Root cause analysis method and device for optical network fault, equipment and storage medium

Abstract

The invention relates to an optical network fault root cause analysis method and device, equipment and a storage medium. The method comprises the following steps: acquiring network control data and physical layer data of a target optical network system; respectively extracting management and control features of the network management and control data and physical layer features of the physical layer data, and performing attention processing on the management and control features and the physical layer features to obtain fusion features; according to the fusion features, performing root cause analysis by adopting a pre-trained fault root cause analysis model to obtain at least one fault root cause analysis result and respective corresponding confidence; and according to the historical reasoning result, adjusting the confidence coefficient corresponding to each of the at least one fault root cause analysis result, and according to the adjusted confidence coefficient, determining a target fault root cause analysis result of the target optical network system. By adopting the method, the root cause analysis accuracy and comprehensiveness of optical network faults can be improved.

Description

Technical Field

[0001] This application relates to the field of optical network fault diagnosis and analysis technology, and in particular to a root cause analysis method, apparatus, device and storage medium for optical network faults. Background Technology

[0002] With the continuous expansion of optical network scale and the increasing complexity of services, optical network fault diagnosis and root cause analysis face significant challenges. Traditional optical network root cause analysis methods mainly rely on conventional network data such as topology, alarms, and performance. However, these data often only reflect the surface state of the network. When faced with latent faults not triggered by alarms, such as optical module aging and gradual degradation of fiber transmission performance, traditional methods, due to limitations in data dimensions and analytical methods, cannot effectively identify and locate the root cause of the fault. This results in long fault investigation times and low repair efficiency, seriously affecting the quality of service and service continuity of optical networks. Therefore, there is an urgent need for a method that can improve the accuracy and comprehensiveness of optical network fault root cause analysis. Summary of the Invention

[0003] Therefore, it is necessary to provide a method, apparatus, computer equipment, computer-readable storage medium, and computer program product for root cause analysis of optical network faults, which can improve the accuracy and comprehensiveness of root cause analysis of optical network faults.

[0004] In a first aspect, this application provides a root cause analysis method for optical network faults, including:

[0005] Acquire network control data and physical layer data of the target optical network system;

[0006] The control features of network control data and the physical layer features of physical layer data are extracted separately. Attention processing is then applied to the control features and physical layer features to obtain fused features.

[0007] Based on the fusion characteristics, a pre-trained fault root cause analysis model is used to perform root cause analysis, and at least one fault root cause analysis result and its corresponding confidence level are obtained.

[0008] Based on historical reasoning results, the confidence levels corresponding to at least one root cause analysis result of a fault are adjusted, and the target root cause analysis result of the target optical network system is determined based on the adjusted confidence levels.

[0009] In one embodiment, the method further includes:

[0010] Based on the root cause analysis results of the target failure, determine the operation and maintenance decision;

[0011] The system acquires the real-time resource status of the target optical network system. If the real-time resource status meets the resource status requirements for operation and maintenance decisions, it generates configuration adjustment instructions based on the operation and maintenance decisions and sends the configuration adjustment instructions to the optical network devices to instruct the optical network devices to execute the configuration adjustment instructions.

[0012] In one embodiment, the method further includes:

[0013] Obtain the execution results corresponding to the configuration adjustment commands sent by the optical network device;

[0014] If the execution result indicates that the fault repair is completed, the target fault root cause analysis results, network management and control data, and physical layer data are added to the training samples. The fault root cause analysis model is then optimized based on the supplemented training samples to obtain the optimized fault root cause analysis model.

[0015] In one embodiment, the confidence levels corresponding to at least one root cause analysis result are adjusted based on historical inference results, including:

[0016] Determine multiple historical inference results corresponding to network control data and physical layer data;

[0017] For any root cause analysis result, if the number of historical inference results that match the root cause analysis result exceeds a preset number, the confidence level corresponding to the root cause analysis result will be increased by a preset value.

[0018] In one embodiment, the confidence levels corresponding to at least one root cause analysis result are adjusted based on historical inference results, including:

[0019] The rationality of at least one root cause analysis result of a fault is verified based on the topology data of the target optical network system.

[0020] If the validity verification result of the root cause analysis result indicates that the validity verification fails, the corresponding root cause analysis result shall be removed from at least one root cause analysis result.

[0021] The confidence levels of the processed root cause analysis results are adjusted based on historical inference results.

[0022] In one embodiment, the training steps of the root cause analysis model include:

[0023] Acquire training samples, which include network management data samples, physical layer data samples, and root cause labels.

[0024] The control feature samples of the network control data samples and the physical layer feature samples of the physical layer data samples are extracted separately. Attention processing is then applied to the control feature samples and the physical layer feature samples to obtain fused feature samples.

[0025] By using the fused feature samples as input to the initial model, the fault prediction results are obtained.

[0026] The model loss is determined based on the fault prediction results, fault root cause labels, and preset loss function;

[0027] The parameters of the initial model are updated based on the model loss until a preset stopping condition is met, thus obtaining the root cause analysis model of the fault.

[0028] Secondly, this application also provides a root cause analysis device for optical network faults, comprising:

[0029] The acquisition module is used to acquire network control data and physical layer data of the target optical network system;

[0030] The processing module is used to extract the control features of network control data and the physical layer features of physical layer data respectively, and to perform attention processing on the control features and physical layer features to obtain fused features;

[0031] The analysis module is used to perform root cause analysis based on the fusion features using a pre-trained fault root cause analysis model, and obtain at least one fault root cause analysis result and its corresponding confidence level.

[0032] The determination module is used to adjust the confidence level of each of the at least one root cause analysis results of the fault based on the historical inference results, and determine the target root cause analysis result of the target optical network system based on the adjusted confidence level.

[0033] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0034] Acquire network control data and physical layer data of the target optical network system;

[0035] The control features of network control data and the physical layer features of physical layer data are extracted separately. Attention processing is then applied to the control features and physical layer features to obtain fused features.

[0036] Based on the fusion characteristics, a pre-trained fault root cause analysis model is used to perform root cause analysis, and at least one fault root cause analysis result and its corresponding confidence level are obtained.

[0037] Based on historical reasoning results, the confidence levels corresponding to at least one root cause analysis result of a fault are adjusted, and the target root cause analysis result of the target optical network system is determined based on the adjusted confidence levels.

[0038] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0039] Acquire network control data and physical layer data of the target optical network system;

[0040] The control features of network control data and the physical layer features of physical layer data are extracted separately. Attention processing is then applied to the control features and physical layer features to obtain fused features.

[0041] Based on the fusion characteristics, a pre-trained fault root cause analysis model is used to perform root cause analysis, and at least one fault root cause analysis result and its corresponding confidence level are obtained.

[0042] Based on historical reasoning results, the confidence levels corresponding to at least one root cause analysis result of a fault are adjusted, and the target root cause analysis result of the target optical network system is determined based on the adjusted confidence levels.

[0043] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0044] Acquire network control data and physical layer data of the target optical network system;

[0045] The control features of network control data and the physical layer features of physical layer data are extracted separately. Attention processing is then applied to the control features and physical layer features to obtain fused features.

[0046] Based on the fusion characteristics, a pre-trained fault root cause analysis model is used to perform root cause analysis, and at least one fault root cause analysis result and its corresponding confidence level are obtained.

[0047] Based on historical reasoning results, the confidence levels corresponding to at least one root cause analysis result of a fault are adjusted, and the target root cause analysis result of the target optical network system is determined based on the adjusted confidence levels.

[0048] The aforementioned root cause analysis methods, apparatus, computer equipment, computer-readable storage media, and computer program products for optical network faults acquire network control data and physical layer data of the target optical network system. They extract control features from the network control data and physical layer features from the physical layer data, and perform attention processing on these features to obtain fused features. This process fully explores the correlation between explicit alarms and implicit physical layer data of the target optical network system, strengthens the weight of key information through an attention mechanism, and improves the accuracy and comprehensiveness of fault root cause analysis. Based on the fused features, a pre-trained fault root cause analysis model is used for root cause analysis, and the confidence level of each fault root cause analysis result is adjusted based on historical inference results. This makes the fault root cause analysis results more consistent with the actual fault situation of the optical network, improving the reliability of the fault root cause analysis results. Attached Figure Description

[0049] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0050] Figure 1 This is a diagram illustrating the application environment of a root cause analysis method for optical network faults in one embodiment.

[0051] Figure 2 This is a flowchart illustrating a root cause analysis method for optical network faults in one embodiment.

[0052] Figure 3 This is a schematic diagram of the overall process of the root cause analysis method for optical network faults in one embodiment;

[0053] Figure 4 This is a structural block diagram of a root cause analysis device for optical network faults in one embodiment.

[0054] Figure 5 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0055] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0056] It should be noted that the terms "first," "second," etc., used in this application can be used to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from the second element. The terms "comprising" and "having," and any variations thereof, used in this application, are intended to cover non-exclusive inclusion. The term "multiple" used in this application refers to two or more. The term "and / or" used in this application refers to one of the embodiments, or any combination of multiple embodiments.

[0057] The root cause analysis method for optical network faults provided in this application can be applied to, for example... Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated on server 104 or placed on the cloud or other network servers. This embodiment uses the method applied to terminal 102 as an example for illustration. It is understood that this method can also be applied to servers, and can also be applied to systems including terminals and servers, and implemented through the interaction between the terminal and the server. Terminal 102 acquires network control data and physical layer data of the target optical network system; extracts control features from the network control data and physical layer features from the physical layer data respectively; performs attention processing on the control features and physical layer features to obtain fused features; based on the fused features, uses a pre-trained fault root cause analysis model to perform root cause analysis, obtaining at least one fault root cause analysis result and its corresponding confidence level; adjusts the confidence level corresponding to each of the at least one fault root cause analysis results based on historical inference results; and determines the target fault root cause analysis result of the target optical network system based on the adjusted confidence level. Terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, drones, low-altitude aircraft, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, smart in-vehicle devices, and projection equipment. Portable wearable devices can include smartwatches, smart bracelets, and head-mounted displays. Head-mounted displays can be virtual reality (VR) devices, augmented reality (AR) devices, and smart glasses. Server 104 can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services.

[0058] In one exemplary embodiment, such as Figure 2 As shown, a root cause analysis method for optical network faults is provided, which is then applied to... Figure 1 Taking terminal 102 as an example, the explanation includes the following steps 202 to 208. Wherein:

[0059] Step 202: Obtain network control data and physical layer data of the target optical network system.

[0060] In this context, an optical network refers to a communication network system that uses optical signals as the information transmission carrier. The target optical network system refers to the optical network for which fault analysis is currently required. It can be categorized based on coverage or service type, such as a metropolitan area optical network for a city, a corporate campus optical network, or a specific backbone transmission link.

[0061] Network control data refers to logical layer data recorded by the target optical network system related to device operation, performance status, alarm information, etc. Network control data includes, but is not limited to, the target optical network system's topology data (e.g., node AB as the core link), network data, alarm data (e.g., alarm ID = ALM20250704001, alarm level = "urgent"), performance data (e.g., throughput = 10Gbps). Table 1 shows network control data in some embodiments.

[0062] Table 1 Network Management Data

[0063]

[0064]

[0065]

[0066] Physical layer data refers to the data from the physical layer of the target optical network system, which directly reflects the physical characteristics of optical signal transmission. For example, physical layer data originates from hardware such as optical modules and fiber optic sensing devices, and can be acquired using dedicated monitoring equipment and instruments, such as optical power meters, spectrum analyzers, and dispersion testers.

[0067] Physical layer data can be numerical or sequential. For example, numerical data includes dispersion values ​​(e.g., 18.5 ps / (nm·km)), fiber attenuation coefficients (e.g., 0.42 dB / km), and optical module operating temperatures (e.g., 52°C); sequential data includes raw data of optical power fluctuation curves (e.g., (-3.2, -3.3, -8.7)) and raw data of signal-to-noise ratio degradation trends (e.g., (25.0, 24.5, 20.0)). Table 2 shows physical layer data from some embodiments.

[0068] Table 2 Physical Layer Data

[0069]

[0070]

[0071] In some embodiments, since physical layer data is measured and therefore contains certain errors, it can be preprocessed to obtain preprocessed physical layer data. Subsequent feature extraction, attention processing, and root cause analysis are then performed based on network management data and the preprocessed physical layer data. Preprocessing includes at least one of the following: data cleaning, format conversion, time alignment, outlier removal, interpolation completion, and format standardization.

[0072] Data cleaning refers to removing outliers from physical layer data (such as invalid values ​​where optical power suddenly exceeds the instrument's range) and completing missing fields in traditional data (such as marking "not cleared" when the End Time of alarm data is empty). Format conversion refers to converting physical layer data according to a preset format, for example, converting the array format of optical power fluctuation curves into "time-power" key-value pairs, and uniformly retaining two decimal places for physical parameters such as dispersion values ​​and attenuation coefficients. Time alignment refers to associating network management data and physical layer data of the same device / link based on timestamps (such as aligning "node A alarm occurrence time" with "node A optical power sudden drop time"). Outlier removal refers to removing invalid values ​​from physical layer data, for example, deleting invalid values ​​(such as -50dBm) in the optical power sequence that exceed the instrument's range (such as -40dBm to +10dBm). Interpolation completion refers to filling in missing data in sequential physical layer data. For example, for the missing 15:00 data in a signal-to-noise ratio sequence, it is filled in using the mean of the preceding and following values ​​(e.g., the mean of 24.5 and 20.0 is 22.25). Format standardization refers to converting sequential physical layer data into a predefined standardized format. For example, sequential data is converted into a JSON array format, such as standardizing the optical power fluctuation curve as {"timestamp":["15:00","15:01","15:02"],"power":[-3.2,-3.3,-8.7]}.

[0073] The preprocessed physical layer data is a standardized multimodal dataset. For example, numerical data retains two decimal places, and sequence data is a structured array.

[0074] Step 204: Extract the control features of the network control data and the physical layer features of the physical layer data respectively, and perform attention processing on the control features and physical layer features to obtain the fused features.

[0075] Among them, control characteristics refer to key information reflecting the operating status of the target optical network system extracted from network control data, including but not limited to alarm characteristics, topology characteristics, performance characteristics, etc.

[0076] Physical layer characteristics refer to quantitative indicators extracted from physical layer data that reflect the physical characteristics of optical signal transmission. They are directly related to the hardware status and signal quality of the optical network, including but not limited to dispersion values, fiber attenuation coefficients, and optical power curves.

[0077] Attention processing refers to using attention mechanisms (such as self-attention and cross-attention) to assign different attention weights to different features, strengthen the influence of key features, weaken the interference of secondary features, and achieve feature fusion.

[0078] For example, a feature encoder converts control features and physical layer features into feature vectors, resulting in control feature vectors and physical layer feature vectors. An attention mechanism is then used to calculate the attention weights of the control feature vectors and physical layer feature vectors respectively. Based on their respective attention weights, the control feature vectors and physical layer feature vectors are weighted and summed, and the result is used as the fused feature.

[0079] Step 206: Based on the fusion features, perform root cause analysis using a pre-trained fault root cause analysis model to obtain at least one fault root cause analysis result and its corresponding confidence level.

[0080] The root cause analysis model refers to a machine learning model trained on a large amount of historical fault data, which can be a neural network model, a large model, etc. The fused features serve as input to the root cause analysis model, and the model outputs at least one root cause analysis result and its corresponding confidence level.

[0081] The root cause analysis results indicate the possible causes of failure in the target optical network system. The confidence level indicates the reliability of the corresponding root cause analysis results; the higher the confidence level, the higher the reliability.

[0082] Step 208: Adjust the confidence level of each of the at least one root cause analysis results of the fault according to the historical reasoning results, and determine the target root cause analysis result of the target optical network system based on the adjusted confidence level.

[0083] The historical inference results refer to the confidence levels of the model's outputs for root cause analysis of optical network system failures at historical moments, as well as the actual conclusions reached after subsequent verification (such as manual investigation). Historical inference results record the model's performance in similar scenarios and can be used to optimize current analysis results.

[0084] The confidence levels of each root cause analysis result are adjusted based on historical inference results. For example, the confidence levels of root cause analysis results existing in historical inference results are increased by a preset value. Based on the adjusted confidence levels, the target root cause analysis result is determined. For example, the root cause analysis result with the highest adjusted confidence level is taken as the target root cause analysis result. Since historical experience is incorporated, this helps to improve the accuracy and reliability of the target root cause analysis result.

[0085] In some embodiments, a correlation mapping is constructed from the network management data, physical layer data, target fault root cause analysis results, and corresponding confidence levels of the target optical network system. The correlation analysis mapping is added to the mapping table, and the mapping table can be updated after each inference. Each correlation mapping in the mapping table can be used to update the fault root cause analysis model, thereby continuously optimizing the fault root cause analysis model to improve the performance of the fault root cause analysis model and the accuracy of the analysis results.

[0086] In the aforementioned root cause analysis method for optical network faults, network control data and physical layer data of the target optical network system are acquired. Control features of the network control data and physical layer features of the physical layer data are extracted separately. Attention processing is applied to these two features to obtain fused features, fully exploring the correlation between explicit alarms and implicit physical layer data of the target optical network system. The attention mechanism strengthens the weight of key information, improving the accuracy and comprehensiveness of the root cause analysis. Based on the fused features, a pre-trained root cause analysis model is used for root cause analysis. The confidence level of each root cause analysis result is adjusted based on historical inference results, making the root cause analysis results more consistent with the actual fault situation of the optical network and improving the reliability of the root cause analysis results.

[0087] In an exemplary embodiment, the root cause analysis method for optical network faults further includes: determining an operation and maintenance decision based on the root cause analysis results of the target fault; obtaining the real-time resource status in the target optical network system; and generating a configuration adjustment instruction based on the operation and maintenance decision when the real-time resource status meets the resource status required for the operation and maintenance decision, and sending the configuration adjustment instruction to the optical network device to instruct the optical network device to execute the configuration adjustment instruction.

[0088] Among them, operation and maintenance decision refers to the specific operation and maintenance plan formulated based on the root cause analysis results of the target fault, which is used to resolve the faults of the target optical network system.

[0089] In some embodiments, the terminal stores a mapping relationship between fault root cause analysis results and operation and maintenance decisions. Through this mapping relationship, the operation and maintenance decision corresponding to the target fault root cause analysis result can be determined. For example, the operation and maintenance decision corresponding to "attenuation coefficient exceeding the standard" is "adjust the optical amplifier gain", and the operation and maintenance decision corresponding to "abnormal dispersion value" is "enable the dispersion compensation module".

[0090] Real-time resource status refers to the real-time status of the resources contained in the target optical network system. Executing operational and maintenance (O&M) decisions often requires certain resource conditions to be met. The real-time resource status is compared with the resource status required for the O&M decision. If the real-time resource status meets the requirements, the O&M decision is executed to ensure its effective implementation. If the real-time resource status does not meet the requirements, a manual O&M instruction is generated to instruct O&M personnel to handle the issue manually. For example, if the root cause analysis of the target fault indicates a fiber optic cable break, and the required resource status for the O&M decision is the existence of a backup link, and the real-time resource status indicates the existence of a backup link, then the O&M decision is executed, triggering automatic backup link switching. If the real-time resource status indicates no backup link, manual O&M is triggered.

[0091] Configuration adjustment commands refer to operation instructions generated based on operational decisions that can be recognized by optical network devices. The terminal sends the configuration adjustment commands to the optical network devices, which can then execute the commands to resolve faults in the target optical network system.

[0092] In this embodiment, by determining the operation and maintenance decision corresponding to the root cause analysis result of the target fault, it is beneficial to carry out targeted operation and maintenance operations based on the root cause analysis result of the target fault; ensuring that the real-time resource status meets the resource status required by the operation and maintenance decision, and executing the configuration adjustment instructions corresponding to the operation and maintenance decision, can avoid operation and maintenance failure due to insufficient resources and improve the probability of operation and maintenance success.

[0093] In an exemplary embodiment, the root cause analysis method for optical network faults further includes: obtaining the execution result corresponding to the configuration adjustment instruction sent by the optical network device; when the execution result indicates that the fault repair is completed, supplementing the training sample with the target fault root cause analysis result, network management and control data, and physical layer data; and optimizing the fault root cause analysis model based on the supplemented training sample to obtain the optimized fault root cause analysis model.

[0094] The execution result refers to the result generated after the optical network device executes the configuration adjustment command, indicating whether the configuration adjustment command was executed successfully. Successful execution of the configuration adjustment command indicates that the fault repair is complete and the root cause analysis of the target fault is accurate.

[0095] Training samples refer to the data used to train the root cause analysis model. By supplementing the training samples with the target root cause analysis results, network management data, and physical layer data, the training samples are continuously enriched, and the root cause analysis model is continuously optimized. This helps to reduce the probability of misjudgment by the root cause analysis model and improve the accuracy of the analysis.

[0096] In this embodiment, the execution results sent by the optical network device determine whether the fault has been repaired. The target fault root cause analysis results, network management data, and physical layer data corresponding to the fault repair completion are then added to the training samples to continuously optimize the model. The model can continuously learn new fault modes and improve the model's recognition accuracy.

[0097] In an exemplary embodiment, adjusting the confidence level of each of at least one root cause analysis result based on historical inference results includes: determining multiple historical inference results corresponding to network control data and physical layer data; and for any root cause analysis result, increasing the confidence level corresponding to the root cause analysis result by a preset value if the number of historical inference results matching the root cause analysis result exceeds a preset number.

[0098] Specifically, the terminal identifies the root cause of the fault from historical data that matches at least one of the network control data and physical layer data, serving as the historical inference result. The historical inference result reflects the correlation between network control data, physical layer data, and the root cause of the fault.

[0099] If historical inference results may align with root cause analysis results, the confidence level corresponding to those root cause analysis results can be increased by a preset value. For example, if there are 50 historical inference results matching either network control data or physical layer data, and 30 of them indicate optical amplifier aging, then the confidence level corresponding to optical amplifier aging in the root cause analysis results can be increased by a preset value. The preset number and preset value can be flexibly adjusted according to actual circumstances.

[0100] In this embodiment, the confidence level of the root cause analysis results is adjusted by the number of historical inference results that match the root cause analysis results. This helps to select target root cause analysis results that are more consistent with the actual fault situation, thereby improving the accuracy of root cause analysis.

[0101] In an exemplary embodiment, adjusting the confidence level of each of the at least one root cause analysis results based on historical inference results includes: performing a rationality verification on at least one root cause analysis result based on the topology data of the target optical network system; removing the corresponding root cause analysis result from the at least one root cause analysis result if the rationality verification result indicates that the rationality verification fails; and adjusting the confidence level of each of the processed root cause analysis results based on historical inference results.

[0102] Topology data refers to structured data describing the connections between devices, links, and nodes in the target optical network. Reasonableness analysis determines whether the root cause analysis results conform to the topological patterns of the target optical network system, eliminating unreasonable root cause analysis results that contradict the topology data. For example, if the root cause analysis results indicate that a faulty optical module at node D causes signal degradation in the link A→B, but the topology data shows no connection between node D and the A→B link, then this root cause analysis result fails the reasonableness verification. The terminal removes the root cause analysis result that fails reasonableness verification from at least one root cause analysis result to avoid interfering with subsequent operation and maintenance decisions.

[0103] The processed root cause analysis results are the remaining root cause analysis results after removing those that failed the rationality verification. The terminal continues to perform operations such as confidence adjustment based on the processed root cause analysis results to determine the target root cause analysis results.

[0104] In this embodiment, the rationality of the root cause analysis results is verified by using topological data. Root cause analysis results that fail the rationality verification are eliminated, which helps to focus on root cause analysis results that conform to topological relationships and improves the efficiency and reliability of fault location.

[0105] In an exemplary embodiment, the training steps of the fault root cause analysis model include: acquiring training samples, which include network control data samples, physical layer data samples, and fault root cause labels; extracting control feature samples from the network control data samples and physical layer feature samples from the physical layer data samples, respectively, and performing attention processing on the control feature samples and physical layer feature samples to obtain fused feature samples; using the fused feature samples as input to the initial model to obtain fault prediction results; determining the model loss based on the fault prediction results, fault root cause labels, and a preset loss function; and updating the parameters of the initial model based on the model loss until a preset stopping condition is met, thereby obtaining the fault root cause analysis model.

[0106] The terminal selects network management data samples, physical layer data samples, and corresponding root cause labels from historical data. After feature extraction and attention processing on the network management data samples and physical layer data samples, the fused feature samples are input into the initial model to obtain fault prediction results. Based on the fault prediction results, fault root cause labels, and a preset loss function, the model loss is determined. The parameters of the initial model are then updated based on the model loss to obtain the fault root cause analysis model. The preset loss function can be a cross-entropy loss function, etc.

[0107] In this embodiment, by fusing control feature samples and physical layer feature samples, and strengthening key information through attention processing, the model can simultaneously capture explicit logical faults and implicit physical layer faults, avoiding the one-sidedness of a single feature dimension and improving the accuracy of root cause analysis of the model; through multiple rounds of iterative training, the model parameters are continuously optimized to improve model performance.

[0108] To illustrate the root cause analysis method and its effectiveness for optical network faults in this solution, a detailed implementation example is provided below:

[0109] like Figure 3 The diagram illustrates the overall flow of root cause analysis methods for optical network faults in some embodiments. The terminal acquires network management data from the target optical network system and obtains physical layer data through multimodal data acquisition, forming a multimodal dataset. The multimodal dataset is preprocessed to obtain preprocessed data. The network management data and the preprocessed data are then concatenated into a structured dataset. Root cause analysis is performed based on the structured dataset and historical reasoning data to generate operational decisions. Specifically, the control features of network control data and the physical layer features of preprocessed physical layer data are extracted separately. Attention processing is applied to the control features and physical layer features to obtain fused features. Based on the fused features, a pre-trained root cause analysis model is used to perform root cause analysis, obtaining at least one root cause analysis result and its corresponding confidence level. The confidence levels corresponding to each of the at least one root cause analysis result are adjusted based on historical inference results. Based on the adjusted confidence levels, the target root cause analysis result of the target optical network system is determined. Based on the target root cause analysis result, an operation and maintenance decision is determined. The real-time resource status of the target optical network system is obtained. If the real-time resource status meets the resource status required for the operation and maintenance decision, a configuration adjustment instruction is generated based on the operation and maintenance decision and sent to the optical network device to instruct the optical network device to execute the configuration adjustment instruction. The execution result corresponding to the configuration adjustment instruction sent by the optical network device is obtained.

[0110] In some embodiments, alarm correlation analysis enhancement refers to constructing the correlation between the target fault root cause analysis result, network management data, and physical layer data when the execution result indicates that the fault repair is completed, and supplementing the training sample with the alarm correlation analysis result. The training sample is structured data. The fault root cause analysis model is optimized based on the supplemented training sample to obtain the optimized fault root cause analysis model.

[0111] In one embodiment, adjusting the confidence level of each of at least one root cause analysis result based on historical inference results includes: determining multiple historical inference results corresponding to network control data and physical layer data; and increasing the confidence level of any root cause analysis result by a preset value if the number of historical inference results matching the root cause analysis result exceeds a preset number.

[0112] In one embodiment, adjusting the confidence level of each of the at least one root cause analysis results based on historical inference results includes: performing a rationality verification on at least one root cause analysis result based on the topology data of the target optical network system; removing the corresponding root cause analysis result from the at least one root cause analysis result if the rationality verification result indicates that the rationality verification fails; and adjusting the confidence level of each of the processed root cause analysis results based on historical inference results.

[0113] In one embodiment, the training steps of the fault root cause analysis model include: acquiring training samples, which include network control data samples, physical layer data samples, and fault root cause labels; extracting control feature samples from the network control data samples and physical layer feature samples from the physical layer data samples, respectively, and performing attention processing on the control feature samples and physical layer feature samples to obtain fused feature samples; using the fused feature samples as input to the initial model to obtain fault prediction results; determining the model loss based on the fault prediction results, fault root cause labels, and a preset loss function; and updating the parameters of the initial model based on the model loss until a preset stopping condition is met, thereby obtaining the fault root cause analysis model.

[0114] This explanation uses an access-type OTN (Optical Transport Network) management system as an example. The root cause analysis method for optical network faults includes the following steps:

[0115] 1. Data Acquisition Phase: The network management system synchronously collects the following data using specialized equipment such as optical power monitors and dispersion testers:

[0116] Physical layer data: Optical power fluctuation curve (link from node A to node B, dropping sharply from -3.2dBm to -8.7dBm between 15:00 and 15:30), dispersion value (18.5ps / (nm·km) at node C, exceeding the threshold of 15±2ps / (nm·km)), fiber attenuation coefficient (0.42dB / km for link segment D, standard value ≤0.35dB / km);

[0117] Network management data: alarm data (node ​​B sends out an optical signal loss alarm, timestamp 15:20), topology information (nodes A, B, C are linked in a chain, and link D is a branch of node B), and performance data (node ​​B's throughput drops from 10Gbps to 2Gbps).

[0118] The above data is simultaneously transmitted to the alarm correlation analysis module and the network data collection module.

[0119] 2. Alarm Correlation and Data Integration: The alarm correlation analysis module outputs: Root Alarm Confidence is 92% (optical signal loss at node B), Derived Alarms include performance degradation alarms at node C, and the number of Affected Connections is 3; the network data collection module integrates physical layer data with correlation analysis results to generate a fused dataset (example fields: optical power drop of 5.5dBm; dispersion exceeding the standard of 3.5ps / (nm·km); correlation alarm confidence of 92%), which is then input into the root cause analysis module (such as a large model).

[0120] 3. Root Cause Analysis: After receiving the fused dataset, the large model outputs the following based on the trained question-and-answer pairs (including physical layer data features): Root Cause: The optical fiber link from node B to node C experiences a sudden drop in optical power due to excessive attenuation (0.42dB / km), triggering a signal loss alarm; Root Cause Confidence: 96%.

[0121] 4. Decision-making and execution: The decision-making module determines that the network resources meet the automatic adjustment conditions (the backup link E has sufficient bandwidth); the execution unit outputs a configuration adjustment instruction: switch the service from node B to node C to the backup link E, and adjust the optical amplifier gain to +2dB; after the network management system executes the instruction, the throughput of node B is restored to 9.8Gbps, and the alarm is cleared.

[0122] 5. Model Optimization: The historical data collection module combines the current physical layer data (optical power curve, attenuation coefficient), root cause analysis results, and execution instructions into training samples, updates the large model dataset, and enhances the ability to identify the correlation features of "excessive attenuation - optical power".

[0123] The newly added physical layer data in the above embodiments is seamlessly integrated through existing module interfaces and can be directly reproduced and verified in the access-type OTN management and control system.

[0124] The aforementioned root cause analysis method for optical network faults acquires network control data and physical layer data of the target optical network system. It extracts control features from the network control data and physical layer features from the physical layer data, and applies attention processing to these features to obtain fused features. This fully explores the correlation between explicit alarms and implicit physical layer data of the target optical network system, strengthening the weight of key information through an attention mechanism to improve the accuracy and comprehensiveness of the root cause analysis. Based on the fused features, a pre-trained root cause analysis model is used for root cause analysis. The confidence level of at least one root cause analysis result is adjusted based on historical inference results, making the root cause analysis results more consistent with the actual fault situation of the optical network and improving the reliability of the root cause analysis results.

[0125] Furthermore, constructing a multimodal dataset of network management and control data and physical layer data overcomes the superficial limitations of traditional solutions in describing network status. By capturing subtle power oscillations in the 30 minutes prior to an alarm through optical power fluctuation curves, and combining this with the slow upward trend of fiber attenuation coefficients, it can identify latent fault causes such as "chronic fiber aging" that traditional data cannot reflect, filling the gap in the analysis of the "transformation process from physical layer degradation to network faults" in traditional methods. A large model is used to jointly encode the multimodal data, and an attention mechanism is employed to strengthen the cross-dimensional correlation between physical layer features (such as the magnitude of sudden drops in optical power) and network alarms (such as signal loss) and performance indicators (such as throughput reduction). This mechanism enables the large model to identify the complete causal chain of "abnormal physical parameters → network status degradation → fault alarm triggering," especially providing accurate analysis capabilities for latent physical faults not triggered by alarms (such as chronic aging of optical modules), solving the problem of delayed response to physical layer-driven faults in traditional solutions. This allows for rapid and accurate location of fault root causes, reducing fault investigation time and improving fault repair efficiency. For example, when optical networks experience performance degradation, root cause analysis can quickly determine whether the issue stems from excessive fiber attenuation or deterioration of optical module performance. Maintenance personnel can then perform targeted maintenance and replacements, significantly improving the efficiency and quality of maintenance work. Accurate root cause analysis can avoid unnecessary equipment replacement and maintenance, reducing maintenance costs. For instance, for problems that appear to be caused by equipment malfunctions, if analysis determines that the issue is due to changes in the physical characteristics of the fiber optic link, only the fiber optic link needs repair, eliminating the need to replace expensive equipment and saving substantial maintenance costs. Timely detection and resolution of faults in optical networks effectively ensures the continuity and stability of network services, improving user experience. For services with high network quality requirements, such as high-definition video transmission and financial transactions, it ensures timely handling of potential optical network faults, preventing service interruptions and data loss, demonstrating strong practicality and real-world application value.

[0126] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps. It is understood that the steps in different embodiments can be freely combined as needed, and all non-contradictory solutions formed by such combinations are within the scope of protection of this application.

[0127] Based on the same inventive concept, this application also provides a root cause analysis apparatus for optical network faults to implement the root cause analysis method for optical network faults described above. The solution provided by this apparatus is similar to the implementation scheme described in the above method. Therefore, the specific limitations of one or more root cause analysis apparatus embodiments for optical network faults provided below can be found in the limitations of the root cause analysis method for optical network faults described above, and will not be repeated here.

[0128] In one exemplary embodiment, such as Figure 4 As shown, a root cause analysis device 400 for optical network faults is provided, comprising: an acquisition module 420, a processing module 440, an analysis module 460, and a determination module 480, wherein:

[0129] The acquisition module 420 is used to acquire network control data and physical layer data of the target optical network system;

[0130] Processing module 440 is used to extract the control features of network control data and the physical layer features of physical layer data respectively, and to perform attention processing on the control features and physical layer features to obtain fused features;

[0131] Analysis module 460 is used to perform root cause analysis based on the fusion features using a pre-trained fault root cause analysis model, and obtain at least one fault root cause analysis result and its corresponding confidence level.

[0132] The determination module 480 is used to adjust the confidence level corresponding to each of the at least one root cause analysis result of the fault based on the historical inference results, and to determine the target root cause analysis result of the target optical network system based on the adjusted confidence level.

[0133] The aforementioned root cause analysis device for optical network faults acquires network control data and physical layer data of the target optical network system. It extracts control features from the network control data and physical layer features from the physical layer data, and performs attention processing on these features to obtain fused features. This process fully explores the correlation between explicit alarms and implicit physical layer data of the target optical network system, strengthens the weight of key information through an attention mechanism, and improves the accuracy and comprehensiveness of the fault root cause analysis. Based on the fused features, a pre-trained fault root cause analysis model is used for root cause analysis. The confidence level of each fault root cause analysis result is adjusted based on historical inference results, making the fault root cause analysis results more consistent with the actual fault situation of the optical network and improving the reliability of the fault root cause analysis results.

[0134] In one embodiment, the root cause analysis device 400 for optical network faults further includes an operation and maintenance module, which is used to: determine operation and maintenance decisions based on the root cause analysis results of the target fault; obtain the real-time resource status in the target optical network system; and, if the real-time resource status meets the resource status required for the operation and maintenance decision, generate a configuration adjustment instruction based on the operation and maintenance decision, and send the configuration adjustment instruction to the optical network device to instruct the optical network device to execute the configuration adjustment instruction.

[0135] In one embodiment, the operation and maintenance module is further configured to: obtain the execution result corresponding to the configuration adjustment instruction sent by the optical network device; when the execution result indicates that the fault repair is completed, supplement the training sample with the target fault root cause analysis result, network management and control data, and physical layer data, and optimize the fault root cause analysis model based on the supplemented training sample to obtain the optimized fault root cause analysis model.

[0136] In one embodiment, the confidence level corresponding to each of at least one root cause analysis result is adjusted based on the historical inference results. The determining module 480 is further configured to: determine multiple historical inference results corresponding to network control data and physical layer data; and for any root cause analysis result, if the number of historical inference results matching the root cause analysis result exceeds a preset number, increase the confidence level corresponding to the root cause analysis result by a preset value.

[0137] In one embodiment, the confidence level of each of the at least one root cause analysis results is adjusted based on historical inference results. The determining module 480 is further configured to: perform a rationality verification on at least one root cause analysis result based on the topology data of the target optical network system; if the rationality verification result of a root cause analysis result indicates that the rationality verification fails, remove the corresponding root cause analysis result from the at least one root cause analysis result; and adjust the confidence level of each of the processed root cause analysis results based on historical inference results.

[0138] In one embodiment, in training the root cause analysis model, the analysis module 460 is further configured to: acquire training samples, including network control data samples, physical layer data samples, and root cause labels; extract control feature samples from the network control data samples and physical layer feature samples from the physical layer data samples, respectively, perform attention processing on the control feature samples and physical layer feature samples to obtain fused feature samples; use the fused feature samples as input to the initial model to obtain fault prediction results; determine the model loss based on the fault prediction results, fault root cause labels, and a preset loss function; update the parameters of the initial model based on the model loss until a preset stopping condition is met, thereby obtaining the root cause analysis model.

[0139] The modules in the aforementioned root cause analysis device for optical network failures can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of a computer device in hardware form or independent of it, or stored in the memory of a computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0140] In one exemplary embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 5 As shown, the computer device includes a processor, memory, input / output interfaces, a communication interface, a display unit, and an input device. The processor, memory, and input / output interfaces are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The input / output interfaces are used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, Near Field Communication (NFC), or other technologies. When executed by the processor, the computer program implements a root cause analysis method for optical network failures. The display unit is used to form a visually visible image and can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0141] Those skilled in the art will understand that Figure 5 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0142] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0143] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above method embodiments.

[0144] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0145] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0146] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0147] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0148] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A root cause analysis method for optical network faults, characterized in that, The method includes: Acquire network control data and physical layer data of the target optical network system; The control features of the network control data and the physical layer features of the physical layer data are extracted separately, and attention processing is performed on the control features and physical layer features to obtain fused features; Based on the fusion features, a pre-trained fault root cause analysis model is used to perform root cause analysis, and at least one fault root cause analysis result and its corresponding confidence level are obtained. Based on historical reasoning results, the confidence levels corresponding to at least one root cause analysis result of a fault are adjusted, and the target root cause analysis result of the target optical network system is determined based on the adjusted confidence levels.

2. The method according to claim 1, characterized in that, The method further includes: Based on the root cause analysis results of the target failure, determine the operation and maintenance decision; The real-time resource status of the target optical network system is obtained. If the real-time resource status meets the resource status required for the operation and maintenance decision, a configuration adjustment instruction is generated according to the operation and maintenance decision, and the configuration adjustment instruction is sent to the optical network device to instruct the optical network device to execute the configuration adjustment instruction.

3. The method according to claim 2, characterized in that, The method further includes: Obtain the execution result corresponding to the configuration adjustment command sent by the optical network device; If the execution result indicates that the fault repair is completed, the target fault root cause analysis result, the network management and control data, and the physical layer data are added to the training sample. The fault root cause analysis model is then optimized based on the supplemented training sample to obtain the optimized fault root cause analysis model.

4. The method according to claim 1, characterized in that, The adjustment of the confidence levels corresponding to at least one root cause analysis result based on historical reasoning results includes: Determine multiple historical inference results corresponding to the network control data and the physical layer data; For any root cause analysis result, if the number of historical inference results matching the root cause analysis result exceeds a preset number, the confidence level corresponding to the root cause analysis result is increased by a preset value.

5. The method according to claim 1, characterized in that, The adjustment of the confidence levels corresponding to at least one root cause analysis result based on historical reasoning results includes: The rationality of at least one root cause analysis result of a fault is verified based on the topology data of the target optical network system. If the validity verification result of the root cause analysis result indicates that the validity verification fails, the corresponding root cause analysis result shall be removed from at least one root cause analysis result. The confidence levels of the processed root cause analysis results are adjusted based on historical inference results.

6. The method according to claim 1, characterized in that, The training steps of the root cause analysis model include: Obtain training samples, which include network management data samples, physical layer data samples, and root cause labels. The control feature samples of the network control data samples and the physical layer feature samples of the physical layer data samples are extracted respectively. Attention processing is performed on the control feature samples and the physical layer feature samples to obtain fused feature samples. The fused feature samples are used as input to the initial model to obtain the fault prediction result; The model loss is determined based on the fault prediction results, the fault root cause labels, and the preset loss function; The parameters of the initial model are updated based on the model loss until a preset stopping condition is met, thus obtaining the root cause analysis model.

7. A root cause analysis device for optical network faults, characterized in that, The device includes: The acquisition module is used to acquire network control data and physical layer data of the target optical network system; The processing module is used to extract the control features of the network control data and the physical layer features of the physical layer data respectively, and to perform attention processing on the control features and physical layer features to obtain fused features; The analysis module is used to perform root cause analysis using a pre-trained fault root cause analysis model based on the fusion features, and obtain at least one fault root cause analysis result and its corresponding confidence level. The determination module is used to adjust the confidence level corresponding to each of the at least one root cause analysis result of the fault based on the historical reasoning results, and to determine the target root cause analysis result of the target optical network system based on the adjusted confidence level.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images