transmission links between front-end devices deployed on highway sections and the service layer; a link feature extraction module for extracting features of the transmission quality parameters of the network transmission links corresponding to each front-end device; a status judgment module for judging the status of the link status labels of each front-end device; a fault location module for comparing the node's operating data with the historical operating data of the corresponding node to obtain the fault location result; and an inspection output module for analyzing the configuration parameters, status data, and fault link quality feature set to generate a network transmission operation and maintenance inspection report. This invention enables quality monitoring and fault location of the network transmission links between front-end devices and the service layer, improving the efficiency of network transmission operation and maintenance.

Description

Technical Field

[0001] This invention relates to the field of network operation and maintenance inspection technology, and in particular to an intelligent network transmission operation and maintenance inspection system. Background Technology

[0002] Existing network transmission systems typically employ a layered architecture, comprising front-end devices, network transmission links, and a service layer. Front-end devices are responsible for data acquisition and uploading, while the service layer handles data processing and service responses. Data transmission between the front-end devices and the service layer occurs via wired or wireless networks. To ensure overall system stability, maintenance primarily focuses on device status monitoring and service layer resource monitoring. However, existing technologies lack systematic quality monitoring methods for the network transmission links between the front-end devices and the service layer. These links often traverse multiple network nodes, including access networks, aggregation networks, and core networks, involving various transmission media and network protocols. In actual operation, the links may experience packet loss, increased latency, aggravated jitter, bandwidth limitations, or link interruptions. Because the monitoring systems for the front-end devices and the service layer are independent, and no dedicated detection and diagnostic mechanisms are deployed at the link level, maintenance personnel struggle to distinguish whether the source of the fault—the front-end devices, the transmission link, or the service layer—is at fault when service access anomalies or data interruptions occur. Link anomalies typically manifest as indirect application-layer errors or data loss. Fault localization requires collaborative investigation by multiple departments, which is time-consuming. This results in link faults not being detected and located in the first instance, and fault handling lags behind the occurrence of business impacts, extending the mean time to repair. With the expansion of network scale, the increase in the number of front-end devices, and the increasing real-time requirements of services, the impact of link quality on the overall system availability is becoming increasingly prominent. Existing operation and maintenance methods are unable to meet the needs of efficient and accurate operation and maintenance, limiting the improvement of network transmission operation and maintenance efficiency.

[0003] Chinese patent publication CN117975730A discloses a highway intelligent inspection system and method based on AR+AI technology. The system includes: an infrastructure layer configured to access front-end devices for comprehensive monitoring and perception of traffic conditions, completing the access and collection of multi-source data; a service layer configured to provide user management, image management, video tag configuration, and one-device-one-file services, supporting third-party system integration to ensure efficient user authentication, permission management, video preview and playback, device monitoring, tag customization and association, and the establishment and analysis of device data archives; and an application layer configured to integrate application management, video map and augmented reality technology applications, road property maintenance inspection, alarm linkage plans, and situational analysis functions, enabling efficient access and management of front-end devices, data backup and access, video tag display and retrieval, and automatic inspection and alarm handling. However, this solution only inspects the highway front-end devices and system application layer functions, failing to monitor the quality and locate faults in the network transmission link between the front-end devices and the service layer. This results in link faults not being detected and handled in a timely manner, reducing network transmission maintenance efficiency. Summary of the Invention

[0004] To address this issue, the present invention provides an intelligent network transmission operation and maintenance inspection system to overcome the problem in the prior art that it is impossible to monitor the quality and locate faults in the network transmission link between the front-end device and the service layer, resulting in the inability to detect and handle link faults in a timely manner and reducing the efficiency of network transmission operation and maintenance.

[0005] To achieve the above objectives, the present invention provides a network transmission intelligent operation and maintenance inspection system, comprising: The data acquisition module is used to collect data from the network transmission link between the front-end equipment deployed on the highway section and the service layer, and obtain raw link quality data including link status data and transport layer protocol message data. The link feature extraction module is used to preprocess the raw link quality data to obtain preprocessed raw link quality data, and to extract the transmission quality parameters of the network transmission link corresponding to each front-end device based on the preprocessed raw link quality data to obtain the link quality feature set of each front-end device. The status judgment module is used to judge the status of the link status label of each front-end device based on the link quality feature set of each front-end device and the preset link baseline feature library, and obtain the status judgment result. The link status label is used to identify whether the link is in a normal state or a fault state. The fault location module is used to determine the fault link quality feature set corresponding to the front-end device marked as faulty based on the status judgment result, extract the faulty network transmission link corresponding to the front-end device marked as faulty from the faulty link quality feature set, obtain the node operation data of each node in the faulty network transmission link, and compare the node operation data with the historical operation data of the corresponding node to obtain the fault location result containing the fault node identifier and fault type. The inspection output module is used to obtain the configuration parameters and status data corresponding to the fault node identifier, and analyze the configuration parameters, the status data and the fault link quality feature set to generate a network transmission operation and maintenance inspection report.

[0006] The technical principle of this application is as follows: By deploying a data acquisition module, real-time data is collected from the network transmission links between each front-end device and the service layer on a highway section, obtaining raw link quality data including link status data and transport layer protocol message data. A link feature extraction module preprocesses the raw link quality data to eliminate noise and outliers, and then extracts key parameters reflecting link transmission quality to form a link quality feature set for each front-end device. A status judgment module compares the link quality feature set of each device with a preset link baseline feature library, which stores the feature parameter ranges under normal link operation. Difference analysis is used to determine whether the link is in a normal or faulty state. For identified faulty links, a fault location module further extracts the operational data of each node in the network path, comparing the node operational data with their corresponding historical operational data node by node to locate the specific faulty node and its fault type. An inspection output module obtains the configuration parameters and status data of the faulty node, performs comprehensive analysis in conjunction with the link quality feature set, and finally generates a network transmission operation and maintenance inspection report.

[0007] Compared with existing technologies, the beneficial effects of this application are as follows: By collecting link status signaling data and transport layer protocol message data to obtain raw link quality data, it is possible to achieve refined monitoring of the network transmission link between the front-end device and the service layer. By preprocessing the raw link quality data and extracting the link quality feature set, the transmission quality status of each link can be accurately characterized. By comparing the link quality feature set with the preset link baseline feature library for status judgment, it is possible to quickly identify links in a faulty state and achieve automated and accurate identification of link anomalies. By extracting the operating data of each node for the faulty link and comparing it with historical operating data to obtain the fault node identifier and fault type, it is possible to locate the fault to a specific network node and the cause of the fault, which significantly shortens the fault investigation time. By obtaining the configuration parameters and status data corresponding to the faulty node and combining them with the link quality feature set for analysis to generate a network transmission operation and maintenance inspection report, it is possible to provide operation and maintenance personnel with complete diagnostic information covering link quality, faulty nodes, fault types and configuration status, improve the accuracy and response efficiency of operation and maintenance decisions, and ensure that the data collected by the front-end device is reliably transmitted to the service layer, thereby improving the overall operational stability and operation and maintenance efficiency of the highway intelligent inspection system.

[0008] Furthermore, the link feature extraction module includes: The feature extraction unit is used to extract the message interaction frequency, message length distribution entropy and message arrival interval variance of the network transmission link corresponding to each front-end device in the time series based on the transport layer protocol message data in the original link quality data, so as to obtain the flow-level statistical feature sequence. The spectral feature fusion unit is used to perform Fourier transform on the flow-level statistical feature sequence to obtain frequency domain energy spectrum features, and to concatenate the frequency domain energy spectrum features with the link state switching frequency in the link state data to obtain concatenated features. Principal component analysis is then performed on the concatenated features to reduce the dimensionality and obtain the link quality feature set for each front-end device.

[0009] In this scheme, by extracting the flow-level statistical feature sequence based on transport layer protocol message data, obtaining the frequency domain energy spectrum feature through Fourier transform and concatenating it with the link state data feature, and then performing principal component analysis to reduce dimensionality, it can accurately characterize the network transmission link quality of the front-end device, effectively reduce the feature dimensionality, and improve the effectiveness and accuracy of the feature representation.

[0010] Furthermore, the status determination module includes: The deviation vector calculation unit is used to calculate the Mahalanobis distance between the link quality feature set of each front-end device and the corresponding baseline feature vector in the preset link baseline feature library to obtain the first feature deviation vector. The mathematical expression is: In the formula, This represents the feature vector corresponding to the link quality feature set. This represents the mean vector of the corresponding feature vectors in the preset link baseline feature library. This represents the covariance matrix of the corresponding feature vector in the preset link baseline feature library. The feature vector representing the link quality feature set Mean vector of the corresponding feature vector in the preset link baseline feature library The transpose matrix of the differences between them; The status judgment unit is used to compare the first feature deviation vector with a preset first-level deviation threshold, and to judge the status of the link status label of each front-end device based on the comparison result, so as to obtain the status diagnosis result, wherein: When all dimension values ​​in the first feature deviation vector do not exceed the preset first-level deviation threshold, the link status label of the front-end device is determined to be in normal link status, and the normal link status is output as the status diagnosis result. When any dimension value in the first feature deviation vector exceeds a preset first-level deviation threshold, the link status label of the front-end device is determined to be in a pending confirmation state. The second feature deviation vector corresponding to the front-end device in the pending confirmation state is compared with a preset second-level deviation threshold. Based on the comparison result, the link status label of the front-end device in the pending confirmation state is judged to obtain a status diagnosis result. The preset second-level deviation threshold is numerically greater than the preset first-level deviation threshold, wherein: If any dimension value in the second feature deviation vector exceeds the preset second-level deviation threshold, the link status label of the front-end device is determined to be a link fault state, and the link fault state is output as the status diagnosis result. When all dimension values ​​in the second feature deviation vector do not exceed the preset second-level deviation threshold, the link status label of the front-end device is determined to be in normal link status, and the normal link status is output as the status diagnosis result.

[0011] In this solution, the first feature deviation vector between the link quality feature set and the baseline feature vector is calculated using Mahalanobis distance. Combined with two levels of progressively increasing deviation thresholds, a hierarchical comparison is performed. This allows for precise quantification of the degree of abnormal deviation of link features, enabling refined hierarchical judgment of the link status of front-end devices and significantly improving the accuracy of link status diagnosis.

[0012] Furthermore, the fault location module includes: The node trajectory reconstruction unit is used to reconstruct the network path topology based on the node routing identifier and link layer exchange identifier in the node operation data, so as to obtain a fault link topology structure containing node sequence and connection relationship between nodes. The node operation data aggregation unit is used to collect the node operation parameters corresponding to each node according to the fault link topology, and aggregate the node operation parameters of multiple types under the same node according to the parameter type to obtain the node operation parameter aggregation set of each node. The fault location unit is used to perform multi-parameter joint deviation analysis on the aggregated set of node operating parameters of each node and the aggregated set of historical operating parameters of the corresponding node to obtain fault location results including fault node identification and fault type.

[0013] In this solution, by reconstructing the topology of the faulty link, aggregating multiple types of operating parameters of each node, and performing multi-parameter joint deviation analysis between the aggregated parameter set and historical parameters, the faulty node identifier and fault type can be accurately identified, improving the accuracy and pertinence of fault location and enabling rapid and accurate troubleshooting of network faults.

[0014] Furthermore, the fault location unit calculates the node comprehensive deviation degree based on the aggregated set of node operating parameters of each node and the aggregated set of historical operating parameters of the corresponding node. Then, based on the node sequence in the faulty link topology, it performs a cumulative deviation comparison analysis between the current node's node comprehensive deviation degree and the node comprehensive deviation degrees of adjacent upstream nodes to obtain a fault location result containing the faulty node identifier and fault type. The node comprehensive deviation degree... The mathematical expression is: In the formula, This indicates the total number of time windows for the centralized collection of node operation parameters. This indicates the total number of parameter types included within each time window. This represents the real-time value of the j-th type of parameter within the k-th time window. This represents the standard value of the j-th type of parameter within the k-th time window of the historical operational parameter aggregation set. This represents the historical data standard deviation of the j-th parameter within the k-th time window. This represents the joint weight factor corresponding to the j-th type of parameter within the k-th time window. This represents the time interval between the k-th time window and the current time. This represents the time decay coefficient.

[0015] In this scheme, by introducing node comprehensive deviation calculation, multi-parameter weighted fusion and time decay mechanism, and combining node sequence to compare upstream and downstream cumulative deviation, fault nodes can be identified more accurately, the fault location accuracy and reliability can be improved, and the fault point can be quickly located.

[0016] Furthermore, the inspection output module includes: The parameter status extraction unit is used to retrieve the configuration parameters and status data corresponding to the fault node identifier from the configuration management database, and perform time-series slicing on the configuration parameters and status data according to the link quality feature set to obtain configuration status time-series data aligned with the fault occurrence time window. The correlation analysis unit is used to perform correlation analysis between the configuration status time series data and the link quality feature set through the correlation coefficient matrix to obtain the correlation analysis results; The structured output unit is used to filter configuration status data and link quality features whose absolute values ​​of correlation coefficients exceed preset correlation coefficient thresholds from the correlation analysis results, and then fill the filtered configuration status data, link quality features and corresponding correlation coefficients into a structured form according to a preset report template to generate a network transmission operation and maintenance inspection report.

[0017] In this solution, by retrieving the configuration and status data of the faulty node and performing time-series slicing, and combining it with the correlation coefficient matrix to conduct correlation analysis, high-correlation features are selected and output in a structured manner according to a template. This can quickly locate the cause of the fault, automatically generate a standardized operation and maintenance inspection report, and improve operation and maintenance efficiency and the accuracy of fault analysis.

[0018] Furthermore, in the correlation analysis unit, the mathematical expression for the correlation coefficient of the elements in the correlation coefficient matrix is: In the formula, Indicates the first Configuration status data and the first The correlation coefficient between the features Indicates the first Configuration status data at a given time point The value at that location, Indicates the first The average value of configuration status data within a time window. Indicates the first in the link quality feature set This feature at a point in time The value at that location, Indicates the first The mean of the feature within the time window, This indicates the total number of time points collected within the time window.

[0019] In this scheme, by using a correlation coefficient matrix to quantify the correlation between configuration status data and link quality characteristics, the linear correlation between configuration status data and link quality characteristics can be objectively reflected, thereby improving the accuracy of fault correlation analysis. Attached Figure Description

[0020] Figure 1This is a schematic diagram of the structure of a network transmission intelligent operation and maintenance inspection system according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the link feature extraction module in an embodiment of the present invention; Figure 3 This is a schematic diagram of the state determination module in an embodiment of the present invention; Figure 4 This is a schematic diagram of the fault location module in an embodiment of the present invention; Figure 5 This is a schematic diagram of the inspection output module in an embodiment of the present invention. Detailed Implementation

[0021] The following detailed description illustrates the specific implementation method: like Figure 1 As shown, it is a structural schematic diagram of a network transmission intelligent operation and maintenance inspection system according to an embodiment of the present invention, including: The data acquisition module is used to collect data from the network transmission link between the front-end equipment deployed on the highway section and the service layer, and obtain raw link quality data including link status data and transport layer protocol message data. The link feature extraction module is used to preprocess the raw link quality data to obtain preprocessed raw link quality data, and to extract the transmission quality parameters of the network transmission link corresponding to each front-end device based on the preprocessed raw link quality data to obtain the link quality feature set of each front-end device. The status judgment module is used to judge the status of the link status label of each front-end device based on the link quality feature set of each front-end device and the preset link baseline feature library, and obtain the status judgment result. The link status label is used to identify whether the link is in a normal state or a fault state. The fault location module is used to determine the fault link quality feature set corresponding to the front-end device marked as faulty based on the status judgment result, extract the faulty network transmission link corresponding to the front-end device marked as faulty from the faulty link quality feature set, obtain the node operation data of each node in the faulty network transmission link, and compare the node operation data with the historical operation data of the corresponding node to obtain the fault location result containing the fault node identifier and fault type. The inspection output module is used to obtain the configuration parameters and status data corresponding to the fault node identifier, and analyze the configuration parameters, the status data and the fault link quality feature set to generate a network transmission operation and maintenance inspection report.

[0022] In this embodiment, in the network transmission link between the front-end equipment deployed on the highway section and the service layer, the data acquisition module connects the network probe to the mirror port of the core switch through a bypass deployment method to capture all the original data packets carried on the link with zero copy. Based on deep packet inspection technology, link status signaling data, including PPP link negotiation signaling, LDP session keep-alive signaling, and BFD detection signaling, is parsed from the captured data packets. At the same time, transport layer protocol message data, including TCP retransmission messages, out-of-order messages, packet loss messages, and UDP jitter messages, are extracted. The acquisition process timestamps and aggregates the above data at the millisecond level, and finally forms the original link quality data containing link status data and transport layer protocol message data.

[0023] like Figure 2 As shown, it is a structural schematic diagram of the link feature extraction module in an embodiment of the present invention, including: The feature extraction unit is used to extract the message interaction frequency, message length distribution entropy and message arrival interval variance of the network transmission link corresponding to each front-end device in the time series based on the transport layer protocol message data in the original link quality data, so as to obtain the flow-level statistical feature sequence. The spectral feature fusion unit is used to perform Fourier transform on the flow-level statistical feature sequence to obtain frequency domain energy spectrum features, and to concatenate the frequency domain energy spectrum features with the link state switching frequency in the link state data to obtain concatenated features. Principal component analysis is then performed on the concatenated features to reduce the dimensionality and obtain the link quality feature set for each front-end device.

[0024] In this embodiment, in the feature extraction unit, based on the transport layer protocol message data in the original link quality data, for the network transmission link corresponding to each front-end device, a preset 10-second time window is divided according to a fixed time sequence. The total number of transport layer protocol message interactions within each time window is counted and divided by the time window duration to obtain the message interaction frequency. The length of all transport layer protocol messages within each time window is counted, 10 length intervals are divided, and the proportion of messages in each interval is calculated. The message length distribution entropy is obtained by the distribution entropy calculation method. The arrival time difference of adjacent transport layer protocol messages within each time window is recorded, and the dispersion of all time differences is calculated to obtain the message arrival interval variance. The message interaction frequency, message length distribution entropy, and message arrival interval variance corresponding to each time window are arranged in chronological order to obtain the flow-level statistical feature sequence corresponding to each front-end device.

[0025] In the spectral feature fusion unit, for each front-end device's flow-level statistical feature sequence, the three features in the flow-level statistical feature sequence—message interaction frequency, message length distribution entropy, and message arrival interval variance—are treated as independent time-domain signals. Each time-domain signal is then subjected to a Fourier transform to convert the flow-level statistical feature sequence in the time domain into a frequency-domain signal. The energy values ​​corresponding to each frequency component in the frequency-domain signal are extracted, and frequency components with energy values ​​lower than a preset energy threshold of 0.01 are removed. The energy values ​​of the remaining frequency components are arranged in ascending order of frequency to form the frequency-domain energy spectrum feature corresponding to each front-end device, ensuring that the frequency-domain energy spectrum feature completely preserves the frequency characteristics of the flow-level statistical feature sequence. The link state switching frequency of each front-end device's network transmission link is extracted from the link state signaling data. The total number of times the link state switches between different states such as connected, disconnected, and stuck within a 10-second period that coincides with the time window of the flow-level statistical feature sequence is counted to obtain the link state switching frequency. The frequency domain energy spectrum feature of each front-end device is concatenated with the corresponding link state switching frequency. Specifically, the link state switching frequency is treated as an independent feature and directly appended to the end of the frequency domain energy spectrum feature. This concatenated feature contains both the frequency energy information of the frequency domain energy spectrum feature and the state change information of the link state switching frequency, resulting in the final concatenated feature. The concatenated features are standardized by mapping all feature values ​​to the range of 0-1, eliminating dimensional differences between different features. Principal component analysis is then performed on the standardized concatenated features to calculate the covariance matrix of the features. The eigenvalues ​​and corresponding eigenvectors of the covariance matrix are then solved. Principal components with eigenvalues ​​greater than a preset feature threshold of 0.8 are selected. These principal components can retain more than 90% of the effective information of the concatenated features. Redundant features are removed, and the selected principal components are arranged in descending order of eigenvalues ​​to form the link quality feature set corresponding to each front-end device.

[0026] like Figure 3 As shown, it is a structural schematic diagram of the state determination module in an embodiment of the present invention, including: The deviation vector calculation unit is used to calculate the Mahalanobis distance between the link quality feature set of each front-end device and the corresponding baseline feature vector in the preset link baseline feature library to obtain the first feature deviation vector. The mathematical expression is: In the formula, This represents the feature vector corresponding to the link quality feature set. This represents the mean vector of the corresponding feature vectors in the preset link baseline feature library. This represents the covariance matrix of the corresponding feature vector in the preset link baseline feature library. The feature vector representing the link quality feature set Mean vector of the corresponding feature vector in the preset link baseline feature library The transpose matrix of the differences between them; The status judgment unit is used to compare the first feature deviation vector with a preset first-level deviation threshold, and to judge the status of the link status label of each front-end device based on the comparison result, so as to obtain the status diagnosis result, wherein: When all dimension values ​​in the first feature deviation vector do not exceed the preset first-level deviation threshold, the link status label of the front-end device is determined to be in normal link status, and the normal link status is output as the status diagnosis result. When any dimension value in the first feature deviation vector exceeds a preset first-level deviation threshold, the link status label of the front-end device is determined to be in a pending confirmation state. The second feature deviation vector corresponding to the front-end device in the pending confirmation state is compared with a preset second-level deviation threshold. Based on the comparison result, the link status label of the front-end device in the pending confirmation state is judged to obtain a status diagnosis result. The preset second-level deviation threshold is numerically greater than the preset first-level deviation threshold, wherein: If any dimension value in the second feature deviation vector exceeds the preset second-level deviation threshold, the link status label of the front-end device is determined to be a link fault state, and the link fault state is output as the status diagnosis result. When all dimension values ​​in the second feature deviation vector do not exceed the preset second-level deviation threshold, the link status label of the front-end device is determined to be in normal link status, and the normal link status is output as the status diagnosis result.

[0027] In this embodiment, when the deviation vector calculation unit is implemented, it first calls the preset link baseline feature library. The specific structure of the preset link baseline feature library is to store the set of baseline feature vectors collected multiple times by each front-end device under normal link conditions. The baseline feature vector corresponding to the current front-end device is extracted from it. The mean vector of the baseline feature vector and the covariance matrix of the corresponding feature vector in the preset link baseline feature library are calculated. The mean vector is the arithmetic mean of all baseline feature vectors of the front-end device under normal link conditions. The covariance matrix is ​​used to describe the correlation between the dimensions of the corresponding feature vector in the preset link baseline feature library. Then, the feature vector corresponding to the link quality feature set of the front-end device is obtained. The difference between the feature vector corresponding to the link quality feature set and the mean vector of the corresponding feature vector in the preset link baseline feature library is calculated. The transpose matrix of the difference is obtained. The transpose matrix is ​​multiplied by the inverse matrix of the covariance matrix of the corresponding feature vector in the preset link baseline feature library, and then multiplied by the above difference to finally obtain the first feature deviation vector of each front-end device.

[0028] When implementing the status judgment unit, a preset first-level deviation threshold of 4.5 and a preset second-level deviation threshold of 9.0 are first set, with the second-level deviation threshold being numerically greater than the first-level deviation threshold. Then, the first feature deviation vector of each front-end device is compared dimension-by-dimensionally with the preset first-level deviation threshold. The comparison examines the relationship between each dimension value in the first feature deviation vector and the preset first-level deviation threshold. When all dimension values ​​in the first feature deviation vector do not exceed the preset first-level deviation threshold of 4.5, the link status label of the front-end device is determined to be in a normal link state, and this normal link state is used as the status diagnosis result. The results are output as follows: When any dimension value in the first feature deviation vector exceeds the preset first-level deviation threshold of 4.5, the link status label of the front-end device is determined to be in a state to be confirmed. Then, the second feature deviation vector corresponding to the front-end device in the state to be confirmed is extracted, and each dimension value in the second feature deviation vector is compared with the preset second-level deviation threshold of 9.0. If any dimension value exceeds 9.0, the link status label of the front-end device is determined to be in a link fault state, and the link fault state is output as the status diagnosis result. When all dimension values ​​do not exceed 9.0, the link is determined to be in a normal state, and the link normal state is output as the status diagnosis result.

[0029] like Figure 4 As shown, it is a structural schematic diagram of the fault location module in an embodiment of the present invention, including: The node trajectory reconstruction unit is used to reconstruct the network path topology based on the node routing identifier and link layer exchange identifier in the node operation data, so as to obtain a fault link topology structure containing node sequence and connection relationship between nodes. The node operation data aggregation unit is used to collect the node operation parameters corresponding to each node according to the fault link topology, and aggregate the node operation parameters of multiple types under the same node according to the parameter type to obtain the node operation parameter aggregation set of each node. The fault location unit is used to perform multi-parameter joint deviation analysis on the aggregated set of node operating parameters of each node and the aggregated set of historical operating parameters of the corresponding node to obtain fault location results including fault node identification and fault type.

[0030] Specifically, the fault location unit calculates the node comprehensive deviation degree based on the aggregated set of node operating parameters of each node and the aggregated set of historical operating parameters of the corresponding node. Then, based on the node sequence in the faulty link topology, it performs a cumulative deviation comparison analysis between the current node's node comprehensive deviation degree and the node comprehensive deviation degrees of adjacent upstream nodes to obtain a fault location result containing the faulty node identifier and fault type. The node comprehensive deviation degree... The mathematical expression is: In the formula, This indicates the total number of time windows for the centralized collection of node operation parameters. This indicates the total number of parameter types included within each time window. This represents the real-time value of the j-th type of parameter within the k-th time window. This represents the standard value of the j-th type of parameter within the k-th time window of the historical operational parameter aggregation set. This represents the historical data standard deviation of the j-th parameter within the k-th time window. This represents the joint weight factor corresponding to the j-th type of parameter within the k-th time window. This represents the time interval between the k-th time window and the current time. This represents the time decay coefficient.

[0031] In this embodiment, when the node operation data aggregation unit is implemented, it uses the fault link topology obtained by the node trajectory reconstruction unit as a basis to determine all nodes in the fault link topology. It collects multiple types of node operation parameters for each node in the fault link topology, specifically including node CPU utilization, memory usage, link bandwidth utilization, and data packet loss rate. The collection cycle is set to 5 seconds / time, and the data is collected continuously for 30 times. Then, the multiple types of node operation parameters collected under the same node are aggregated according to the parameter type. The average value of the node operation parameters of the same type is taken, and abnormal fluctuation data is removed to form a node operation parameter aggregation set for each node, ensuring that the node operation parameter aggregation set can fully reflect the real-time operation status of each node.

[0032] When the fault location unit is implemented, it first calls the historical operating parameter aggregation set of each node. The specific structure of the historical operating parameter aggregation set is the aggregation data set of node operating parameters under normal operating conditions of each node in the past 30 days. The node operating parameter aggregation set of each node is compared with the corresponding historical operating parameter aggregation set of the corresponding node. The deviation of each type of parameter in the node operating parameter aggregation set is compared with the corresponding type of parameter in the historical operating parameter aggregation set. The deviation threshold is set to 15%. When the deviation of any type of parameter exceeds 15%, the node is determined to be a fault node. The fault type is determined by combining the parameter deviation type. Finally, the fault location result containing the fault node identifier and the fault type is obtained.

[0033] like Figure 5 As shown, it is a structural schematic diagram of the inspection output module in an embodiment of the present invention, including: The parameter status extraction unit is used to retrieve the configuration parameters and status data corresponding to the fault node identifier from the configuration management database, and perform time-series slicing on the configuration parameters and status data according to the link quality feature set to obtain configuration status time-series data aligned with the fault occurrence time window. The correlation analysis unit is used to perform correlation analysis between the configuration status time series data and the link quality feature set through the correlation coefficient matrix to obtain the correlation analysis results; The structured output unit is used to filter configuration status data and link quality features whose absolute values ​​of correlation coefficients exceed preset correlation coefficient thresholds from the correlation analysis results, and then fill the filtered configuration status data, link quality features and corresponding correlation coefficients into a structured form according to a preset report template to generate a network transmission operation and maintenance inspection report.

[0034] Specifically, in the correlation analysis unit, the mathematical expression for the correlation coefficient of the elements in the correlation coefficient matrix is: In the formula, Indicates the first Configuration status data and the first The correlation coefficient between the features Indicates the first Configuration status data at a given time point The value at that location, Indicates the first The average value of configuration status data within a time window. Indicates the first in the link quality feature set This feature at a point in time The value at that location, Indicates the first The mean of the feature within the time window, This indicates the total number of time points collected within the time window.

[0035] In this embodiment, when the parameter status extraction unit is implemented, it first accesses the configuration management database. The specific structure of the configuration management database is to store the configuration parameters and status data of each node according to the fault node identifier. The configuration parameters and status data corresponding to the fault node identifier in the fault location result are retrieved from it. The configuration parameters include node port configuration and routing protocol parameters, and the status data includes the real-time running status of the node. Then, the link quality feature set corresponding to the front-end device is obtained. According to the time dimension corresponding to the link quality feature set, the retrieved configuration parameters and status data are time-series sliced. The fault occurrence time window is set to 10 seconds. After slicing, the data aligned with the 10-second fault occurrence time window is retained, and finally the configuration status time-series data aligned with the fault occurrence time window is obtained.

[0036] When implementing the correlation analysis unit, the configuration status time-series data obtained from the parameter status extraction unit and the link quality feature set corresponding to the front-end device are first acquired. The configuration status time-series data and the link quality feature set are taken as two analysis objects. Correlation analysis is performed on the configuration status time-series data and the link quality feature set through a correlation coefficient matrix. A correlation coefficient matrix containing each dimension of the configuration status time-series data and each dimension of the link quality feature set is constructed. The degree of correlation between the two is reflected by matrix calculation. The focus is on analyzing the correlation between changes in configuration status time-series data and fluctuations in the link quality feature set. Unrelated redundant data is eliminated, and finally, the correlation analysis results that can reflect the degree of correlation between the two are obtained.

[0037] When implementing the structured output unit, a preset correlation coefficient threshold of 0.7 is first set. Configuration status data and link quality features with an absolute correlation coefficient exceeding 0.7 are selected from the correlation analysis results to ensure that the selected content has a significant correlation. Then, a preset report template is called. The specific structure of the preset report template includes four core modules: fault node identifier, associated configuration status data, associated link quality features, and corresponding correlation coefficient. The selected configuration status data, link quality features, and corresponding correlation coefficients are structured and filled in according to the requirements of each module of the template. After supplementing the basic operation and maintenance information, a complete network transmission operation and maintenance inspection report is generated.

[0038] The above are merely embodiments of the present invention. Commonly known structures and characteristics are not described in detail here. Those skilled in the art are aware of all common technical knowledge in the field prior to the application date or priority date, are aware of all existing technologies in that field, and have the ability to apply conventional experimental methods prior to that date. Those skilled in the art can, under the guidance of this application, improve and implement this solution in combination with their own capabilities. Some typical known structures or methods should not be obstacles for those skilled in the art to implement this application. It should be noted that those skilled in the art can make several modifications and improvements without departing from the structure of the present invention. These should also be considered within the scope of protection of the present invention, and will not affect the effectiveness of the implementation of the present invention or the practicality of the patent. The scope of protection claimed in this application should be determined by the content of its claims, and the specific embodiments described in the specification can be used to interpret the content of the claims.

Claims

1. A network transmission intelligent operation and maintenance inspection system, characterized in that: include: The data acquisition module is used to collect data from the network transmission link between the front-end equipment deployed on the highway section and the service layer, and obtain raw link quality data including link status data and transport layer protocol message data. The link feature extraction module is used to preprocess the raw link quality data to obtain preprocessed raw link quality data, and to extract the transmission quality parameters of the network transmission link corresponding to each front-end device based on the preprocessed raw link quality data to obtain the link quality feature set of each front-end device. The status judgment module is used to judge the status of the link status label of each front-end device based on the link quality feature set of each front-end device and the preset link baseline feature library, and obtain the status judgment result. The link status label is used to identify whether the link is in a normal state or a fault state. The fault location module is used to determine the fault link quality feature set corresponding to the front-end device marked as faulty based on the status judgment result, extract the faulty network transmission link corresponding to the front-end device marked as faulty from the faulty link quality feature set, obtain the node operation data of each node in the faulty network transmission link, and compare the node operation data with the historical operation data of the corresponding node to obtain the fault location result containing the fault node identifier and fault type. The inspection output module is used to obtain the configuration parameters and status data corresponding to the fault node identifier, and analyze the configuration parameters, the status data and the fault link quality feature set to generate a network transmission operation and maintenance inspection report.

2. The intelligent network transmission operation and maintenance inspection system according to claim 1, characterized in that: The link feature extraction module includes: The feature extraction unit is used to extract the message interaction frequency, message length distribution entropy and message arrival interval variance of the network transmission link corresponding to each front-end device in the time series based on the transport layer protocol message data in the original link quality data, so as to obtain the flow-level statistical feature sequence. The spectral feature fusion unit is used to perform Fourier transform on the flow-level statistical feature sequence to obtain frequency domain energy spectrum features, and to concatenate the frequency domain energy spectrum features with the link state switching frequency in the link state data to obtain concatenated features. Principal component analysis is then performed on the concatenated features to reduce the dimensionality and obtain the link quality feature set for each front-end device.

3. The intelligent operation and maintenance inspection system for network transmission according to claim 1, characterized in that: The status determination module includes: The deviation vector calculation unit is used to calculate the Mahalanobis distance between the link quality feature set of each front-end device and the corresponding baseline feature vector in the preset link baseline feature library to obtain the first feature deviation vector. The mathematical expression is: In the formula, This represents the feature vector corresponding to the link quality feature set. This represents the mean vector of the corresponding feature vectors in the preset link baseline feature library. This represents the covariance matrix of the corresponding feature vector in the preset link baseline feature library. The feature vector representing the link quality feature set Mean vector of the corresponding feature vector in the preset link baseline feature library The transpose matrix of the differences between them; The status judgment unit is used to compare the first feature deviation vector with a preset first-level deviation threshold, and to judge the status of the link status label of each front-end device based on the comparison result, so as to obtain the status diagnosis result, wherein: When all dimension values ​​in the first feature deviation vector do not exceed the preset first-level deviation threshold, the link status label of the front-end device is determined to be in normal link status, and the normal link status is output as the status diagnosis result. When any dimension value in the first feature deviation vector exceeds a preset first-level deviation threshold, the link status label of the front-end device is determined to be in a pending confirmation state. The second feature deviation vector corresponding to the front-end device in the pending confirmation state is compared with a preset second-level deviation threshold. Based on the comparison result, the link status label of the front-end device in the pending confirmation state is judged to obtain a status diagnosis result. The preset second-level deviation threshold is numerically greater than the preset first-level deviation threshold, wherein: If any dimension value in the second feature deviation vector exceeds the preset second-level deviation threshold, the link status label of the front-end device is determined to be a link fault state, and the link fault state is output as the status diagnosis result. When all dimension values ​​in the second feature deviation vector do not exceed the preset second-level deviation threshold, the link status label of the front-end device is determined to be in normal link status, and the normal link status is output as the status diagnosis result.

4. The intelligent operation and maintenance inspection system for network transmission according to claim 1, characterized in that: The fault location module includes: The node trajectory reconstruction unit is used to reconstruct the network path topology based on the node routing identifier and link layer exchange identifier in the node operation data, so as to obtain a fault link topology structure containing node sequence and connection relationship between nodes. The node operation data aggregation unit is used to collect the node operation parameters corresponding to each node according to the fault link topology, and aggregate the node operation parameters of multiple types under the same node according to the parameter type to obtain the node operation parameter aggregation set of each node. The fault location unit is used to perform multi-parameter joint deviation analysis on the aggregated set of node operating parameters of each node and the aggregated set of historical operating parameters of the corresponding node to obtain fault location results including fault node identification and fault type.

5. The intelligent network transmission operation and maintenance inspection system according to claim 4, characterized in that: The fault location unit calculates the node comprehensive deviation degree based on the aggregated set of node operating parameters of each node and the aggregated set of historical operating parameters of the corresponding node. Then, based on the node sequence in the faulty link topology, it performs a cumulative deviation comparison analysis between the current node's node comprehensive deviation degree and the node comprehensive deviation degrees of adjacent upstream nodes to obtain a fault location result containing the faulty node identifier and fault type. The node comprehensive deviation degree... The mathematical expression is: In the formula, This indicates the total number of time windows for the centralized collection of node operation parameters. This indicates the total number of parameter types included within each time window. This represents the real-time value of the j-th type of parameter within the k-th time window. This represents the standard value of the j-th type of parameter within the k-th time window of the historical operational parameter aggregation set. This represents the standard deviation of the historical data for the j-th parameter within the k-th time window. This represents the joint weight factor corresponding to the j-th type of parameter within the k-th time window. This represents the time interval between the k-th time window and the current time. This represents the time decay coefficient.

6. The intelligent operation and maintenance inspection system for network transmission according to claim 1, characterized in that: The inspection output module includes: The parameter status extraction unit is used to retrieve the configuration parameters and status data corresponding to the fault node identifier from the configuration management database, and perform time-series slicing on the configuration parameters and status data according to the link quality feature set to obtain configuration status time-series data aligned with the fault occurrence time window. The correlation analysis unit is used to perform correlation analysis between the configuration status time series data and the link quality feature set through the correlation coefficient matrix to obtain the correlation analysis results; The structured output unit is used to filter configuration status data and link quality features whose absolute values ​​of correlation coefficients exceed preset correlation coefficient thresholds from the correlation analysis results, and then fill the filtered configuration status data, link quality features and corresponding correlation coefficients into a structured form according to a preset report template to generate a network transmission operation and maintenance inspection report.

7. The intelligent operation and maintenance inspection system for network transmission according to claim 6, characterized in that: In the association analysis unit, the mathematical expression for the correlation coefficient of the elements in the correlation coefficient matrix is: In the formula, Indicates the first Configuration status data and the first The correlation coefficient between the features Indicates the first Configuration status data at a given time point The value at that location, Indicates the first The average value of configuration status data within a time window. Indicates the first in the link quality feature set This feature at a point in time The value at that location, Indicates the first The mean of the feature within the time window, This indicates the total number of time points collected within the time window.

Images