Method and system for judging fault type of communication network management and control system

Abstract

The invention relates to the technical field of communication networks, in particular to a communication network management and control system fault type judgment method and system, and the method comprises the steps: obtaining and carrying out the characterization processing of network operation data of a communication network management and control system, and obtaining network operation parameters; obtaining information entropies of the network operation parameters, and screening the network operation parameters based on the information entropies of the network operation parameters to obtain target network operation parameters; obtaining a fault occurrence moment of the network management and control system, and extracting m groups of data of the target network operation parameters before and after the fault occurrence moment to generate a target network operation parameter sequence; based on the box dimension and the Spearman rank correlation coefficient of the target network operation parameter sequence, obtaining the target network operation parameter sequence after feature extraction; and inputting the target network operation parameter sequence after feature extraction into an SVM fault classification model to obtain a fault type. According to the invention, by simplifying the data set of the network management system, the efficiency of network performance evaluation and fault location is improved.

Description

Technical Field

[0001] This invention relates to the field of communication network technology, specifically to a method and system for determining fault types in a communication network management and control system. Background Technology

[0002] In the field of communication network operation and maintenance management, existing technologies have addressed the issues of high redundancy and insufficient representation of key information in network status data collected from multi-source heterogeneous communication devices. Some solutions have been developed. For example, Chinese patent application CN202411699415.2 proposes a method and system for fault location and tracking in communication network operation and maintenance. This method uses a fault feature knowledge base for fault location and diagnosis. However, it has two shortcomings: first, it does not fully consider the negative impact of data redundancy on fault location efficiency; second, it fails to comprehensively analyze key status information that significantly affects network operation, such as device power status, fan status, CPU utilization, and temperature. Chinese patent application CN202111681649.0 proposes a method and apparatus for locating service faults in optical transmission networks. It identifies anomalies by investigating target information and querying fault parameters. Its core problem is that it fails to effectively cope with the complexity of data characteristics in network operation and maintenance scenarios. Specifically, it lacks a mechanism to distinguish between data with significant values ​​but limited actual reference value and high-value data with small values ​​but significant fluctuation characteristics. It relies solely on human experience to set a uniform numerical threshold to judge abnormal data, resulting in a highly subjective and inaccurate screening process. Summary of the Invention

[0003] The purpose of this invention is to provide a method and system for judging the fault type of a communication network management and control system in order to solve at least one of the above-mentioned technical problems, so as to help operation and maintenance personnel better screen high-value abnormal data and conduct network fault analysis and status assessment, promptly discover key factors that lead to faults, and improve the overall network maintenance quality.

[0004] The present invention achieves the above objectives through the following technical solutions: A method for determining fault types in a communication network management and control system, comprising: Obtain network operation data from the communication network management and control system, and perform feature processing on the network operation data to obtain network operation parameters; Obtain the information entropy of each network operating parameter, and filter the network operating parameters based on the information entropy of each network operating parameter to obtain the target network operating parameters; Obtain the time of failure of the network management system, extract m sets of data of the target network operation parameters before and after the time of failure to generate the target network operation parameter sequence; where m is a natural number greater than or equal to 1; Based on the box dimension and Spearman rank correlation coefficient of the target network operating parameter sequence, the target network operating parameter sequence after feature extraction is obtained; The target network running parameter sequence after feature extraction is input into a pre-trained SVM fault classification model to obtain the fault type.

[0005] Furthermore, the network operation data includes: device health status parameters and network link performance parameters; The device health status parameters include: clock synchronization status, device power supply status, heat dissipation system status, processor load rate, device temperature index, and engine redundancy status. The network link performance parameters include: transmission latency, channel occupancy, and packet loss rate.

[0006] Furthermore, the network operation data is characterized, including: standardizing and scoring each network operation data point, with a score range of 0-1.

[0007] Furthermore, the target network operating parameters are obtained by filtering the network operating parameters based on the information entropy of each network operating parameter, including: sorting the information entropy of each network operating parameter in descending order, and retaining the top k network operating parameters as the target network operating parameters; where k is a natural number less than or equal to 5.

[0008] Furthermore, the target network operating parameters are extracted from m sets of data before and after the fault occurrence time to generate a target network operating parameter sequence, including: Five sets of data were acquired before and after the time of the fault occurrence as the target network operating parameter sequence, with a sampling interval of 5 minutes between each two adjacent sets of target network operating parameter sequences.

[0009] Furthermore, the classification function of the SVM fault classification model The calculation formula is: in, Represents support vectors; Indicates the number of fault types; Represents the Lagrange multipliers; Represents the kernel function. This represents the sequence of target network operating parameters after feature extraction. , These represent the first and second support vectors in a specific category, respectively. This represents the classification threshold.

[0010] Furthermore, the fault types include: optical equipment failure, network configuration error, line blockage, equipment hardware failure, CPU overload, equipment frequency reduction, clock oscillation, and other fault causes.

[0011] A fault type determination system for a communication network management and control system, comprising: The acquisition module is used to acquire network operation data from the communication network management and control system, and to perform feature processing on the network operation data to obtain network operation parameters. The parameter filtering module is used to obtain the information entropy of each network operating parameter, and to filter the network operating parameters based on the information entropy of each network operating parameter to obtain the target network operating parameters; The feature extraction module is used to obtain the time of failure of the network management system, extract the target network operating parameters before and after the time of failure, and generate the target network operating parameter sequence; where m is a natural number greater than or equal to 1. The feature extraction module is also used to obtain the target network operating parameter sequence after feature extraction based on the box dimension and Spearman rank correlation coefficient of the target network operating parameter sequence; The SVM judgment module is used to input the target network running parameter sequence after feature extraction into the pre-trained SVM fault classification model to obtain the fault type.

[0012] An electronic device includes a processor, a memory, and a computer program stored in the memory and running on the processor. When the computer program is executed by the processor, it implements the fault type determination method for a communication network management system as described above.

[0013] A computer-readable storage medium storing a computer program, which, when executed by a processor, implements the fault type determination method for a communication network management and control system as described above.

[0014] The beneficial effects of this invention are as follows: This invention quantifies the degree of data change in network operating parameters through information entropy, dynamically filters out high-value parameters with drastic changes (Top-k and k≤5), combines Spearman rank correlation coefficient for redundancy removal with cross-validation with an expert whitelist, effectively solves the evaluation interference problem caused by data redundancy in communication networks, and improves the efficiency of key feature extraction. Based on 5 sets of time series before and after the fault time, with an interval of 5 minutes, it accurately portrays the parameter change pattern before and after the fault through dual feature extraction of box dimension (quantifying data complexity) and Spearman rank correlation coefficient (quantifying parameter correlation), solving the technical problem of unclear judgment of the degree of system state change. Finally, it uses a one-to-many strategy SVM fault classification model (using RBF kernel function) to efficiently classify small sample, nonlinear network management data, achieving accurate judgment of 7 types of faults such as optical equipment faults and CPU overload. The overall solution simplifies the processing dimensions while maintaining data quality, significantly improves the efficiency of network fault location and the accuracy of operation and maintenance decisions, and provides reliable technical support for intelligent operation and maintenance in complex heterogeneous network environments. Attached Figure Description

[0015] Figure 1 This is a schematic flowchart of a fault type determination method for a communication network management and control system according to one embodiment of the present invention; Figure 2 This is a schematic diagram of a fault type determination method for a communication network management and control system according to another embodiment of the present invention; Figure 3 This is a schematic diagram of the fault type judgment system of a communication network management and control system according to one embodiment of the present invention; Figure 4 This is a schematic diagram illustrating the principle of a feature extraction module according to one embodiment of the present invention. Detailed Implementation

[0016] The invention will now be discussed with reference to exemplary embodiments. It should be understood that the described embodiments are merely intended to enable those skilled in the art to better understand and thus implement the invention, and are not intended to imply any limitation on the scope of the invention.

[0017] As used herein, the term "comprising" and its variations are to be interpreted as open-ended terms meaning "including but not limited to". The term "based on" is to be interpreted as "at least partially based on". The terms "one embodiment" and "an embodiment" are to be interpreted as "at least one embodiment".

[0018] Example 1 Figure 1 This is a schematic flowchart of a fault type determination method for a communication network management and control system according to one embodiment of the present invention. Figure 1 As shown, according to one embodiment of the present invention, a method for determining the fault type of a communication network management and control system includes the following steps: Step S102: Obtain network operation data from the communication network management and control system, and perform feature processing on the network operation data to obtain network operation parameters; Step S104: Obtain the information entropy of each network operating parameter, and filter the network operating parameters based on the information entropy of each network operating parameter to obtain the target network operating parameters; Step S106: Obtain the time of the network control system failure, and extract m sets of data before and after the time of the failure to generate a target network operation parameter sequence; where m is a natural number greater than or equal to 1. Step S108: Based on the box dimension and Spearman rank correlation coefficient of the target network running parameter sequence, obtain the target network running parameter sequence after feature extraction; Step S110: Input the target network running parameter sequence after feature extraction into the pre-trained SVM fault classification model to obtain the fault type.

[0019] In this embodiment, addressing the issues of high data redundancy, difficulty in feature extraction, and weak model generalization in fault location of communication network management and control systems, a complete process of "data filtering - feature extraction - classification decision" is proposed. First, in step S102, network operation data (such as traffic, latency, packet loss rate, CPU / memory usage, etc.) from the integrated network management center and various professional network management subsystems are acquired in real time. These heterogeneous, multi-source data are preprocessed, including normalization, missing value imputation, and sliding window smoothing, to obtain network operation parameters with uniform dimensions. Next, in step S104, information entropy is used to quantify the uncertainty of each dimension parameter, filtering out low-change or noisy parameters with information entropy below a threshold, retaining only the "target network operation parameters" sensitive to faults, thereby reducing subsequent computation while preserving data quality. Subsequently, in step S106, the system accurately locates the fault occurrence time based on maintenance logs or alarm timestamps, and uses this time as the center, extracting m sets (m≥1) of data points before and after it to construct the target network operation parameter sequence, providing a temporal context for subsequent time-series feature analysis. In step S108, dual feature extraction is performed using the box dimension from fractal theory (describing sequence complexity) and the Spearman rank correlation coefficient from statistics (characterizing nonlinear correlation between parameters) to obtain feature vectors that combine dynamic complexity and correlation stability. This step calculates the local box dimension through a sliding window and combines significance testing to eliminate spurious correlations, further suppressing the influence of outliers. Finally, in step S110, the feature vector is input into a pre-trained SVM fault classification model (the kernel function is preferably RBF, and the hyperparameters are determined through grid search and cross-validation), and the model outputs fault type labels (such as link interruption, routing loop, hardware aging, software defects, etc.). This invention achieves "redundancy removal, quality improvement, and timeliness assurance" of fault data in heterogeneous network environments by filtering sensitive parameters through information entropy, refining features through box dimension and Spearman coefficient, and using SVM for efficient classification. While retaining key information, this invention reduces data dimensionality and shortens the average fault location time, providing highly reliable, low-latency, and scalable decision support for complex network operation and maintenance.

[0020] According to one embodiment of the present invention, in step S102, the network operation data includes: device health status parameters and network link performance parameters; The device health status parameters include: clock synchronization status, device power supply status, cooling system status, processor load rate, device temperature index, and engine redundancy status. Network link performance parameters include: transmission delay, channel occupancy, and packet loss rate.

[0021] Preferably, the network operation data is characterized, including: standardizing and scoring each type of network operation data, with a score range of 0-1.

[0022] In this embodiment, in step S102, the system simultaneously collects two types of raw data at fixed intervals (e.g., 30 seconds) using common protocols such as SNMP, Syslog, and NetFlow: device health status parameters and network link performance parameters. Then, a [0,1] standardized "score" is performed on all parameters: physical quantities are mapped to dimensionless scores, where 1 represents "completely normal" and 0 represents "extreme abnormality," thus eliminating dimensional differences and facilitating subsequent entropy calculation.

[0023] This invention transforms the heterogeneous device status and link performance into directly comparable information entropy inputs by unifying 0-1 estimation. After refinement by dual features of box dimension and rank correlation, the fault type is quickly located by SVM.

[0024] According to one embodiment of the present invention, step S104 includes: sorting the information entropy of each network operating parameter in descending order, and retaining the network operating parameters before the sorting by k as the target network operating parameters; wherein, k is a natural number less than or equal to 5.

[0025] The higher the information entropy, the more drastic the fluctuation of the parameter within the observation window, often corresponding to the dimension with the most significant fault symptoms. In this embodiment, in step S104, after calculating the information entropy of all network operating parameters, the upper bound of the 95% confidence interval of the entropy value distribution is first calculated, and the k value (k≤5) is dynamically determined. Then, it is cross-validated with the whitelist of business experts to prevent critical but low-entropy alarms from being mistakenly deleted. For example, if k=5 is determined, the parameters are sorted from largest to smallest entropy value, and only the top 5 parameters are retained as "target network operating parameters". Through the "Top-k" strategy, the dependence on the dimensions and distribution of the traditional fixed threshold method is avoided, and the defect of the proportional threshold may retain too many or too few dimensions is also avoided. If there is a "tied fifth" which results in more than 5 parameters, the Spearman correlation coefficient is used to remove redundancy to ensure that the final dimensions meet the set requirements.

[0026] This invention employs a three-step closed loop of "Top-k dynamic entropy selection + whitelist verification + related redundancy removal". Under the strict constraint of k≤5, it can completely retain key fault information using only 5 or fewer high-entropy parameters, thereby reducing input dimension, halving SVM training time, and improving fault classification values. This significantly reduces computational overhead and enhances model interpretability while ensuring accuracy.

[0027] According to one embodiment of the present invention, step S106 includes: acquiring 5 sets of data before and after the time of the fault occurrence as a target network operating parameter sequence, with a sampling interval of 5 minutes between every two adjacent sets of target network operating parameter sequences.

[0028] In this embodiment, in step S106, the fault occurrence time is obtained by cross-verifying the network management system logs and data records. Based on this, five sets of sampling points are taken forward and five sets backward, for a total of 10 sets of target network operating parameter sequences, with the interval between adjacent sets fixed at 5 minutes.

[0029] This invention accurately locates the fault moment through cross-validation and obtains five sets of data before and after the fault at five-minute intervals to form a time series. This not only preserves the potential degradation trend before the fault but also covers the instantaneous and short-term recovery process after the fault. It provides sufficient and aligned time series context for box dimension and Spearman correlation calculation, effectively improving the accuracy of fault type judgment and the reliability of feature analysis.

[0030] According to one embodiment of the present invention, in step S110, the classification function of the SVM fault classification model... The calculation formula is: in, Represents support vectors; Indicates the number of fault types; Represents the Lagrange multipliers; Represents the kernel function. This represents the sequence of target network operating parameters after feature extraction. , These represent the first and second support vectors in a specific category, respectively. This represents the classification threshold.

[0031] Preferably, the fault types include: optical equipment failure, network configuration error, line blockage, equipment hardware failure, CPU overload, equipment frequency reduction, clock oscillation and other fault causes.

[0032] In this embodiment, labeled samples from a known knowledge base are used as the training set. A one-to-many strategy is employed to construct an SVM fault classification model. The model input is the target network operating parameter sequence after dual feature extraction using box dimension and Spearman rank correlation. The output consists of 7 fault labels (optical equipment fault, network configuration error, line blockage, equipment hardware fault, CPU overload, equipment frequency reduction, clock oscillation, and other fault causes). The target network operating parameter sequence after feature extraction obtained through steps S102-S108 is used as the classification function. The input of the calculation formula is based on The output fault labels determine the fault type of the communication network management and control system.

[0033] This invention constructs an SVM fault classification model through a one-to-many strategy. It takes the time-series parameters extracted from the dual features of box dimension and Spearman rank correlation as input, and trains based on known labeled samples to achieve accurate classification of seven types of faults, including optical equipment faults and network configuration errors. This effectively improves the ability to judge potential fault causes and the efficiency of operation and maintenance under small sample and nonlinear network management data.

[0034] Example 2 Figure 2 This is a schematic flowchart illustrating a fault type determination method for a communication network management and control system according to another embodiment of the present invention. Figure 2 As shown, according to one embodiment of the present invention, a method for determining the fault type of a communication network management and control system includes the following steps: Step S202 involves acquiring network operation data from the communication network management and control system, and performing feature processing on the network operation data to obtain network operation parameters; specifically including: Based on common protocol standards such as SNMP, Syslog, and Netflow in integrated network management systems, network operation data is collected and characterized according to a preset period. In this embodiment, the network operation data includes device health status parameters and network link performance parameters; specifically, Equipment health status parameters (standardized assessment value range: 0-1 points, 1 point indicates no abnormalities), including: NTP clock, power status, fan status, CPU utilization, temperature, and main / standby engine status. Network link performance parameters (standardized evaluation value range: 0-1 points, 1 point indicates the best performance), including: link latency, bandwidth utilization, and packet loss rate.

[0035] Step S204 involves obtaining the information entropy of each network operating parameter, and filtering the network operating data based on the information entropy of each network operating parameter to obtain the target network operating parameters; specifically including: Step S2041: Calculate the information entropy of each network operating parameter; specifically, First, the nth network running data is set to include N random variables. Calculate the probability of each random variable occurring. The number of random variables included in each network operation data can be different. In this embodiment, the nth network operation data includes N random variables. in, This indicates that the nth network operating parameter in the uncertain system F has the i-th random variable. Quantity; This represents the total number of times the nth network operating parameter appears in the uncertain system F; Then, based on the probability of each random variable occurring... Further calculations are performed to obtain the information entropy of the nth network operating parameter. ; Step S2042: Sort the information entropy of all network operating parameters in descending order, and retain the top five network operating parameters as the target network operating parameters; Information entropy calculations are used to filter out parameters with drastic changes from numerous network operating parameters, achieving the goal of input parameter selection. The information entropy of each network operating parameter category is calculated independently; the higher the information entropy of a single network operating parameter, the more drastic the data change. After information entropy calculation, only the five network operating parameters with the highest information entropy in the sample to be evaluated are retained for subsequent calculations.

[0036] For example, if the information entropy of the network operation data items 1, 3, 4, 5, and 6 is the highest, then the original samples... Input samples are obtained after filtering. It is also equivalent to .

[0037] The raw data samples with the highest information entropy can better reflect the differences between normal and fault states by utilizing these drastically changing data. For example, when an optical device fails or an optical cable is cut by human intervention, the information entropy of optical transmission power is relatively high. When a station's power and environmental equipment fails, the information entropy of fan status and equipment temperature is relatively high. This is helpful for subsequent analysis of the cause of the fault.

[0038] Step S206: Obtain the time of the network control system failure, and extract 5 sets of data before and after the time of the failure to form the target network operating parameter sequence. Calculate the target network running data sequence The box dimension and Spearman rank correlation coefficient; specifically including: Step S2061: Retrieve logs and data records from the network management system, and determine the time of the fault based on the logs and data records. ; Step S2062: Obtain the time of fault occurrence. Five sets of data before and after are used as the target network's operating parameter sequence. (t = 1,2,…,10), 10 sets of target network operating parameter sequences In this context, the interval between any two adjacent time groups is 5 minutes. Step S2063: Calculate the target network operating parameter sequence. The box dimension and Spearman rank correlation coefficient; First, calculate the target network's operating parameter sequence. The box dimension; In the metric space, when the graph For signal parameter sequence (t = 1,2,…, A simplified algorithm can be used to place the sequence within a unit square, with the minimum interval of the x-coordinates. In this embodiment, =10; At this point, the number of closed squares completely covering the figure... Represented as: The box dimension is represented as: The target network operation parameters are obtained from the target network operation data sequence using the above method. box dimension ; Simultaneously, calculate the target network's operational data sequence. Data for each target network operation Spearman rank correlation coefficient ; The data samples are divided into samples from before the failure occurred. and samples after the failure , = 1, 2, ..., 5, and the Spearman rank correlation coefficient between the two is expressed as: in, It is the first The rank difference of an observation between two consecutive samples In this embodiment, the total number of observations is [number]. Take 5.

[0039] A smaller Spearman rank correlation coefficient indicates a weaker correlation of signal characteristic parameters before and after interference. When the data is highly correlated and of consistent rank; when At that time, the data were basically unrelated; when At this time, the data are highly correlated but inversely ranked. The correlation is used to evaluate the degree of difference in network management data characteristics caused by faults, and thereby to determine the cause of the fault.

[0040] Step S208, based on box dimension and Spearman rank correlation coefficient Obtain the target network's running parameter sequence after feature extraction. ; Step S210: Extract the target network's running parameter sequence after feature extraction. Input a pre-trained SVM fault classification model to obtain the fault type; specifically including: Pre-build an SVM fault classification model; The SVM classifier is trained using a known knowledge base, and the mapping relationship between the network operating parameters and the fault judgment results is obtained as the SVM fault classification model. Each fault type in the fault diagnosis results is labeled with a different tag, including: Optical equipment failure – Label 1; Network configuration error - Tag 2; Line blockage – Label 3; Equipment hardware failure – Tag 4; CPU overload - Tag 5; Device frequency reduction – Tag 6; Clock oscillation and other factors are the causes of the malfunction – tag 7.

[0041] SVM fault classification model classification function for: in, Represents support vectors; Indicates the number of fault types; Represents the Lagrange multipliers; Represents the kernel function. This represents the sequence of target network operating parameters after feature extraction. , These represent the first and second support vectors in a specific category, respectively. This represents the classification threshold.

[0042] The target network running parameter sequence after feature extraction Input classification function Get the fault type.

[0043] Table 1. Fault diagnosis results for a certain optical equipment failure. Taking a specific optical equipment failure as an example, the bandwidth utilization, packet loss rate, link latency, CPU utilization, and equipment temperature, calculated using information entropy, are the five parameters with the greatest fluctuations before and after the failure. These parameters are used for subsequent failure cause analysis, while the remaining parameters are discarded. Table 1 shows the data samples obtained after extraction using box dimension and Spearman rank correlation coefficient, along with the judgment results of the SVM failure classification model. The failure causes obtained using the communication network management system failure judgment method of this invention are completely consistent with the actual failure causes.

[0044] This invention provides a method for determining the fault type of a communication network management and control system. This method quantifies the degree of data change through information entropy, effectively addressing the challenges of complex and redundant communication network data, and improving the efficiency of key information extraction and subsequent data processing. Based on the joint analysis of box dimension and correlation coefficient, it performs multi-dimensional quantitative comparison of data before and after network faults, accurately locating high-value parameters associated with the faults. Combined with the SVM classification algorithm, it enhances the ability to identify fault patterns in small-sample, nonlinear network management data, enabling rapid determination of potential fault causes, thereby comprehensively improving network operation and maintenance efficiency and quality.

[0045] Example 3 Figure 3 This is a schematic diagram of the fault type judgment system of a communication network management and control system according to one embodiment of the present invention; Figure 4 This is a schematic diagram illustrating the principle of a feature extraction module according to one embodiment of the present invention. Figure 3-4 As shown, according to one embodiment of the present invention, a fault type determination system for a communication network management and control system includes: The acquisition module is used to acquire network operation data from the communication network management and control system, and to perform feature processing on the network operation data to obtain network operation parameters. The parameter filtering module is used to obtain the information entropy of each network operating parameter, and to filter the network operating parameters based on the information entropy of each network operating parameter to obtain the target network operating parameters; The feature extraction module is used to obtain the time of failure of the network management system, extract the target network operating parameters before and after the time of failure, and generate the target network operating parameter sequence; where m is a natural number greater than or equal to 1. The feature extraction module is also used to obtain the target network running parameter sequence after feature extraction based on the box dimension and Spearman rank correlation coefficient of the target network running parameter sequence; The SVM judgment module is used to input the sequence of target network running parameters after feature extraction into a pre-trained SVM fault classification model to obtain the fault type.

[0046] This embodiment provides a fault type judgment system for a communication network management and control system, including: an acquisition module, a parameter filtering module, a feature extraction module, and an SVM judgment module. The acquisition module first collects device health status parameters (clock synchronization, power supply, heat dissipation, CPU load, temperature, engine redundancy) and network link performance parameters (latency, channel occupancy, packet loss rate) at 30s to 5min intervals using common protocols such as SNMP / Syslog / NetFlow. All raw indicators are then standardized to 0-1 to eliminate dimensions and unify score ranges. The parameter filtering module performs preliminary filtering and cleaning of the device health status parameters and network link performance parameters collected in real time by the integrated network management system, focusing on drastically changing abnormal data. It retains the five dimensions with the highest entropy values ​​according to a dynamic strategy of "Top-k and k≤5". If ties occur, Spearman's rank correlation coefficient is used to remove redundancy, and cross-validation is performed with a whitelist of operation and maintenance experts to ensure that the most drastically changing data is captured while not missing low-entropy but business-critical alarms. The feature extraction module further compresses and simplifies the filtered data, highlighting the most intuitive information that reflects the occurrence of network faults; the SVM evaluation module uses an SVM classifier to intelligently judge the data after feature extraction and outputs the final fault cause judgment result.

[0047] This invention collects and standardizes device and link parameters using protocols such as SNMP, dynamically filters highly variable parameters using information entropy, removes redundancy by combining Spearman coefficients and expert whitelists, generates a time series sequence before and after a fault, extracts features using both box dimension and Spearman coefficients, and finally uses an SVM classifier to achieve accurate fault type identification under small sample and nonlinear network management data, effectively improving network operation and maintenance efficiency and fault location accuracy.

[0048] According to one embodiment of the present invention, an electronic device includes a processor, a memory, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, it implements any of the fault type judgment methods of the communication network management and control system of the present invention.

[0049] According to one embodiment of the present invention, a computer-readable storage medium stores a computer program, which, when executed by a processor, implements any of the fault type determination methods of the communication network management and control system of the present invention.

[0050] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and media described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0051] The above description is merely a preferred embodiment of this application and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in this application is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the inventive concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this application.

[0052] It should be understood that the sequence number of each step in the invention and embodiments of the present invention does not absolutely imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

Claims

1. A method for judging a fault type of a communication network management and control system, characterized in that, The method comprises the following steps: obtaining network operation data of a communication network management system, and performing feature processing on the network operation data to obtain network operation parameters; obtaining information entropy of each network operation parameter, and performing screening on the network operation parameters based on the information entropy to obtain target network operation parameters; obtaining a fault occurrence time of the network management system, and extracting m groups of data before and after the fault occurrence time to generate a target network operation parameter sequence, wherein m is a natural number greater than or equal to 1; obtaining a target network operation parameter sequence after feature extraction based on a box dimension and a Spearman rank correlation coefficient of the target network operation parameter sequence; inputting the target network operation parameter sequence after feature extraction into a pre-trained SVM fault classification model to obtain a fault type. 2.The method of claim 1, wherein, The network operation data comprises device health state parameters and network link performance parameters. The device health state parameters comprise clock synchronization state, device power supply state, heat dissipation system state, processor load rate, device temperature index, and engine redundancy state. The network link performance parameters comprise transmission delay index, channel occupancy rate, and data packet loss rate. 3.The method of claim 2, wherein, The feature processing on the network operation data comprises standardization estimation of each network operation data, and the estimation range is 0-1. 4.The method of claim 1, wherein, The screening on the network operation parameters based on the information entropy comprises sorting the information entropy of each network operation parameter in descending order, and retaining the top k network operation parameters as the target network operation parameters, wherein k is a natural number less than or equal to 5. 5.The method of claim 1, wherein, The extraction of m groups of data before and after the fault occurrence time to generate the target network operation parameter sequence comprises: obtaining five groups of data before and after the fault occurrence time as the target network operation parameter sequence, and the sampling interval of each two adjacent target network operation parameter sequences is 5 minutes. 6.The method of claim 1, wherein, a classification function of the svm fault classification model The calculation formula is: wherein, denotes a support vector; denotes a number of fault types; denotes a Lagrange multiplier; denotes a kernel function, denotes a sequence of target network operating parameters after feature extraction; , denote a first support vector and a second support vector in a specific class, respectively, denotes a classification threshold. 7.The method of claim 1, wherein, The fault type comprises optical device fault, network configuration error, line blockage, device hardware fault, CPU overload, device frequency reduction, clock oscillation, and other fault reasons. 8.A system for judging a fault type of a communication network management and control system, characterized by comprising: The method comprises the following steps: a obtaining module is configured to obtain network operation data of a communication network management system, and perform feature processing on the network operation data to obtain network operation parameters; a parameter screening module is configured to obtain information entropy of each network operation parameter, and perform screening on the network operation parameters based on the information entropy to obtain target network operation parameters; a feature extraction module is configured to obtain a fault occurrence time of the network management system, and extract m groups of data before and after the fault occurrence time to generate a target network operation parameter sequence, wherein m is a natural number greater than or equal to 1; the feature extraction module is further configured to obtain a target network operation parameter sequence after feature extraction based on a box dimension and a Spearman rank correlation coefficient of the target network operation parameter sequence; an SVM judgment module is configured to input the target network operation parameter sequence after feature extraction into a pre-trained SVM fault classification model to obtain a fault type.

9. An electronic device, comprising: The communication network management system fault type judging method according to any one of claims 1-7 is implemented by a processor, a memory and a computer program stored on the memory and running on the processor, wherein the computer program is executed by the processor.

10. A computer-readable storage medium, characterized in that, The communication network management system fault type judging method according to any one of claims 1-7 is implemented by a computer program stored on a computer readable storage medium and executed by a processor.

Images