Fault processing method and apparatus, network device, and storage medium

Abstract

The fault processing method, device, network equipment and storage medium provided by the embodiment of the application analyze the historical data of each optical module according to the model of the optical module, determine the historical data that mutates when the optical module of the corresponding model is in a fault condition, extract the corresponding original fault data from the historical data that mutates, and train a fault model according to the original fault data and the model of the optical module. Thus, by dividing according to the model of the optical module, the flexibility of collecting the historical data of the fault optical module is improved, and the application range of fault detection is further improved. In actual use, automatic updating can be achieved, the detection accuracy is continuously improved with the accumulation of fault data, and the operation and maintenance cost of the optical module is effectively reduced.

Description

TECHNICAL FIELD

[0001] Embodiments of the present application relate to, but are not limited to, the field of optical network communication, in particular, relate to, but are not limited to, a fault processing method, device, network equipment and storage medium. BACKGROUND

[0002] With the continuous development of communication technology, in order to meet the needs of the majority of users for broadband services such as integrated communication services and multimedia communication services, optical communication networks are increasingly widely used, which has well solved the bandwidth bottleneck of the transmission system in the communication network.

[0003] As a key component of optical-electric signal conversion, the optical module is an important device for connecting optical networks. With the continuous deployment of optical networks, the use of optical modules is increasing. Once the optical module fails, it will directly affect the service quality and even cause service interruption. Although the failure rate of optical module devices is very low, the inventory of optical modules in the existing network is often in the tens of thousands or even hundreds of thousands, so the number of failures is still high. Moreover, the geographical distribution of failures is random, and if a failure occurs, the maintenance personnel often cannot repair it in the first time. In short, it brings very high maintenance cost to the operator. Therefore, the fault detection of the optical module becomes a necessary means. However, in the related art, there are still some problems in the fault detection of the optical module. First, different manufacturers' optical modules often have design differences, resulting in different failure phenomena, and a unified detection method cannot cover all models. In addition, the current main detection method only detects by setting certain thresholds for a single indicator such as temperature and current, and the effect of detection is greatly different at different times due to the influence of external environment by using the same set of thresholds. Therefore, in the related art, there is a lack of a universal optical module fault detection scheme, which makes the maintenance cost of the optical module always high. SUMMARY

[0004] The fault processing method, device, network equipment and storage medium provided by the embodiments of the present application mainly solve the problem of low universality of the optical module fault detection scheme and high optical module maintenance cost in the related art.

[0005] To solve the above technical problems, the present application provides a fault processing method, comprising:

[0006] According to the type of the optical module, the historical data of each optical module is analyzed to determine the historical data that suddenly changes when the optical module of the corresponding type fails;

[0007] From the historical data that suddenly changes, the corresponding original fault data is extracted;

[0008] According to the original fault data, the fault model is trained according to the type of the optical module.

[0009] The embodiment of the present application also provides a fault processing device, comprising:

[0010] a data analysis module, used for analyzing historical data of each optical module according to a model of the optical module, and determining historical data that is mutated in case of a fault of the optical module of the corresponding model;

[0011] a fault extraction module, used for extracting corresponding original fault data from the historical data that is mutated;

[0012] a model training module, used for training a fault model according to the original fault data according to the model of the optical module.

[0013] The embodiment of the present application also provides a network device, comprising a processor, a memory and a communication bus;

[0014] The communication bus is used for realizing connection communication between the processor and the memory;

[0015] The processor is used for executing one or more computer programs stored in the memory, so as to realize steps of the fault processing method.

[0016] The embodiment of the present application also provides a computer storage medium, wherein the computer readable storage medium stores one or more programs, and the one or more programs can be executed by one or more processors, so as to realize steps of the fault processing method.

[0017] According to the fault processing method, device, network device and storage medium provided by the embodiment of the present application, the historical data of each optical module is analyzed according to the model of the optical module, the historical data that is mutated in case of a fault of the optical module of the corresponding model is determined, the corresponding original fault data is extracted from the historical data that is mutated, and the fault model is trained according to the original fault data according to the model of the optical module. Therefore, by dividing according to the model of the optical module, the flexibility of collecting the historical data of the fault optical module is improved, and the application range of the fault detection is further improved, and the operation and maintenance cost of the optical module is effectively reduced.

[0018] Other features and corresponding advantages of the present application are described in the latter part of the specification, and it should be understood that at least part of the advantages is apparent from the description of the present application in the specification. BRIEF DESCRIPTION OF DRAWINGS

[0019] Figure 1 A flow chart of the fault processing method of the embodiment one of the present application;

[0020] Figure 2Flow chart of the fault processing method of the embodiment two of the present application;

[0021] Figure 3 Deployment flow chart of the fault processing method of the embodiment three of the present application;

[0022] Figure 4 Detection flow chart of the optical module fault of the fault processing method of the embodiment four of the present application;

[0023] Figure 5 Monitoring update flow chart of the fault processing method of the embodiment five of the present application;

[0024] Figure 6 Composition schematic diagram of the fault processing device of the embodiment six of the present application;

[0025] Figure 7 Composition schematic diagram of the fault processing system of the embodiment six of the present application;

[0026] Figure 8 Composition schematic diagram of the network device of the embodiment seven of the present application. DETAILED DESCRIPTION

[0027] In order to make the objects, technical solutions and advantages of the present application clearer, further detailed description will be made to the embodiments of the present application in combination with the accompanying drawings. It should be understood that the specific embodiments described herein are only used to explain the present application, and are not used to limit the present application.

[0028] Embodiment one:

[0029] The embodiment provides a fault processing method, please refer to Figure 2 , the method comprises:

[0030] S101, according to the type of optical module, the historical data of each optical module is analyzed, and the historical data of the corresponding type of optical module is determined to occur mutation under the condition of failure;

[0031] S102, the corresponding original fault data is extracted from the historical data of mutation;

[0032] S103, according to the original fault data, the fault model training is carried out according to the type of optical module.

[0033] For the applicability of the failure processing method of the optical module in the embodiment, the optical module produced by various manufacturers can be well applied. In the embodiment, the historical data of the optical module is analyzed according to the model of the optical module, that is, the analysis results of the historical data are different for optical modules of different models. The causes and effects of the failure of the optical module may be similar, but the data changes when the failure occurs may be different for optical modules of different models, which brings the difference of the historical data whose mutation occurs in the analysis results of the historical data of the optical module. In the embodiment, considering this part of the difference, the analysis is performed according to the model of the optical module, and the historical data whose mutation occurs when the failure occurs for optical modules of different models is obtained.

[0034] Among them, the historical data whose mutation occurs indicates that the index of the historical data changes obviously in a short time when the optical module fails, which can be obviously increased or obviously decreased. In the optical module failure detection problem, it can be defined as "in the historical data of the optical module, at a certain time point, due to a special event-failure, some data before and after the time point changes obviously", and the some data referred to here is the historical data whose mutation occurs. The obvious change mentioned here can be determined according to the actual data itself, for example, for normal indicators, the increase or decrease amplitude in a short time exceeds 50%. The embodiment only lists one possible mutation amplitude, and the mutation amplitude of different data cannot be generalized. The mutation amplitude can be determined according to the actual system operation and the type of historical data, and the embodiment does not limit it.

[0035] In addition, the mutation can also indicate that the trend of the data in the time period from the time point of the failure to the time when the optical module cannot work and is replaced is obviously different from the trend before the time point of the failure, for example, before a certain time point, the trend is always stable, after a certain time point, the trend continuously rises, or falls, or fluctuates, then the time point can be determined as a mutation point, and the event that causes the mutation point to appear has a large probability of being a failure.

[0036] In the analysis of the mutation of each historical data, the most mutated indicators are counted, and the historical data that does not mutate or does not mutate obviously can be directly ignored. Of course, in order to improve the comprehensiveness of detection and avoid the adverse consequences caused by missed judgment and misjudgment, these data can be retained, and the failure model can be updated in the subsequent update stage.

[0037] The historical data of the optical module in the embodiment can be performance data, inspection data, etc. Before analyzing the historical data of each optical module according to the model of the optical module, the optical modules can be classified according to the model of the optical module. The classification standard of the optical modules of various models can be classification according to the manufacturer, classification according to the working principle of the optical module, classification according to the specifications and applicable range of the optical module, etc.

[0038] Not all of the historical data in the mutated historical data is related to the fault. For example, for the optical modules of the same model that have failed, there are some mutated historical data that are occasional, and only individual faulty optical modules can produce such data, which should not be regarded as fault data. In other words, the fault data should be the historical data that has a high mutation rate when the optical module of the corresponding model fails. Such historical data can be used as original fault data. Specifically, extracting the corresponding original fault data from the mutated historical data can include:

[0039] According to the mutated historical data, analyzing the mutation time point in the time sequence;

[0040] According to the mutation time point, determining the historical data in which the mutation rate exceeds a preset proportion;

[0041] The historical data with a mutation proportion exceeding a preset proportion is taken as the original fault data. This scheme for determining the original fault data can greatly exclude occasional mutation historical data and retain historical data that is more likely to be related to faults. The mutation time points are analyzed. The mutation time points can be considered as time points at which the optical module fails. The mutation historical data is more concentrated at these time points. However, even the mutation historical data at the mutation time points is not all caused by faults, and therefore can be further screened. The historical data with a mutation proportion exceeding a preset proportion is taken as the original fault data. The mutation proportion refers to that the proportion of the historical data that mutates at the mutation time points exceeds a threshold. For example, for a mutation time point, the proportion of the historical data that mutates on a plurality of optical modules of the same type and that fails exceeds a certain proportion (for example, 80%). It can be considered that the historical data is original fault data. Alternatively, for the same optical module, the proportion of the historical data that mutates at a plurality of mutation time points and exceeds a certain proportion (for example, 80%) can also be considered as original fault data. For this embodiment, because the optical modules are classified according to the type, the original fault data determined for optical modules of different types is different. This reflects the difference between devices of different manufacturers and the excellent universality of the fault processing method in the embodiments of the present application.

[0042] After the original fault data is determined, the fault model can be trained according to the original fault data and according to the type of the optical module. Similarly, the fault model of the optical module of different types is different.

[0043] In some embodiments, training the fault model according to the type of the optical module according to the original fault data can specifically include:

[0044] The original fault data is screened to obtain target fault data;

[0045] The target fault data is sampled according to the normal-to-fault ratio of the optical module to obtain sampling data;

[0046] According to the sampling data, the model is trained. Although the original fault data can be directly used for fault model training of the optical module, there may still be some abnormalities, such as gaps in the analysis of historical data, data abnormalities, or non-fault related data when the original fault data is obtained from the sudden historical data. Therefore, the original fault data can be further screened to obtain target fault data. The target fault data has a higher correlation with the optical module fault, and the target fault data can more accurately reflect the fault condition of the optical module and can more accurately detect and analyze the fault of the optical module.

[0047] Because the probability of optical module failure in the actual system is low, most of the optical modules are not faulty, so the target fault data can be sampled according to the normal-to-fault ratio of the optical module to obtain sampling data. Then, the model is trained according to the sampling data.

[0048] In some embodiments, the screening of the original fault data specifically includes:

[0049] Deleting abnormal data in the original fault data; and / or,

[0050] Extracting data representing the timing characteristics of the optical module from the original fault data. The abnormal data mainly refers to missing data or data with obvious errors in the collection process; the data representing the timing characteristics of the optical module refers to the data representing the significant features in time that are extracted for the characteristics of the optical module timing.

[0051] In some embodiments, the fault model training according to the type of the optical module can include:

[0052] According to the sampling data corresponding to optical modules of different types, the fault model is trained for optical modules of different types respectively. The fault data corresponding to the faults of optical modules of different types is different, and the corresponding fault models are also different.

[0053] In some embodiments, the fault model can include at least one of a machine learning model, a statistical model, and a deep learning model.

[0054] In some embodiments, after the fault model training according to the type of the optical module, the method can further include:

[0055] According to the result of the fault model training, a fault risk prediction result of the corresponding optical module within a preset time is output. The result of the fault model training can be directly used for prediction, that is, to predict whether some optical modules have failed, or when the failure will occur, what kind of failure will occur, and the like, to provide convenience for system operation and maintenance.

[0056] In some embodiments, the method can further include:

[0057] The prediction result of the fault risk is evaluated through the preset evaluation index. The result of the training based on the fault model is always likely to have a deviation, and therefore the prediction result can be monitored by setting the evaluation index; the actual fault condition can be cross-compared with the prediction result to obtain the evaluation index corresponding to the fault model. Once the evaluation index triggers a condition, the fault model needs to be updated.

[0058] In some embodiments, the method further comprises:

[0059] collecting supplementary historical data corresponding to the mis-detected and missed-detected optical modules;

[0060] updating the fault model according to the supplementary historical data when the preset evaluation index reaches an update threshold. When the historical data of the optical modules are analyzed, the optical modules that are originally normal may be mistakenly considered as faulty optical modules, that is, mis-detection, or the historical data of the faulty optical modules may be missed, that is, missed detection. Mis-detection and missed detection can also occur when the fault conditions of the optical modules are predicted. When mis-detection or missed detection occurs, the mis-detected and missed-detected data, that is, the supplementary historical data, can be collected first, and the fault model can be updated according to the supplementary historical data when the evaluation index reaches the update threshold.

[0061] The fault processing method provided in this embodiment analyzes the historical data of each optical module according to the model of the optical module, determines the historical data that suddenly changes when the optical module of the corresponding model is faulty, extracts the corresponding original fault data from the historical data that suddenly changes, and trains the fault model according to the original fault data according to the model of the optical module. Therefore, by dividing according to the model of the optical module, the flexibility of collecting the historical data of the faulty optical module is improved, and the application range of fault detection is further improved, effectively reducing the operation and maintenance cost of the optical module.

[0062] Embodiment Two

[0063] The fault processing method provided in this embodiment is described below with reference to Figure 2 The method comprises the following steps:

[0064] S201, historical data (such as performance data, inspection data, etc.) during the operation of the optical module is input into a processing system of the fault processing method;

[0065] S202, the historical data is classified according to the model of the optical module to obtain the historical data corresponding to each optical module classified according to the model of the optical module;

[0066] S203, for the historical data of different types of optical modules, analyze the "mutation points" of each optical module in the time sequence, that is, determine the time point of failure occurrence, and obtain the historical data of the optical modules that appear mutation proportion reaching a certain proportion as the original fault data;

[0067] S204, screening the original fault data, removing unnecessary data, and rejecting the collected incorrect data;

[0068] S205, further screening the original fault data, extracting the data capable of representing the time sequence characteristics of the optical module as the target fault data;

[0069] S206, according to the normal and fault optical module proportion of the current data, determine the sampling algorithm, and generate the sampling data;

[0070] S207, according to the upstream input sampling data, classify according to the type of optical module, and obtain the optical module sampling data divided by type;

[0071] S208, according to different types of optical modules and corresponding sampling data, train different fault models, the fault model adopts a machine learning model, which can also be replaced by a statistical model or a deep learning model according to demand and computing power; wherein, in the prediction stage, the fault model can be directly used to output the prediction result;

[0072] S209, the prediction result is summarized as the system output result for output;

[0073] S210, through the preset evaluation index, the model detection ability of the current system output result is evaluated;

[0074] S211, collect the misdetected (normal optical module detected as fault) and missed (fault optical module detected as normal) optical modules for subsequent model updating;

[0075] S212, when the evaluation index meets the incremental update condition, trigger the system to perform incremental update;

[0076] S213, when the evaluation index meets the full update condition, trigger the system to perform full update;

[0077] S214, the output result is directly output to the user for viewing, or output to other downstream interfacing systems.

[0078] Embodiment three

[0079] This embodiment provides a deployment process of a fault processing method, please refer to Figure 3 , the method comprises:

[0080] S301, read historical data (such as performance data or inspection data) of the optical module from the upstream docking system;

[0081] S302, classify the historical data according to the model information of the optical module;

[0082] S303, for each type of optical module, extract the historical data of the optical module that has failed;

[0083] S304, analyze each historical data of each failed module, and analyze whether there is a "trend mutation point" in each historical data;

[0084] S305, if there is, add +1 to the statistical counter;

[0085] S306, if there is not, continue to traverse other historical data until all analysis is completed;

[0086] S307, after the statistics are completed, use the top n historical data with the highest statistical proportion as the original fault data;

[0087] S308, clean the data, including deleting abnormal data and other operations;

[0088] S309, for the time sequence characteristics of the optical module data, extract data that can represent the characteristics in time as target fault data;

[0089] S310, analyze the proportion of normal and fault samples in the data, and sample according to the sampling algorithm;

[0090] S311, start the training process, train the fault model according to the optical module data of different types, and after the training is completed, the deployment is completed.

[0091] Embodiment four

[0092] This embodiment provides a fault processing method for detecting optical module faults, please refer to Figure 4 , the method comprises:

[0093] S401, periodically run the fault detection task, and read the historical data of the optical module in the recent period from the upstream docking system;

[0094] S402, classify the historical data according to the model of the optical module;

[0095] S403, for the time sequence characteristics of the optical module data, extract data that can represent the characteristics in time;

[0096] S404, for the data of each type of optical module, respectively input into the corresponding fault model which has been trained, and output the probability of possible failure;

[0097] S405, monitoring the prediction result;

[0098] S406, outputting the prediction result to an end user or other downstream system.

[0099] Embodiment five

[0100] The embodiment provides a monitoring update process of a fault processing method, please refer to Figure 5 The method comprises the following steps:

[0101] S501, the system reaches the set task execution time, and runs the monitoring task;

[0102] S502, the collected prediction results are counted;

[0103] S503, the fault condition of the current optical module is counted, such as whether there is a serious alarm, whether it is replaced, whether there is a maintenance order, etc.;

[0104] S504, the actual fault condition is compared with the prediction result, and the evaluation index is calculated;

[0105] S505, whether the evaluation index reaches the update threshold (which may be full update or incremental update) is judged;

[0106] S506, if not, continue to collect the currently missed and missed module data;

[0107] S507, if the threshold is reached, the current historical data is prepared, and the collected missed and missed data is supplemented;

[0108] S508, whether the full update threshold is reached is determined;

[0109] S509, if the threshold is reached, the data in S507 is used to update the fault model;

[0110] S510, if not, whether the incremental update threshold is reached is determined;

[0111] S511, if the threshold is reached, the data in S507 is used to update the fault model.

[0112] Embodiment six

[0113] The embodiment provides a fault processing device, please refer to Figure 6 The device comprises:

[0114] The data analysis module 61 is used for analyzing the historical data of each optical module according to the type of the optical module, and determining the historical data of the corresponding type of optical module which is mutated when the optical module is faulty;

[0115] a fault extraction module 62, configured to extract corresponding original fault data from the history data with mutation;

[0116] a model training module 63, configured to train a fault model according to the original fault data and according to the model of the optical module.

[0117] In order to improve the applicability of the optical module fault processing device in the embodiment, the optical module fault processing device in the embodiment can be well applied to optical modules produced by various manufacturers. In the embodiment, the history data of the optical module is analyzed according to the model of the optical module, that is, the analysis results of the history data are different for optical modules of different models. The causes and effects of the faults of the optical modules can be similar, but the data changes when the faults occur can be different for optical modules of different models, which brings the difference of the history data with mutation in the analysis results of the history data of the optical module. In the embodiment, the difference is considered, and the history data with mutation when the faults occur is obtained for optical modules of different models according to the model of the optical module.

[0118] The history data with mutation indicates that the index of the history data changes obviously in a short time when the optical module fails, which can be obviously increased or obviously decreased. In the optical module fault detection problem, it can be defined as "in the history data of the optical module, at a certain time point, due to a special event-failure, some data before and after the time point changes obviously". The some data referred to here is the history data with mutation. The obvious change mentioned here can be determined according to the actual data itself, for example, for normal indicators, the increase or decrease amplitude in a short time exceeds 50%. The embodiment only lists one possible mutation amplitude, and the mutation amplitude of different data cannot be generalized. The mutation amplitude can be determined according to the actual system operation and the type of the history data, and the embodiment does not limit it.

[0119] When analyzing the mutation of each history data, the most mutated indicators are counted, and the history data that does not mutate or does not mutate obviously can be directly ignored. Of course, in order to improve the comprehensiveness of detection and avoid the adverse consequences caused by missed judgment and misjudgment, these data can be retained, and the fault model can be updated in the subsequent update stage.

[0120] The historical data of the optical module in the embodiment can be performance data, inspection data, etc. Before analyzing the historical data of each optical module according to the model of the optical module, the optical modules can be classified according to the model of the optical module. The classification standard of the optical modules of various models can be classification according to the manufacturer, classification according to the working principle of the optical module, classification according to the specifications and applicable range of the optical module, etc.

[0121] Not all of the historical data in the mutated historical data is related to the fault. For example, for the optical modules of the same model that have failed, there are some mutated historical data that are occasional, and only individual failed optical modules can generate such data, which should not be regarded as fault data. In other words, the fault data should be the historical data that has a high mutation rate when the optical module of the corresponding model fails. Such historical data can be used as original fault data. Specifically, extracting the corresponding original fault data from the mutated historical data can include:

[0122] analyzing the mutation time point in the time sequence according to the mutated historical data;

[0123] determining the historical data in which the mutation rate exceeds a preset proportion according to the mutation time point;

[0124] The historical data with a mutation proportion exceeding a preset proportion is taken as the original fault data. This scheme for determining the original fault data can greatly exclude occasional mutation historical data and retain historical data that is more likely to be related to faults. The mutation time points are analyzed. The mutation time points can be considered as time points at which the optical module fails. The mutation historical data is more concentrated at these time points. However, even the mutation historical data at the mutation time points is not all caused by faults, and therefore can be further screened. The historical data with a mutation proportion exceeding a preset proportion is taken as the original fault data. The mutation proportion refers to that the proportion of the historical data that mutates at the mutation time points exceeds a threshold. For example, for a mutation time point, the proportion of the historical data that mutates on a plurality of optical modules of the same type and that fails exceeds a certain proportion (for example, 80%). It can be considered that the historical data is original fault data. Alternatively, for the same optical module, the proportion of the historical data that mutates at a plurality of mutation time points and exceeds a certain proportion (for example, 80%) can also be considered as original fault data. For this embodiment, because the optical modules are classified according to the type, the original fault data determined for optical modules of different types is different. This reflects the difference between devices of different manufacturers and the excellent universality of the fault processing device in the embodiments of the present application.

[0125] After the original fault data is determined, the fault model can be trained according to the original fault data and according to the type of the optical module. Similarly, the fault model of the optical module of different types is different.

[0126] In some embodiments, training the fault model according to the type of the optical module according to the original fault data can specifically include:

[0127] The original fault data is screened to obtain target fault data;

[0128] The target fault data is sampled to obtain sampling data according to a normal-to-fault ratio of the optical module;

[0129] According to the sampling data, the model is trained. Although the original fault data can be directly used for fault model training of the optical module, there may still be some abnormalities, such as gaps in the analysis of historical data, data abnormalities, or non-fault related data when the original fault data is obtained from the sudden historical data. Therefore, the original fault data can be further screened to obtain target fault data. The target fault data has a higher correlation with the optical module fault, and the target fault data can more accurately reflect the fault condition of the optical module and can more accurately detect and analyze the fault of the optical module.

[0130] Because the probability of optical module failure in the actual system is low, most of the optical modules are not faulty, so the target fault data can be sampled according to the normal-to-fault ratio of the optical module to obtain sampling data. Then, the model is trained according to the sampling data.

[0131] In some embodiments, the screening of the original fault data specifically includes:

[0132] Delete abnormal data in the original fault data; and / or,

[0133] Extract data representing the timing characteristics of the optical module from the original fault data. The data representing the timing characteristics of the optical module refers to data that can represent significant features in time with respect to the characteristics of the timing of the optical module.

[0134] In some embodiments, the fault model training according to the model of the optical module can include:

[0135] According to the sampling data corresponding to optical modules of different models, the fault model is trained for optical modules of different models respectively. The fault data corresponding to the faults of optical modules of different models is different, and the corresponding fault models are also different.

[0136] In some embodiments, the fault model can include at least one of a machine learning model, a statistical model, and a deep learning model.

[0137] In some embodiments, after the fault model training according to the model of the optical module, it can further include:

[0138] According to the result of the fault model training, the fault risk prediction result of the corresponding optical module within a preset time is output. The result of the fault model training can be directly used for prediction, that is, to predict whether some optical modules have failed, or when they will fail, what kind of failure will occur, and the like, to facilitate the operation of the system.

[0139] In some embodiments, it can further include:

[0140] The prediction result of the fault risk is evaluated through the preset evaluation index. The result of the training based on the fault model is always likely to have deviation, and therefore the prediction result can be monitored by setting the evaluation index; the actual fault condition can be cross-compared with the prediction result to obtain the evaluation index corresponding to the fault model. Once the evaluation index triggers a condition, the fault model needs to be updated.

[0141] In some embodiments, the method further comprises:

[0142] Collecting the supplementary historical data corresponding to the mis-detected and missed-detected optical modules;

[0143] When the preset evaluation index reaches an update threshold, updating the fault model according to the supplementary historical data. When analyzing the historical data of the optical modules, the optical modules that are originally normal may be mistakenly considered as faulty optical modules, i.e. mis-detection, or the historical data of the faulty optical modules may be missed, i.e. missed detection. The missed detection and mis-detection can also occur when predicting the fault conditions of the optical modules. When the mis-detection or missed detection occurs, the mis-detected and missed-detected data, i.e. the supplementary historical data, can be collected first, and when the evaluation index reaches the update threshold, the fault model can be updated according to the supplementary historical data.

[0144] The fault processing apparatus provided in the embodiment comprises a data analysis module configured to analyze the historical data of each optical module according to the model of the optical module, determine the historical data that has a mutation when the optical module of the corresponding model is faulty, a fault extraction module configured to extract the original fault data corresponding to the historical data that has a mutation, and a model training module configured to train a fault model according to the original fault data according to the model of the optical module. In this way, by dividing according to the model of the optical module, the flexibility of collecting the historical data of the faulty optical module is improved, and the application range of the fault detection is further improved, and the operation and maintenance cost of the optical module is effectively reduced.

[0145] Please refer to Figure 7 , Figure 7 Fig. 1 shows the detailed composition of the fault processing apparatus in the fault processing system in the embodiment, wherein:

[0146] The data analysis module 61 is mainly responsible for classifying data and extracting fault data indicators of different types of optical modules through a mutation point analysis algorithm; the data analysis module 61 includes a first device classification submodule 611 and an indicator extraction submodule 612, wherein the first device classification submodule 611 is mainly responsible for classifying data according to the type and other information of the optical module; the indicator extraction submodule 612 analyzes the "mutation point" on the time sequence for the fault data of different types of optical modules, obtains historical data with a high mutation proportion in the fault module, and inputs the historical data as original fault data to the downstream.

[0147] The fault extraction module 62 is mainly responsible for simplifying the data, cleaning the abnormal data, extracting new features, and sampling the data according to the original fault data provided by the upstream subsystem; the fault extraction module 62 includes an abnormal cleaning submodule 621, a time sequence extraction submodule 622, and a sampling submodule 623, wherein the abnormal cleaning submodule 621 uses the original fault data extracted by the upstream to remove unnecessary data and eliminate data with errors; the time sequence extraction submodule 622 extracts data that can represent significant features in time according to the characteristics of the optical module time sequence; the sampling submodule 623 formulates a sampling algorithm according to the proportion of normal and fault optical modules in the current data, generates sampling data, and delivers the sampling data to the downstream.

[0148] The model training module 63 is mainly responsible for training the fault detection model and completing subsequent prediction tasks; the model training module 63 includes a second device classification submodule 631 and a result processing submodule 632, wherein the second device classification submodule 631 classifies the sampling data input by the upstream according to the type of the optical module; different models can be trained according to different types of modules and input data; the model adopts a machine learning model, and can be replaced by a statistical model or a deep learning model according to the demand and computing power; in the prediction stage, the model output result can be directly used; the result processing submodule 632 is used to summarize the results as system results output, and simultaneously transmits the results to the downstream.

[0149] The model updating module 64 is mainly responsible for monitoring the results of the upstream output, triggering system model updating when the evaluation index decreases; the model updating module 64 includes a result evaluation submodule 641, a collection submodule 644, an incremental updating trigger 642 and a full updating trigger 643, wherein the result evaluation submodule 641 is used for model detection capability evaluation of the current output result through the evaluation index; the collection submodule 644 is used for collecting the mis-detection (i.e. normal module detection as fault) and the missed detection (i.e. fault module detection as normal) of the optical module data for subsequent model updating; the incremental updating trigger 642 is controlled by the result evaluation submodule 641, triggering the system to perform incremental updating; the full updating trigger 643 is also controlled by the result evaluation submodule 641, triggering the system to perform full updating.

[0150] The model updating module 64 can also output the results directly to the user.

[0151] Embodiment Seven

[0152] The embodiment also provides a network device, please refer to Figure 8 which includes a processor 81, a memory 82 and a communication bus 83;

[0153] The communication bus 83 is used to realize the connection communication between the processor 81 and the memory 82;

[0154] The processor 81 is used to execute one or more computer programs stored in the memory 82 to realize the steps in the fault processing method in the above-mentioned embodiments, which will not be repeated here.

[0155] The embodiment also provides a computer readable storage medium, which includes a volatile or non-volatile, removable or non-removable medium implemented in any method or technology for storing information (such as computer readable instructions, data structures, computer program modules or other data). The computer readable storage medium includes but is not limited to RAM (Random Access Memory), ROM (Read-Only Memory), EEPROM (Electrically Erasable Programmable read only memory), flash memory or other memory technology, CD-ROM (Compact Disc Read-Only Memory), digital versatile disc (DVD) or other optical disc storage, magnetic cassette, magnetic tape, magnetic disk storage or other magnetic storage device, or any other medium that can be used to store desired information and can be accessed by a computer.

[0156] The computer readable storage medium in the embodiment can be used to store one or more computer programs, which can be executed by one or more processors to implement the steps of the fault processing method in the above embodiments.

[0157] The embodiment also provides a computer program (or computer software), which can be distributed on a computer readable medium and executed by a computing device to implement the steps of the fault processing method in the above embodiments; and in some cases, at least one step shown or described can be executed in an order different from that described in the above embodiments.

[0158] The embodiment also provides a computer program product, which includes a computer readable device on which the computer program shown above is stored. The computer readable device in the embodiment can include the computer readable storage medium shown above.

[0159] It can be seen that all or some steps in the method disclosed above, the functions of the modules / units in the system and the device can be implemented as software (which can be implemented as computer program codes executable by a computing device), firmware, hardware and appropriate combinations thereof. In the hardware implementation, the division between the functional modules / units mentioned in the above description does not necessarily correspond to the division of physical components; for example, one physical component can have multiple functions, or one function or step can be performed by several physical components in cooperation. Some or all of the physical components can be implemented as software executed by a processor, such as a central processing unit, a digital signal processor or a microprocessor, or as hardware, or as an integrated circuit, such as an application specific integrated circuit.

[0160] In addition, it is well known to those skilled in the art that a communication medium generally contains computer readable instructions, data structures, computer program modules or other data in a modulated data signal such as a carrier wave or other transmission mechanism, and can include any information delivery medium. Therefore, the present application is not limited to any specific combination of hardware and software.

[0161] The above is a further detailed description of the embodiments of the present application in combination with specific embodiments, and cannot be considered as limiting the specific implementation of the present application to these descriptions. For those skilled in the art, without departing from the concept of the present application, a number of simple deductions or substitutions can be made, which should be considered as falling within the protection scope of the present application.

Claims

1. A fault handling method, comprising: Analyze historical data of each optical module according to the model of the optical module to determine historical data of sudden changes in the optical module of the corresponding model when a fault occurs; Extracting corresponding original fault data from the historical data that has undergone mutation; Screening the original fault data to obtain target fault data; Sampling the target fault data to obtain sampled data according to a normal-to-fault ratio of the optical module; Model training is performed based on the sampled data.

2. The fault handling method according to claim 1, wherein: Before analyzing the historical data of each optical module according to the optical module model, the following steps are also included: The optical modules are classified based on their models.

3. The fault handling method according to claim 1, wherein: The extracting corresponding original fault data from the historical data of the sudden change includes: Analyze the mutation time points in the time series based on the historical data of the mutation; Determine, based on the mutation time point, historical data in which a mutation ratio exceeds a preset ratio in the historical data; The historical data in which the mutation ratio exceeds a preset ratio is used as the original fault data.

4. The fault handling method according to claim 1, wherein: The screening of the original fault data includes: Deleting abnormal data in the original fault data; and / or, Data characterizing the timing characteristics of the optical module is extracted from the original fault data.

5. The fault handling method according to claim 4, characterized in that: The fault model training according to the optical module model includes: According to the sampling data corresponding to different types of optical modules, fault model training is performed for different types of optical modules.

6. The fault handling method according to claim 5, characterized in that: The fault model includes at least one of a machine learning model, a statistical model, and a deep learning model.

7. The fault handling method according to any one of claims 1 to 3, characterized in that: After the fault model training is performed according to the optical module model, the method further includes: According to the results of fault model training, the fault risk prediction results of the corresponding optical module within the preset time are output.

8. The fault handling method according to claim 7, wherein: Also includes: The fault risk prediction result is evaluated by using preset evaluation indicators.

9. The fault handling method according to claim 8, characterized in that: Also includes: Collect supplementary historical data corresponding to misdetected and missed optical modules; When the preset evaluation index reaches an update threshold, the fault model is updated according to the supplementary historical data.

10. A fault handling device, comprising: A data analysis module is used to analyze the historical data of each optical module according to the model of the optical module, and determine the historical data of sudden changes in the optical module of the corresponding model when a fault occurs; A fault extraction module is used to extract corresponding original fault data from the historical data that has undergone mutation; A model training module is used to screen the original fault data to obtain target fault data; and to sample the target fault data to obtain sampled data based on the normal-to-fault ratio of the optical module; Model training is performed based on the sampled data.

11. A network device comprising a processor, a memory and a communication bus; The communication bus is used to realize the connection and communication between the processor and the memory; The processor is configured to execute one or more computer programs stored in the memory to implement the steps of the fault handling method according to any one of claims 1 to 9.

12. A computer-readable storage medium, characterized in that The computer-readable storage medium stores one or more computer programs, and the one or more computer programs can be executed by one or more processors to implement the steps of the fault handling method according to any one of claims 1 to 9.

Images