Network group fault detection method and network group fault detection model training method and device

Abstract

The embodiment of the invention provides a network group fault detection method and a network group fault detection model training method and device. The method comprises the following steps: acquiring a first equipment state of first network equipment in a network system and a second equipment state of second network equipment managed by the first network equipment; the first equipment state comprises that the first network equipment is in an online or offline state, and the second equipment state comprises that the second network equipment is in an online or offline state; acquiring a comprehensive equipment state of the network system according to the first equipment state and the second equipment state; inputting the state of the comprehensive equipment into the trained network group fault detection model, and obtaining a network group fault detection result of the network system; the network group fault detection model is obtained by training a comprehensive equipment state sample, the comprehensive equipment state sample is a sample obtained according to a first equipment state sample of the first network equipment and a second equipment state sample of the second network equipment, and network group fault detection can be rapidly, accurately and automatically carried out.

Description

Technical Field

[0001] The present invention belongs to the technical field of network detection, and particularly relates to a network group fault detection method, a network group fault detection model training method, a device, an electronic device, and a readable storage medium. Background Art

[0002] The network group fault phenomenon refers to the phenomenon that multiple network devices all have network faults within the same time period. Through network group fault detection, the fault points can be promptly investigated when a network group fault event occurs in the network system to ensure the normal operation of the network.

[0003] In the related art, users report network fault events, and when the number of network fault events reaches a certain number threshold, they are determined as network group fault events and reported for maintenance personnel to handle the faults.

[0004] This processing method has problems such as difficulty in promptly discovering network group faults, long network group fault positioning time, and high consumption of a large amount of labor costs. Summary of the Invention

[0005] The present invention provides a network group fault detection method, a network group fault detection model training method, a device, an electronic device, and a readable storage medium to solve the technical problems in the related art that it is difficult to promptly discover network group faults and it is difficult to locate network group faults.

[0006] In a first aspect, the present invention provides a network group fault detection method, including:

[0007] Obtain the first device state of a first network device in the network system and the second device state of a second network device managed by the first network device; the first device state includes whether the first network device is in an online or offline state, and the second device state includes whether the second network device is in an online or offline state;

[0008] Obtain the comprehensive device state of the network system according to the first device state and the second device state;

[0009] Input the comprehensive device state into a trained network group fault detection model to obtain the network group fault detection result of the network system; the network group fault detection model is a model trained through comprehensive device state samples, and the comprehensive device state samples are samples obtained according to the first device state samples of the first network device and the second device state samples of the second network device.

[0010] In a second aspect, the present invention provides a network group model training method for training the network group model training model in the network group identification method as described in the first aspect, including:

[0011] Obtain the first device status sample of the first network device in the network system, as well as the second device status sample of the second network device managed by the first network device; the first device status sample includes whether the first network device is in an online or offline state, and the second device status sample includes whether the second network device is in an online or offline state;

[0012] Obtain the comprehensive device status sample of the network system according to the first device status sample and the second device status sample;

[0013] Train the network group fault detection model with the comprehensive device status sample to obtain a trained network group fault detection model.

[0014] In a third aspect, the present invention provides a network group fault detection device, and the device includes:

[0015] A first acquisition module, configured to acquire the first device status of the first network device in the network system, as well as the second device status of the second network device managed by the first network device; the first device status includes whether the first network device is in an online or offline state, and the second device status includes whether the second network device is in an online or offline state;

[0016] A second acquisition module, configured to acquire the comprehensive device status of the network system according to the first device status and the second device status;

[0017] A third acquisition module, configured to input the comprehensive device status into the trained network group fault detection model to obtain the network group fault detection result of the network system; the network group fault detection model is a model trained with a comprehensive device status sample, and the comprehensive device status sample is a sample obtained according to the first device status sample of the first network device and the second device status sample of the second network device.

[0018] In a fourth aspect, the present invention provides a network group fault model training device for training the network group fault model training model in the network group fault recognition method as described in the first aspect, including:

[0019] A fourth acquisition module, configured to acquire the first device status sample of the first network device in the network system, as well as the second device status sample of the second network device managed by the first network device; the first device status sample includes whether the first network device is in an online or offline state, and the second device status sample includes whether the second network device is in an online or offline state;

[0020] A fifth acquisition module, configured to acquire the comprehensive device status sample of the network system according to the first device status sample and the second device status sample;

[0021] A sixth acquisition module, configured to obtain a trained network group fault detection model by training a network group fault detection model through the comprehensive device status.

[0022] In a fifth aspect, the present invention provides an electronic device, including:

[0023] A processor, a memory, and a computer program stored on the memory and executable on the processor, where when the processor executes the program, the method described in any one of the first or second aspects is implemented.

[0024] In a sixth aspect, the present invention provides a readable storage medium, when instructions in the storage medium are executed by a processor of an electronic device, enabling the electronic device to execute the method described in any one of the first or second aspects.

[0025] In this embodiment, the first device status of the first network device in the network system and the second device status of the second network device managed by the first network device are obtained; the first device status includes whether the first network device is in an online or offline state, and the second device status includes whether the second network device is in an online or offline state; according to the first device status and the second device status, the comprehensive device status of the network system is obtained; the comprehensive device status is input into the trained network group fault detection model to obtain the network group fault detection result of the network system; the network group fault detection model is a model trained through comprehensive device status samples, and the comprehensive device status samples are samples obtained according to the first device status samples of the first network device and the second device status samples of the second network device. The method of this embodiment can quickly and accurately detect network group faults, thereby greatly improving the user experience and reducing the impact of group faults. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0027] Figure 1 is a flowchart of the steps of a network group fault inspection method provided by an embodiment of the present invention;

[0028] Figure 2 is a flowchart of the steps of another network group fault inspection method provided by an embodiment of the present invention;

[0029] Figure 3 is a flowchart of the steps of yet another network group fault inspection method provided by an embodiment of the present invention;

[0030] Figure 4 It is a schematic diagram of a network system provided by an embodiment of the present invention;

[0031] Figure 5 It is a grayscale binary image provided by an embodiment of the present invention;

[0032] Figure 6 It is a schematic diagram of the detection result of network group faults provided by an embodiment of the present invention;

[0033] Figure 7 It is a flowchart of the steps of a method for training a network group fault inspection model provided by an embodiment of the present invention;

[0034] Figure 8 It is a flowchart of the steps of a method for inspecting network group faults provided by an embodiment of the present invention;

[0035] Figure 9 It is a structural diagram of a network group fault inspection device provided by an embodiment of the present invention;

[0036] Figure 10 It is a structural diagram of a device for training a network group fault inspection model provided by an embodiment of the present invention;

[0037] Figure 11 It is a structural diagram of an electronic device provided by an embodiment of the present invention. Detailed implementation manners

[0038] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0039] Figure 1 It is a flowchart of the steps of a broadband group fault method provided by an embodiment of the present invention. As Figure 1 shown, the method may include:

[0040] Step 101, obtain the first device status of the first network device in the network system and the second device status of the second network device managed by the first network device.

[0041] Among them, the first device status includes whether the first network device is online or offline, and the second device status includes whether the second network device is online or offline.

[0042] Exemplarily, the first device state includes: the first network device corresponding to the first device state is in an online or offline state; the second device state includes: the second network device corresponding to the second device state is in an online state, or in a state of transitioning from online to offline.

[0043] Exemplarily, the device state can be represented by its corresponding eigenvalue. For example, the online state is represented by the eigenvalue 1, and the offline state is represented by the eigenvalue 0. Exemplarily, the first network device can be an optical line terminal (OLT device), and the second network device can be an optical network unit (ONU device).

[0044] Furthermore, the OLT device periodically detects the ONU device in the offline state. If within the set time, it is detected that the ONU device transitions from the online state to the offline state, the device state of the ONU device is represented by the eigenvalue 0; if the ONU device is normally online, the device state of the ONU device is represented by the eigenvalue 1; if the ONU device is offline for more than the preset time, the ONU device is ignored, and in the case of ignoring the ONU device, the device state of the ONU device is not recorded in the subsequent processing.

[0045] Among them, the offline of a single ONU device may be caused by a fault, or by the user shutting down or a single household power outage. When the ONU device is offline, it indicates that the network where the ONU device is located (such as home broadband) is in an unavailable state. Among them, the offline of a single ONU device is not necessarily caused by a network group fault, but is a factor used to determine whether a network group fault has occurred.

[0046] Exemplarily, the first network device is an OLT device, and the second network device is an ONU device. The OLT device can periodically execute the ONU device status query command to obtain the second device state of the ONU device, or detect the uplink data to count the number of uplink frames of the ONU device, and obtain the second device state of the ONU device according to the statistical result.

[0047] Step 102, obtain the comprehensive device state of the network system according to the first device state and the second device state.

[0048] Exemplarily, integrate the first device information and the second device state to obtain the comprehensive device state of the network system.

[0049] Specifically, if the first network device information indicates that the first network device is online, then according to the second device status of the second network devices subordinate to the first network device, construct a first comprehensive device status in a one-dimensional data format corresponding to the first network device, and then integrate the first comprehensive device statuses in the one-dimensional data format corresponding to all first network devices to obtain a second comprehensive device status in a two-dimensional data format. Determine the second comprehensive device status in the two-dimensional data format as the comprehensive device status in this step.

[0050] Furthermore, the network system further includes a third network device, which can be a network element, and can integrate the second comprehensive device statuses in the two-dimensional data format corresponding to all first network devices subordinate to the third network device to obtain a second comprehensive device status in a three-dimensional data format. Determine the second comprehensive device status in the three-dimensional data format as the comprehensive device status in this step.

[0051] If the first network device information indicates that the first network device is offline, then mark the second device statuses of all second network devices subordinate to the first network device as offline statuses. For example, if the device status is characterized by eigenvalues, then in this case, set the eigenvalues corresponding to the second device statuses of all second network devices subordinate to the first network device to 0.

[0052] Step 103: Input the comprehensive device status into the trained network group fault detection model to obtain the network group fault detection result of the network system.

[0053] Among them, the network group fault detection model is a model trained through comprehensive device status samples, and the comprehensive device status samples are samples obtained based on the first device status samples of the first network device and the second device status samples of the second network device.

[0054] Exemplarily, the comprehensive device status includes multiple eigenvalues, each eigenvalue is used to characterize the device status of the corresponding network device. Construct a grayscale binary image according to the comprehensive device status including multiple eigenvalues, and input the grayscale binary image into the trained network group fault detection model to obtain the network group fault detection result of the network system. Among them, in the grayscale binary image, the grayscale binary includes eigenvalue 1 and 0, eigenvalue 1 indicates that the corresponding network device is in an online state, and eigenvalue 0 indicates that the corresponding network device is in an offline state.

[0055] Further, inputting the comprehensive device status into the trained network group fault detection model to obtain the network group fault detection results of the network system may include: whether a group fault occurs in the network system, the type of group fault, the first network device related to the group fault, the second network device, etc. Further, by associating the group fault type with a preset knowledge base, the cause of the group fault and the solution suggestions for solving the group fault can be determined. Based on the network devices related to the group fault and the cause of the group fault, the fault point can be quickly determined. Among them, the preset knowledge base stores at least the preset group fault types and their corresponding group fault causes and solution suggestions for solving the group fault. Correspondingly, when training the network group fault detection model, the comprehensive device status samples are used as training samples, and whether a group fault occurs in the network system and the network devices affected by the fault are used as labels. The network devices affected by the fault are specifically manifested in that the online / offline status of the network devices changes due to the group fault.

[0056] Exemplarily, the obtained image is called through the service interface of the image classification model to achieve real-time fault detection and problem location.

[0057] Exemplarily, the method of this embodiment can be applied to the application scenario of home broadband group fault detection, and can also be applied to the application scenario of network group fault detection in enterprises and the like.

[0058] In the related art, when a fault occurs in the home broadband, it is necessary for users to report the fault, and then the customer service confirms and notifies the telecommunications staff to conduct fault troubleshooting. Finally, the cause of the fault is confirmed and the fault is processed, and the group fault event alarm is triggered only when the number of user-reported faults reaches a certain level. The group fault detection method in the related art has disadvantages such as late discovery of network group faults, low efficiency, high labor costs, and long time-consuming for fault location.

[0059] In this embodiment, the first device status of the first network device in the network system and the second device status of the second network device managed by the first network device are obtained; the first device status includes whether the first network device is in an online or offline state, and the second device status includes whether the second network device is in an online or offline state; according to the first device status and the second device status, the comprehensive device status of the network system is obtained; the comprehensive device status is input into the trained network group fault detection model to obtain the network group fault detection results of the network system; the network group fault detection model is a model trained through comprehensive device status samples, and the comprehensive device status samples are samples obtained according to the first device status samples of the first network device and the second device status samples of the second network device. The method of this embodiment can quickly and accurately detect group faults, thereby greatly improving the user experience and reducing the impact of group faults.

[0060] Refer to Figure 2 , this embodiment provides a network group fault detection method, and the method may include the following steps:

[0061] Step 201, obtain the first device status of the first network device in the network system, and the second device status of the second network device managed by the first network device.

[0062] Among them, the first device status includes that the first network device is in an online or offline state, and the second device status includes that the second network device is in an online or offline state.

[0063] Exemplarily, there is at least one first network device.

[0064] Step 202, obtain the first comprehensive device status corresponding to the first network device according to the first device status of the first network device and the second device status of the second network device managed by the first network device.

[0065] Exemplarily, step 202 may include sub-steps A1 to A3:

[0066] Sub-step A1, if the first device status is that the first network device is in an online state, obtain the first comprehensive device status corresponding to the first network device according to the second device status of the second network device managed by the first network device.

[0067] Exemplarily, integrate the second device status of the second network device managed by the first network device into the first comprehensive device status in the form of a one-dimensional array, and the elements in the first comprehensive device status are the second device status of the second network device managed by the first network device.

[0068] Sub-step A2, if the first device status is that the first network device is in an offline state, mark the second device status of the second network device managed by the first network device as the offline state to obtain the status marking result of the second network device managed by the first network device.

[0069] Exemplarily, according to the marking result marked as the offline state, obtain the status marking result of the second network device managed by the first network device.

[0070] Sub-step A3, obtain the first comprehensive device status corresponding to the first network device according to the status marking result of the second network device managed by the first network device.

[0071] Specifically, integrate the status marking result of the second network device managed by the first network device to obtain the first comprehensive device status in the form of a one-dimensional array corresponding to the first network device. The elements in the first comprehensive device status are the status marking results of the second network device.

[0072] Exemplarily, there are multiple second network devices managed by the first network device; step 202 may include sub-steps B1 to B2:

[0073] Sub-step B1: According to the second device status, obtain the target second network device that has been continuously offline within a preset time period from multiple second network devices.

[0074] Sub-step B2: According to the first device status of the first network device and the second device status of other second network devices except the target second network device, obtain the first comprehensive device status corresponding to the first network device.

[0075] According to Sub-steps B1 to B2, when the target second network device is in a long-term offline state, ignore its second device status. In other words, when obtaining the first comprehensive device status corresponding to the first network device, do not include the second device status of the target second network device. Thus, the influence of excessive data on the sensitivity of state changes can be avoided, and the data processing accuracy can be improved.

[0076] Step 203: According to the first comprehensive device status corresponding to each first network device, obtain the comprehensive device status of the network system.

[0077] Exemplarily, integrate the first comprehensive device status corresponding to each first network device to obtain the comprehensive device status of the network system.

[0078] Step 204: Input the comprehensive device status into the trained network group fault detection model to obtain the network group fault detection result of the network system.

[0079] Among them, the network group fault detection model is obtained by training with comprehensive device status samples, and the comprehensive device status samples are obtained according to the first device status samples of the first network devices and the second device status samples of the second network devices.

[0080] Furthermore, the network group fault detection model is obtained by training with comprehensive device status samples and labels; the labels include: whether the network system corresponding to the comprehensive device status sample has a network group fault, and the type of the network group fault. In addition, the labels can also include the network devices affected by the fault. Thus, input the comprehensive device status into the trained network group fault detection model to obtain the network group fault detection result of the network system, and the network group fault detection result includes: whether there is a network group fault, the network devices affected by the fault, and the type of the network group fault.

[0081] Further, associating the network group barrier type in the group barrier detection result with a preset knowledge base can obtain specific group barrier causes and solution suggestions. Based on the network devices affected by the group barrier in the group barrier detection result and the group barrier causes obtained from the preset knowledge base, the fault points in the network system can be quickly determined. Among them, the preset knowledge base at least includes the correspondence between network group barrier types and group barrier causes and solution suggestions. Based on the method of this embodiment, the rapid autonomous discovery and problem location of network group barriers in network systems such as home broadband can be realized, and the impact of group barriers on network stability can be reduced.

[0082] The triggering of group barriers in related technologies depends on fixed threshold settings and cannot detect different forms of network group barriers. Based on the method of this embodiment, through the trained group barrier detection model, not only can it be detected whether a group barrier has occurred in the network system, but also the group barrier type can be obtained. In addition, through this embodiment, the change in the offline state of communication devices throughout the network can also be detected.

[0083] Exemplarily, the network system includes multiple third network devices, and the third network devices are used to manage the first network device. Refer to Figure 3 , the method may include the following steps:

[0084] Step 301, obtain the first device state of the first network device in the network system and the second device state of the second network device managed by the first network device.

[0085] Among them, the first device state includes whether the first network device is in an online or offline state, and the second device state includes whether the second network device is in an online or offline state.

[0086] Obtain the offline state change information of ONU devices, OLT devices, and network element devices of the whole network broadband, form a grayscale binary image, and perform dataset processing on historical operation and maintenance group barrier data. Among them, for the detection value of the offline state change, within the set time, if the device changes from the online state to the offline state, 0 in the binary value represents that the device becomes offline, 1 represents that the device is normally online, and if the device exceeds the set time, the device is ignored.

[0087] The OLT device collects the offline change binary values of the ONU devices under it, combines them into a one-dimensional array, and reports them to the network element; the network element periodically detects the online or offline state of the OLT device, and the one-dimensional arrays of multiple OLT devices are combined into a two-dimensional array.

[0088] It should be noted that when the network element detects that the OLT device is offline, the row where the OLT device is located is set to all 0, and the length is the number of ONU devices that the OLT device can support to be hung under it;

[0089] The two-dimensional arrays of multiple network elements can be combined into a three-dimensional array to represent the view of all network devices;

[0090] Use the full - network three - dimensional array when historical group faults occur and the full - network three - dimensional array when no group faults occur as the two classifications of the training set;

[0091] Mark the fault location and the corresponding pixel points of the fault in the data of historical group faults. For example, an ONU device is a pixel point, an OLT device is a row, and a network element is a plane;

[0092] Because ONU devices with an offline time exceeding the preset duration are ignored, the shapes of the training sets are not unified. The ROIPooling algorithm (existing technology) is used to integrate the irregular data so that it can be input into the algorithm for training;

[0093] Using the three - dimensional array as the input and whether group faults occur and the fault points as the annotations, input the pre - processed data set into the CNN model to train and obtain an image detection model;

[0094] Step 302, obtain the third device status of the third network device in the network system.

[0095] Among them, the third device status includes that the third network device is in an online or offline state.

[0096] Exemplarily, the first network device in the network system can be an OLT device, the second network device can be an ONU device, and the third network device can be a network element device.

[0097] In this embodiment, whether the ONU device changes from online to offline within a preset time period is used as a key change point to form a grayscale binary image, and the historical operation and maintenance group - fault data is processed into a data set. Through the CNN image multi - classification model, rapid discovery and fault location of group faults are realized

[0098] Step 303, obtain the second comprehensive device status corresponding to the third network device according to the third device status of the third network device, the first device status of the first network device managed by the third network device, and the second device status of the second network device managed by the first network device;

[0099] Exemplarily, step 303 may include sub - steps C1 to C2:

[0100] Sub - step C1, if the third device status is that the third network device is online, obtain the first comprehensive device status corresponding to the first network device according to the first device status of the first network device managed by the third network device and the second device status of the second network device managed by the first network device;

[0101] Sub - step C2, integrate the first comprehensive device statuses respectively corresponding to each first network device managed by the third network device to obtain the second comprehensive device status corresponding to the third network device.

[0102] Step 304: Obtain the comprehensive device status of the network system according to the second comprehensive device status corresponding to each third network device respectively.

[0103] Exemplarily, the data format of the first comprehensive device status is a one-dimensional data format, and the data format of the second comprehensive device status is a two-dimensional data format. Exemplarily, step 304 may include sub-step D1:

[0104] Sub-step D1: Integrate the second comprehensive device status corresponding to each third network device respectively to obtain the comprehensive device status of the network system in a three-dimensional data format.

[0105] Exemplarily, the data format of the second comprehensive device status is a two-dimensional data format, and the data format of the comprehensive device status is a three-dimensional data format; step 304 may include sub-steps E1 to E2:

[0106] Sub-step E1: Based on a preset regional normalization algorithm, perform regional normalization processing on the second comprehensive device status corresponding to each third network device respectively to obtain the processed second comprehensive device status.

[0107] Among them, the ROIPooling algorithm is a common method in object detection tasks. It was first proposed in Faster R-CNN. Its function is to project a series of RoIs (regions of interest) of different sizes onto the feature map, and then process them into a consistent size through pooling operations, so as to facilitate the subsequent network layers for processing. In previous network structures, the last few layers were often fully connected layers, so a fixed input size was required. Through regional normalization processing, the role of accelerating calculations is also played.

[0108] Sub-step E2: Obtain the comprehensive device status of the network system according to the processed second comprehensive device status.

[0109] Among them, the number of elements of each processed second comprehensive device status in the same dimension is correspondingly equal.

[0110] For example, the second comprehensive device status is in a two-dimensional data format, and the number of elements in the same row and the same column of each processed second comprehensive device status is equal.

[0111] In a network system, the number of network devices attached to different regions is inconsistent, resulting in the possibility that the number of second network devices attached to different third network devices may not be equal to each other, the number of first network devices attached to different third network devices may not be equal to each other, and the number of second network devices attached to different first network devices may not be equal to each other. This will lead to inconsistent shapes of the data sets of the second comprehensive device states. Through the regional normalization algorithm, the shapes of the data sets of each second comprehensive device state can be made consistent. Further, through the regional normalization algorithm, the states of the data sets of the comprehensive device states of different network systems can be made consistent. For example, if the comprehensive device state is in a three-dimensional data format, through the regional normalization algorithm, the comprehensive device state can be unified into the form of a×b×c. Where a, b, and c are the lengths of the three dimensions respectively.

[0112] Step 305, input the comprehensive device state into the trained network group fault detection model to obtain the network group fault detection result of the network system.

[0113] The network group fault detection result of this embodiment at least includes whether a group fault occurs in the network system.

[0114] Exemplarily, referring to Figure 4 , through an optical network unit (ONU device), an optical line terminal (OLT device) and a network element device, the network group fault detection of the network system can be realized. Exemplarily, whether the ONU device changes from online to offline within a preset time period can be used as a key point of change, and the device state corresponding to this key point is represented by a feature value, and then a grayscale binary image is formed. The grayscale binary image is input into the trained network group fault detection model to obtain the network group fault detection result of the network system.

[0115] Among them, a data set can be constructed according to historical operation and maintenance group fault data. The data set can include network device state samples of ONU devices, OLT devices and network element devices. Referring to the processing method of the model recognition process, grayscale binary image samples are obtained according to the network device state samples of ONU devices, OLT devices and network element devices, and a multi-classification model such as a CNN model is trained through the binary image samples to obtain a trained network group fault detection model.

[0116] Exemplarily, the schematic diagram of the comprehensive device state in a three-dimensional data format is as Figure 5 shown, referring to Figure 5, the eigenvalue of the second device state of ONU devices under the same OLT device constitutes one-dimensional grayscale binary data corresponding to the OLT device; the one-dimensional grayscale binary data of OLT devices under the same network element device constitutes two-dimensional grayscale binary data corresponding to the network element device; the two-dimensional grayscale binary data of multiple network element devices in the network system constitutes three-dimensional grayscale binary data of the network system. Among them, the three-dimensional grayscale binary data is represented in the form of a picture. Among them, the grayscale binary value in the three-dimensional grayscale binary picture can be 0 or 1. The eigenvalue 0 indicates that the corresponding network device is offline, and the eigenvalue 1 indicates that the corresponding network device is online.

[0117] Exemplarily, the group fault detection result may further include the group fault type, and the group fault point can be quickly located according to the group fault type. For example, in Figure 6 the illustrated embodiment, it can be determined that the group fault point is that the OLT device has a group fault.

[0118] Exemplarily, the comprehensive device state includes the eigenvalue for characterizing the network device state of the network device; step 305 may include sub-steps F1 to F2:

[0119] Sub-step F1, according to the eigenvalue, construct a grayscale binary picture for characterizing the comprehensive device state.

[0120] Exemplarily, the eigenvalues are 0 and 1. The eigenvalue 0 indicates that the corresponding network device is in the offline state, and the eigenvalue 1 indicates that the corresponding network device is in the online state. Use the eigenvalue equal to 0 or 1 as the grayscale binary value to construct the grayscale binary picture.

[0121] Sub-step F2, input the grayscale binary picture into the trained network group fault detection model to obtain the network group fault detection result of the network system.

[0122] Among them, the network group fault detection model is obtained by training with the grayscale binary picture sample corresponding to the comprehensive device state sample, and the grayscale binary picture sample corresponding to the comprehensive device state sample is obtained according to the eigenvalue sample of the device state sample for characterizing the network device.

[0123] In the related art, it is possible to obtain home broadband fault alarm information, and then determine the data object points corresponding to the fault alarm information. Each data object point includes seven alarm key index components. Determine the rectangular cell to which the data object point belongs, use the rectangular cell to which the data object point belongs as a dense rectangular cell, and cluster the dense rectangular cell; then, according to the clustering result and the preset alarm association rule, locate the fault source; locate the home broadband fault according to the fault source. This processing method needs to collect seven alarm key index components of the data object points corresponding to the home user faults, the method is complex, and the processing efficiency is low. In this embodiment, it is possible to collect the device status of the three-level network devices (such as ONU devices, OLT devices, and network elements). This application does not need to collect additional information on the user side to determine whether a group fault has occurred. The method is simple. Because the data to be collected is less, it also has the advantages of high processing efficiency and low cost. In addition, in this embodiment, the online or offline status in the device status can be converted into a feature value, which is different from the input data of the related art method.

[0124] In the related art, it is also possible to use the preset alarm association rule for judgment and fault location. In this embodiment, a deep learning model trained with historical operation and maintenance data can be used as a group fault detection model. The judgment method of this embodiment is different from that of the related art, and there is no need to set a preset rule, which reduces the manual operation cost, and can avoid the problem of detection errors caused by human experience differences, and improves the accuracy of group fault detection. The method of the related art can only detect a single broadband fault, while this embodiment can achieve group fault detection.

[0125] In the related art, it is also possible to obtain the quality evaluation information of the broadband networks of multiple household user groups in a target cell and the traffic usage information of the multiple household user groups; then, for each household user group, the quality evaluation data and traffic usage information of the household user group are used as input features to be input into a pre-trained perception model to determine the quality score of the broadband network of the household user group; based on the quality scores of the broadband networks of the multiple household user groups, the overall network quality of the target cell is determined. This method requires collecting multiple broadband quality evaluation information in the same cell, such as broadband fault reports and complaint ratios, and traffic usage conditions, and using these as data information to evaluate the overall network quality of the cell. This method requires obtaining information through manual collection of fault reports and numerous evaluation parameters. In this embodiment, the network device automatically obtains the device status offline, and the method of this embodiment does not require consuming additional labor costs. The method in the related art depends on user fault reports and complaints for the input of evaluation data for the cell, and requires users to initiate it actively. However, the method of this embodiment does not require initiation from the user side, can actively detect the occurrence of group faults, and can perform network group fault detection and processing automatically, accurately, and quickly. In addition, the method in the related art can only evaluate the overall network quality of the cell and cannot perform network group fault checks, while this application can perform network checks quickly, accurately, and at low cost.

[0126] In the related art, in the case of a household broadband failure, it is necessary for the customer service to confirm and notify the network staff to conduct a fault investigation to confirm the cause of the detected fault and perform fault handling. Only when the number of network faults reported by users reaches a certain threshold will a group fault alarm be triggered. This processing method has problems such as difficulty in detecting group faults in a timely manner, long group fault location time, and high labor costs for the location method. In this embodiment, by integrating the offline state changes of the entire network devices including the first network device, the second network device, and the third network device, the comprehensive device state of the network system is obtained. Based on the comprehensive device state and the pre-trained network group fault detection model, group faults can be quickly detected. Further, the group fault point can also be located. In addition, compared with the method in the related art for determining network group faults based on the fault number threshold and the number reported by users, this application does not require setting threshold rules, and the model trained based on historical operation and maintenance information automatically judges network group faults, with high processing efficiency, and can avoid the problem of low detection result accuracy caused by artificially setting thresholds. In addition, in this embodiment, information image processing is performed on three-layer network devices (including ONU devices, OLT devices, and network element devices), and the data aggregation method is concise. Based on the obtained grayscale binary image and the trained network group fault inspection model, network group faults can be identified with high efficiency.

[0127] According to the above analysis, compared with the related art, the method of this embodiment can automatically, quickly, accurately and at low cost detect network group faults in the network system according to the first device state of the first network device, the second device state of the second network device, the third device state of the third network device, and the trained network group fault detection network model, at least solving the problems in the related art that network group faults cannot be detected, or the detection cost is high and the efficiency is low.

[0128] Referring to Figure 7 , the method for training a network group fault detection model provided by an embodiment of the present application is used to train the network group fault model training model in the network group fault identification method of the above-mentioned embodiment. The method may include the following steps:

[0129] Step 401, obtain the first device state sample of the first network device in the network system, and the second device state sample of the second network device managed by the first network device.

[0130] The first device state sample includes that the first network device is in an online or offline state, and the second device state sample includes that the second network device is in an online or offline state.

[0131] Step 402, obtain the comprehensive device state sample of the network system according to the first device state sample and the second device state sample.

[0132] For the method of this step, reference may be made to the description of step 102 above, and details are not described here again.

[0133] Exemplarily, step 402 includes sub-steps G1 to G3:

[0134] Sub-step G1, obtain the third device state sample of the third network device in the network system.

[0135] For the method of this step, reference may be made to the description of step 302 above, and details are not described here again.

[0136] Sub-step G2, obtain the second comprehensive device state sample corresponding to the third network device according to the third device state sample of the third network device, the first device state sample of the first network device managed by the third network device, and the second device state sample of the second network device managed by the first network device.

[0137] For the method of this step, reference may be made to the description of step 303 above, and details are not described here again.

[0138] Sub-step G3, obtain the comprehensive device state sample of the network system according to the second comprehensive device state samples respectively corresponding to each third network device.

[0139] The method of this step can refer to the description of step 304 above and will not be elaborated here.

[0140] Step 403: Train the network group fault detection model with the comprehensive device status samples to obtain the trained network group fault detection model.

[0141] Exemplarily, step 403 may include sub-steps G1 to G3:

[0142] Sub-step H1: Construct a grayscale binary map sample for characterizing the comprehensive device status samples according to the eigenvalue samples.

[0143] The method of this step can refer to the description of sub-step F1 above and will not be elaborated here.

[0144] Sub-step H2: Train the network group fault detection model with the grayscale binary map sample to obtain the trained network group fault detection model.

[0145] The method of this step can refer to the description of sub-step F2 above and will not be elaborated here.

[0146] Exemplarily, the comprehensive device status samples are in the format of a three-dimensional array. The three-dimensional array of the entire network when historical group faults occur and the three-dimensional array of the entire network when no group faults occur can be used as two classification data samples of the training set. Based on these data samples, a grayscale binary map sample is obtained, and the network group fault detection model is trained through the grayscale binary map sample.

[0147] Among them, the grayscale binary map sample is labeled. Specifically, according to the data of historical group faults, the fault location and the pixel points corresponding to the faults are marked in the grayscale binary map. Among them, one pixel point corresponds to a second network device. For example, the second network device (ONU device) is a pixel point, the first network device (OLT device) is a row, and the third network device (network element device) is a plane; thus, a three-dimensional grayscale binary map sample is formed.

[0148] In this embodiment, the second network devices (ONU devices) with an offline duration exceeding the preset duration are ignored. Specifically, when obtaining the second device status samples, the second device status of this second network device (ONU device) is not recorded in the second device status samples. Because the ONU devices with an offline duration exceeding the preset duration are ignored, the shapes of the training sets may not be unified. For this situation, the ROIPooling algorithm is used to achieve the integration of irregular data so that it can be input into the network fault detection model for model training.

[0149] Exemplarily, the sample data of a three-dimensional array is used as the input of the network fault detection model. The three-dimensional array is labeled with information on whether a group fault occurs and the fault points. The preprocessed and labeled dataset is input into the network fault detection model (for example, it can be a CNN model), and a trained image detection model is obtained through training.

[0150] In this embodiment, the first device state sample of the first network device in the network system is obtained, and the second device state sample of the second network device managed by the first network device is obtained. According to the first device state sample and the second device state sample, the comprehensive device state sample of the network system is obtained, and the network group fault detection model is trained through the comprehensive device state. A trained network group fault detection model is obtained. Based on this trained network group fault detection model, network group fault detection of the network system can be accurately and quickly performed.

[0151] Referring to Figure 8 , the network group fault detection method provided by this application may include the following steps:

[0152] Step S1, the OLT device collects the second device state of the ONU devices under it within a preset time period.

[0153] The second device state includes one of the following: online converted to offline, continuously online.

[0154] Exemplarily, the OLT device periodically detects the ONU devices in the offline state. Within the set time, if an ONU device changes from the online state to the offline state, the ONU device is represented by 0 in the binary value. If the ONU device is normally online, it is represented by 1. If the ONU device is offline for more than the set time, the ONU device is ignored.

[0155] Step S2, the network element device collects the first device state of the OLT devices under it within a preset time period, and the second device state of the ONU devices collected by the OLT devices.

[0156] Exemplarily, the OLT device collects the binary values of the online / offline changes of the ONU devices under it, combines the binary values of the online / offline changes to obtain a one-dimensional array, and reports it to the network element device. The network element device periodically detects the online / offline state of the OLT devices, and combines the one-dimensional arrays of multiple OLT devices into a two-dimensional array. Among them, the binary value of the online / offline change is a characteristic value corresponding to the device state of the network device. For example, if the network device is online, the corresponding characteristic value is 1, and if the network device is offline, the corresponding characteristic value is 0.

[0157] It should be noted that if the network element detects that the OLT device is offline, all the grayscale binary values of the row where the OLT device is located are marked as 0. Among them, when the OLT device is offline, the length of the corresponding one-dimensional array is equal to the number of ONU devices that the OLT device can support for hanging.

[0158] Step S3, detect the third device status of all network elements, as well as the first device status of the OLT devices and the second device status of the ONU devices collected by the network elements, and construct a three-dimensional array image based on these three types of device statuses.

[0159] Exemplarily, the three-dimensional array is represented in the format of a grayscale binary image; the three-dimensional array includes two-dimensional arrays of multiple network elements, and the two-dimensional array of a network element is composed of all the one-dimensional arrays respectively corresponding to the OLT devices hung by the network element. The three-dimensional array in the format of a grayscale binary image can accurately and concisely represent the device status of the entire network system, which is equivalent to the whole-network view of the entire network system.

[0160] Step S4, input the three-dimensional array image into the trained group fault detection model to obtain the group fault detection result and obtain the fault occurrence point.

[0161] Specifically, input the three-dimensional array into the group fault detection model trained with historical operation and maintenance data. The group fault detection model outputs the detection result of whether a group fault occurs in the network system, and infers the fault occurrence point based on the detection result.

[0162] Exemplarily, input the three-dimensional array representing the whole-network device view into the network group fault detection model (such as, a CNN model) that has been trained with historical operation and maintenance data, so as to quickly determine whether a group fault occurs in the network system, infer the possible cause of the fault, and quickly locate the fault point according to the cause.

[0163] This application provides a network group fault detection system, which includes an optical network unit (), an optical line terminal, and network element devices.

[0164] Figure 9 It is the structural diagram of a network group fault detection device provided by an embodiment of the present invention. The device 50 may include:

[0165] The first acquisition module 501 is used to acquire the first device status of the first network device in the network system and the second device status of the second network device managed by the first network device; the device status includes: the network device corresponding to the device status is in an online or offline state.

[0166] The second acquisition module 502 is used to acquire the comprehensive device status of the network system according to the first device status and the second device status.

[0167] A third acquisition module 503, configured to input the comprehensive device status into the trained network group fault detection model to obtain the network group fault detection result of the network system; the network group fault detection model is a model trained through comprehensive device status samples, and the comprehensive device status samples are samples obtained according to the first device status samples of the first network device and the second device status samples of the second network device.

[0168] Optionally, there is at least one first network device; the second acquisition module 502 includes:

[0169] A first acquisition sub-module, configured to obtain a first comprehensive device status corresponding to the first network device according to the first device status of the first network device and the second device status of the second network device managed by the first network device.

[0170] A second acquisition sub-module, configured to obtain the comprehensive device status of the network system according to the first comprehensive device status corresponding to each first network device.

[0171] Optionally, the first acquisition sub-module includes:

[0172] A first acquisition unit, configured to, if the first device status is that the first network device is in an online state, obtain the first comprehensive device status corresponding to the first network device according to the second device status of the second network device managed by the first network device.

[0173] Optionally, the first acquisition sub-module includes:

[0174] A second acquisition unit, configured to, if the first device status is that the first network device is in an offline state, mark the second device status of the second network device managed by the first network device as an offline state to obtain a status marking result of the second network device managed by the first network device.

[0175] A third acquisition unit, configured to obtain the first comprehensive device status corresponding to the first network device according to the status marking result of the second network device managed by the first network device.

[0176] Optionally, there are multiple second network devices managed by the first network device; the first acquisition sub-module may include:

[0177] A fourth acquisition unit, configured to obtain a target second network device that has been continuously in an offline state within a preset time period from the multiple second network devices according to the second device status.

[0178] A fifth acquisition unit, configured to obtain the first comprehensive device status corresponding to the first network device according to the first device status of the first network device and the second device status of other second network devices except the target second network device.

[0179] Optionally, the second acquisition module 502 includes:

[0180] A third acquisition sub-module, configured to acquire the third device status of a third network device in the network system; the third device status includes that the third network device is in an online or offline state;

[0181] A fourth acquisition sub-module, configured to obtain a second comprehensive device status corresponding to the third network device according to the third device status of the third network device, the first device status of the first network device managed by the third network device, and the second device status of the second network device managed by the first network device;

[0182] A fifth acquisition sub-module, configured to obtain the comprehensive device status of the network system according to the second comprehensive device status corresponding to each third network device respectively.

[0183] Optionally, the fourth acquisition sub-module includes:

[0184] A sixth acquisition unit, configured to, if the third device status is that the third network device is online, obtain a first comprehensive device status corresponding to the first network device according to the first device status of the first network device managed by the third network device and the second device status of the second network device managed by the first network device;

[0185] A seventh acquisition unit, configured to integrate the first comprehensive device statuses corresponding to each first network device managed by the third network device respectively to obtain a second comprehensive device status corresponding to the third network device.

[0186] Optionally, the data format of the first comprehensive device status is a one-dimensional data format, and the data format of the second comprehensive device status is a two-dimensional data format; the fifth acquisition sub-module includes:

[0187] An eighth acquisition unit, configured to integrate the second comprehensive device statuses corresponding to each third network device respectively to obtain the comprehensive device status of the network system in a three-dimensional data format.

[0188] Optionally, the data format of the second comprehensive device status is a two-dimensional data format, and the data format of the comprehensive device status is a three-dimensional data format; the fifth acquisition sub-module includes:

[0189] A ninth acquisition unit, configured to perform regional normalization processing on the second comprehensive device status corresponding to each third network device respectively based on a preset regional normalization algorithm to obtain the processed second comprehensive device status;

[0190] A tenth acquisition unit, configured to obtain the comprehensive device status of the network system according to the processed second comprehensive device status; wherein, the number of elements of each processed second comprehensive device status in the same dimension is correspondingly equal.

[0191] Optionally, the comprehensive device status includes eigenvalue for characterizing the network device status; the third acquisition module 503 includes:

[0192] The sixth acquisition sub-module is configured to construct a grayscale binary image for characterizing the comprehensive device status according to the eigenvalue;

[0193] The seventh acquisition sub-module is configured to input the grayscale binary image into the trained network group fault detection model to obtain the network group fault detection result of the network system; the network group fault detection model is obtained by training with the grayscale binary image sample corresponding to the comprehensive device status sample, and the grayscale binary image sample corresponding to the comprehensive device status sample is obtained according to the eigenvalue sample of the device status sample for characterizing the network device.

[0194] Refer to Figure 10 , this embodiment further provides a network group fault detection model training device for training the network group fault model training model in the network group fault recognition method of any of the above embodiments. The device 60 includes:

[0195] The fourth acquisition module 601 is configured to acquire the first device status sample of the first network device in the network system and the second device status sample of the second network device managed by the first network device; the first device status sample includes that the first network device is in an online or offline state, and the second device status sample includes that the second network device is in an online or offline state;

[0196] The fifth acquisition module 602 is configured to acquire the comprehensive device status sample of the network system according to the first device status sample and the second device status sample;

[0197] The sixth acquisition module 603 is configured to train the network group fault detection model with the comprehensive device status sample to obtain the trained network group fault detection model.

[0198] Optionally, the fifth acquisition module 602 includes:

[0199] The eighth acquisition sub-module is configured to acquire the third device status sample of the third network device in the network system;

[0200] The ninth acquisition sub-module is configured to obtain the second comprehensive device status sample corresponding to the third network device according to the third device status sample of the third network device, the first device status sample of the first network device managed by the third network device, and the second device status sample of the second network device managed by the first network device;

[0201] The tenth acquisition sub-module is configured to acquire the comprehensive device status sample of the network system according to the second comprehensive device status samples respectively corresponding to each third network device.

[0202] Optionally, the comprehensive device status sample includes eigenvalue samples for characterizing the network device status samples of network devices; the sixth acquisition module 603 includes:

[0203] The eleventh acquisition sub-module is used to construct a grayscale binary map sample for characterizing the comprehensive device status sample according to the eigenvalue samples;

[0204] The twelfth acquisition sub-module is used to train the network group fault detection model through the grayscale binary map sample to obtain a trained network group fault detection model.

[0205] In this embodiment, the first device status sample of the first network device in the network system and the second device status sample of the second network device managed by the first network device are obtained. According to the first device status sample and the second device status sample, the comprehensive device status sample of the network system is obtained, and the network group fault detection model is trained through the comprehensive device status to obtain a trained network group fault detection model. Based on this trained network group fault detection model, the network system can be accurately and quickly detected for network group faults.

[0206] The present invention also provides an electronic device. Refer to Figure 11 , including: a processor 701, a memory 702, and a computer program 7021 stored on the memory and executable on the processor. When the processor executes the program, it implements the network group fault detection method or the network group fault detection model training method of the foregoing embodiments.

[0207] The present invention also provides a readable storage medium. When the instructions in the storage medium are executed by the processor of the electronic device, the electronic device can execute the network group fault detection method or the network group fault detection model training method of the foregoing embodiments.

[0208] For the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple. For the relevant parts, refer to the partial description of the method embodiment.

[0209] It should be noted that all kinds of data obtained in the embodiments of the present invention are obtained under the authorization of the acquisition / data holder.

[0210] The algorithms and displays provided herein are not inherently related to any particular computer, virtual system, or other device. Various general-purpose systems can also be used in conjunction with the teachings herein. The structure required to construct such a system will be apparent from the above description. In addition, the present invention is not directed to any particular programming language. It should be understood that the content of the present invention described herein can be implemented using various programming languages, and the description of the specific language above is to disclose the best mode of the present invention.

[0211] In the description provided herein, numerous specific details are set forth. It will be understood, however, that embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures and techniques have not been shown in detail in order not to obscure an understanding of this description.

[0212] Similarly, it should be understood that in order to streamline the present invention and assist in understanding one or more of the various inventive aspects, in the foregoing description of the exemplary embodiments of the present invention, the various features of the present invention are sometimes grouped together in a single embodiment, figure, or description thereof. However, the disclosed method should not be construed as reflecting an intention that the claimed invention requires more features than are expressly recited in each claim. Rather, as reflected in the following claims, the inventive aspects lie in less than all the features of the single foregoing disclosed embodiment. Thus, the claims following the detailed description are hereby expressly incorporated into the detailed description, with each claim standing on its own as a separate embodiment of the present invention.

[0213] Those skilled in the art will appreciate that the modules in the devices in the embodiments can be adaptively changed and disposed in one or more devices different from the embodiments. The modules or units or components in the embodiments can be combined into one module or unit or component, and in addition, can be divided into multiple sub-modules or sub-units or sub-components. Except that at least some of such features and / or processes or units are mutually exclusive, any combination can be used to combine all the features disclosed in this specification (including the accompanying claims, abstract and drawings) and all the processes or units of any method or device so disclosed. Unless otherwise expressly stated, each feature disclosed in this specification (including the accompanying claims, abstract and drawings) can be replaced by an alternative feature providing the same, equivalent or similar purpose.

[0214] The various component embodiments of the present invention may be implemented in hardware, or in software modules running on one or more processors, or in a combination thereof. Those skilled in the art should understand that in practice, a microprocessor or a digital signal processor (DSP) can be used to implement some or all of the functions of some or all of the components of the sorting device according to the present invention. The present invention can also be implemented as a device or apparatus program for performing part or all of the methods described herein. Such a program for implementing the present invention can be stored on a computer-readable medium, or can be in the form of one or more signals. Such signals can be downloaded from an Internet website, or provided on a carrier signal, or in any other form.

[0215] It should be noted that the above embodiments are illustrative of the present invention rather than restrictive thereof, and those skilled in the art can design alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses shall not be construed as limiting the claim. The word "comprising" does not exclude the presence of elements or steps not listed in a claim. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The present invention can be implemented by means of hardware including several different elements and by means of a suitably programmed computer. In a unit claim listing several devices, several of these devices may be embodied by the same item of hardware. The use of the words first, second, and third, etc. does not denote any order. These words may be interpreted as names.

[0216] The users involved in the present invention (including but not limited to the users' devices, the users themselves, etc.), relevant data, etc. are all authorized by the users or authorized by all parties.

[0217] Those skilled in the art can clearly understand that for the sake of convenience and brevity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0218] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

[0219] The above are only the specific implementation manners of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or replacements within the technical scope disclosed by the present invention and should be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A network group fault detection method, characterized in that Including: Obtaining a first device state of a first network device in a network system and a second device state of a second network device managed by the first network device; The first device state includes that the first network device is in an online or offline state, and the second device state includes that the second network device is in an online or offline state; Obtaining an overall device state of the network system according to the first device state and the second device state; Inputting the overall device state into a trained network group fault detection model to obtain a network group fault detection result of the network system; The network group fault detection model is a model trained through overall device state samples, and the overall device state samples are samples obtained according to a first device state sample of the first network device and a second device state sample of the second network device.

2. The method according to claim 1, characterized in that, There is at least one first network device; the obtaining of the overall device state of the network system according to the first device state and the second device state includes: Obtaining a first overall device state corresponding to the first network device according to the first device state of the first network device and the second device state of the second network device managed by the first network device; Obtaining the overall device state of the network system according to the first overall device state corresponding to each first network device respectively.

3. The method according to claim 2, wherein The obtaining of the first overall device state corresponding to the first network device according to the first device state of the first network device and the second device state of the second network device managed by the first network device includes: If the first device state is that the first network device is in an offline state, then marking the second device state of the second network device managed by the first network device as an offline state to obtain a state marking result of the second network device managed by the first network device; Obtaining the first overall device state corresponding to the first network device according to the state marking result of the second network device managed by the first network device.

4. The method according to claim 1, characterized in that, There are multiple second network devices managed by the first network device; the obtaining of the first overall device state corresponding to the first network device according to the first device state of the first network device and the second device state of the second network device managed by the first network device includes: According to the second device state, obtaining a target second network device that has been continuously in an offline state within a preset time period from the multiple second network devices; Obtaining the first overall device state corresponding to the first network device according to the first device state of the first network device and the second device state of other second network devices except the target second network device.

5. The method according to any one of claims 1 to 4, characterized in that The obtaining of the overall device state of the network system according to the first device state and the second device state includes: Obtaining a third device state of a third network device in the network system; the third device state includes that the third network device is in an online or offline state; Obtain a second comprehensive device state corresponding to the third network device according to the third device state of the third network device, the first device state of the first network device managed by the third network device, and the second device state of the second network device managed by the first network device; Obtain the comprehensive device state of the network system according to the second comprehensive device state corresponding to each of the third network devices respectively.

6. The method according to claim 5, wherein The obtaining a second comprehensive device state corresponding to the third network device according to the third device state of the third network device, the first device state of the first network device managed by the third network device, and the second device state of the second network device managed by the first network device includes: If the third device state is that the third network device is online, obtain a first comprehensive device state corresponding to the first network device according to the first device state of the first network device managed by the third network device and the second device state of the second network device managed by the first network device; Integrate the first comprehensive device states corresponding to each of the first network devices managed by the third network device to obtain the second comprehensive device state corresponding to the third network device.

7. The method according to claim 6, characterized in that, The data format of the first comprehensive device state is a one-dimensional data format, and the data format of the second comprehensive device state is a two-dimensional data format; the obtaining the comprehensive device state of the network system according to the second comprehensive device state corresponding to each of the third network devices respectively includes: Integrate the second comprehensive device states corresponding to each of the third network devices respectively to obtain the comprehensive device state of the network system in a three-dimensional data format.

8. The method according to claim 5, wherein The data format of the second comprehensive device state is a two-dimensional data format, and the data format of the comprehensive device state is a three-dimensional data format; The obtaining the comprehensive device state of the network system according to the second comprehensive device state corresponding to each of the third network devices respectively includes: Based on a preset regional normalization algorithm, perform regional normalization processing on the second comprehensive device state corresponding to each of the third network devices respectively to obtain the processed second comprehensive device state; Obtain the comprehensive device state of the network system according to the processed second comprehensive device state; wherein, the number of elements of each of the processed second comprehensive device states in the same dimension is correspondingly equal.

9. The method according to claim 5, wherein The comprehensive device state includes characteristic values for characterizing the network device state of the network device; Input the comprehensive device state into the trained network group fault detection model to obtain the network group fault detection result of the network system, including: Construct a grayscale binary image for characterizing the comprehensive device state according to the characteristic values; Input the grayscale binary image into the trained network group fault detection model to obtain the network group fault detection result of the network system; the network group fault detection model is trained by the grayscale binary image samples corresponding to the comprehensive device state samples, and the grayscale binary image samples corresponding to the comprehensive device state samples are obtained according to the characteristic value samples of the device state samples for characterizing the network device.

10. A method for training a network group obstacle model, characterized in that, A training model for training a network group fault identification model in the network group fault identification method according to any one of claims 1-9, comprising: Obtaining a first device status sample of a first network device in a network system and a second device status sample of a second network device managed by the first network device; the first device status sample includes whether the first network device is in an online or offline state, and the second device status sample includes whether the second network device is in an online or offline state; Obtaining a comprehensive device status sample of the network system according to the first device status sample and the second device status sample; Training a network group fault detection model through the comprehensive device status sample to obtain a trained network group fault detection model.

11. The method according to claim 10, wherein The obtaining a comprehensive device status sample of the network system according to the first device status sample and the second device status sample includes: Obtaining a third device status sample of a third network device in the network system; Obtaining a second comprehensive device status sample corresponding to the third network device according to the third device status sample of the third network device, the first device status sample of the first network device managed by the third network device, and the second device status sample of the second network device managed by the first network device; Obtaining a comprehensive device status sample of the network system according to the second comprehensive device status samples respectively corresponding to each third network device.

12. The method according to claim 11, wherein The comprehensive device status sample includes a feature value sample for characterizing the network device status sample of the network device; The training a network group fault detection model through the comprehensive device status sample to obtain a trained network group fault detection model includes: Constructing a grayscale binary map sample for characterizing the comprehensive device status sample according to the feature value sample; Training a network group fault detection model through the grayscale binary map sample to obtain a trained network group fault detection model.

13. A network group fault detection device, characterized in that, The device includes: A first obtaining module, configured to obtain a first device status of a first network device in a network system and a second device status of a second network device managed by the first network device; the first device status includes whether the first network device is in an online or offline state, and the second device status includes whether the second network device is in an online or offline state; A second obtaining module, configured to obtain a comprehensive device status of the network system according to the first device status and the second device status; A third obtaining module, configured to input the comprehensive device status into a trained network group fault detection model to obtain a network group fault detection result of the network system; the network group fault detection model is a model trained through a comprehensive device status sample, and the comprehensive device status sample is a sample obtained according to the first device status sample of the first network device and the second device status sample of the second network device.

14. A network group obstacle model training device, characterized in that, A training model for training a network group fault identification model in the network group fault identification method according to any one of claims 1-9, comprising: The fourth acquisition module is configured to acquire a first device status sample of a first network device in a network system and a second device status sample of a second network device managed by the first network device; the first device status sample includes whether the first network device is in an online or offline state, and the second device status sample includes whether the second network device is in an online or offline state; The fifth acquisition module is configured to acquire a comprehensive device status sample of the network system according to the first device status sample and the second device status sample; The sixth acquisition module is configured to train a network fault detection model with the comprehensive device status sample to obtain a trained network fault detection model.

15. An electronic device, characterized in that, Comprising: A processor, a memory, and a computer program stored on the memory and executable on the processor, wherein when the processor executes the program, the method according to any one of claims 1-12 is implemented.

16. A readable storage medium, characterized in that, When the instructions in the storage medium are executed by the processor of the electronic device, the electronic device is enabled to execute the method according to any one of claims 1-12.

Images