WO2019137052A1 - Method and device for network operation and maintenance - Google Patents

Abstract

The present application provides a method and device for network operation and maintenance, and belongs to the technical field of networks. The method comprises: a server acquiring network data of m types of services, where m ≥ 2; then, the server determining n pieces of first fault information according to the network data of the m types of services, each piece of first fault information being used to indicate that a corresponding service has encountered a network fault, where 1 ≤ n ≤ m; subsequently, the server dividing part or all of the n pieces of first fault information into k sets of fault information, a superordinate fault of a network fault indicated by the first fault information in each set of fault information being the same, where 1 ≤ k ≤ n; the server outputting the k sets of fault information and the k superordinate faults, the k superordinate faults corresponding one-to-one to the k sets of fault information. Further, the server can also predict potential faults in a network. The present application resolves the issue in which a plurality of services cannot be comprehensively processed in network operation and maintenance modes in the art, realizes comprehensive processing a plurality of services, improves the accuracy of fault prediction, and increases the efficiency of fault processing, and is applicable to network operation and maintenance.

Description

Network operation and maintenance method and device

The present application claims priority to Chinese Patent Application Serial No. PCT Application No. No. No. No. No. No. No. No. No. No. No. No. No. No. No. No. No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No

Technical field

The present application relates to the field of network technologies, and in particular, to a method and apparatus for network operation and maintenance.

Background technique

In the data service era, the user experience is the core of the service. The stable and reliable network and the good user experience can help operators to rapidly develop services. The network operation and maintenance is used to ensure the safe operation of the network and services, and how to carry out network operation and maintenance. User experience is a very important issue.

There is a network operation and maintenance method in the related art. This network operation and maintenance method first uses an unsupervised learning model to perform abnormality detection on the network data of the service, and then presents the detection result to the staff, and the staff performs the accuracy of the detection result. Judging, the correct detection result is taken as a training sample, and then the training sample is trained to obtain a supervised learning model, and then the supervised learning model is used to perform abnormality detection on the network data of the service.

However, the above network operation and maintenance mode can only process one type of service, and cannot comprehensively process multiple services. With the rapid development of network technology, network services are becoming more and more abundant, and it is urgent to integrate multiple services. The network operation and maintenance method handled.

Summary of the invention

The embodiment of the invention provides a method and a device for network operation and maintenance, which can solve the problem that the network operation and maintenance mode cannot comprehensively process a plurality of services in the related art, and the technical solution is as follows:

The first aspect provides a method for network operation and maintenance, the method includes: the server first acquires network data of the m types of services, m≥2, and then determines n first fault information according to the network data of the m types of services, and each The first fault information is used to indicate that the corresponding service has a network fault, 1≤n≤m. Then, the server divides part or all of the n first fault information into k sets of fault information, and the upper fault of the network fault indicated by the first fault information in each set of fault information is the same, and is indicated by any first fault information. The superior fault of the network fault is a fault causing the network fault indicated by any of the first fault information, 1 ≤ k ≤ n. After that, the server outputs k sets of fault information and k upper level faults, and k upper level faults correspond one-to-one with the k group fault information.

Optionally, the m services may include a predictive service, an alarm compression service, and an abnormality detection service.

Optionally, the server can display k sets of fault information and k superior faults.

In the embodiment of the present invention, the server can determine the first fault information and the superior fault according to the network data of the multiple services, so that the staff can perform fault processing. Further, the worker can also obtain potential faults in the network according to the superior fault and the first fault information output by the server, and process the potential faults.

Optionally, after outputting the k-group fault information and the k-level faults, the method may further include: the server acquiring, according to the k upper-level faults and the first fault information corresponding to each of the upper-level faults, the association related to each superior fault. The network data is further predicted according to the associated network data, and the second fault information is different from the first fault information. Thereafter, the server outputs k superior faults, k sets of fault information, and all predicted second fault information.

In the embodiment of the present invention, the network fault indicated by the second fault information related to the superior fault refers to a network fault that can be caused by the superior fault.

Optionally, the server may display k upper faults, k sets of fault information, and all predicted second fault information.

Because the correlation between the network data is strong, in the embodiment of the present invention, when the server obtains the superior fault and the first fault information, the server may predict the remaining network faults that may be caused by the superior fault according to the superior fault and the first fault information. The superior diffusion labeling selection method enables the staff to timely deal with faults and potential faults in the network according to the superior fault, the first fault information and the second fault information, thereby improving the stability of the network and ensuring the normal operation of the network.

Optionally, after the k-group fault information and the k-level faults are output, the method may further include: the server receiving the first labeling instruction, where the first labeling instruction is used to indicate that the k-group fault information predicts the correct first fault information. Predict the correct superior fault with k superior faults. Next, the server acquires a first sample set based on the first annotation instruction, the first sample set including information indicated by the first annotation instruction. Then, the server acquires the associated network data related to each superior fault in the first sample set according to the first sample set, and then predicts the second fault information related to each superior fault according to the associated network data, and the second fault information and the second fault information A fault message is different. The server then outputs the first sample set and all of the predicted second failure information.

Optionally, the server may send a prompt message for prompting the staff to use the first annotation symbol to mark the server to predict the correct superior failure and predict the correct first failure information, and use the second annotation symbol to mark the superior of the server prediction error. The first fault message for faults and predicted errors.

Optionally, the server may display the first sample set and all the second fault information predicted.

In the embodiment of the present invention, since the server obtains the second fault information according to the first fault information that is correctly predicted in the k group fault information and the correct fault fault in the k upper faults, the accuracy of the second fault information is higher. .

In the embodiment of the present invention, the server can predict the network fault that may be caused by the correct superior fault according to the staff's labeling instruction, so that the staff can predict the correct first fault information, predict the correct superior fault, and predict all. The second fault information timely processes faults and potential faults in the network. Moreover, since the accuracy of the second fault information is high, the processing efficiency of the fault is also improved.

Optionally, the network data of the m types of services corresponds to the m operation and maintenance models, and the m operation and maintenance models are different from each other. Each operation and maintenance model is used to predict the network data of the corresponding service, and output fault information or non- The fault information, after outputting the first sample set and the predicted second fault information, the method may further include: the server determining all the predicted second fault information as the sample set to be labeled, and then receiving the second labeling instruction, and second The labeling instruction is used to indicate the second fault information that is correctly predicted within the sample set to be labeled. The server then acquires a second sample set based on the second annotation instruction, the second sample set includes information indicated by the second annotation instruction, and then the server determines the first sample set and the second sample set as the target sample set, and then the server The evaluation index of the first operation and maintenance model is determined according to the target sample set, and the first operation and maintenance model is any operation and maintenance model of the m operation and maintenance models. When the evaluation index of the first operation and maintenance model does not belong to the specified evaluation index range, the server uses the target sample set to update the first operation and maintenance model.

Optionally, the evaluation index of the first operation and maintenance model may be the accuracy, the precision, or the false discovery rate of the first operation and maintenance model. The specified evaluation index range can be determined according to the determined evaluation index of the first operation and maintenance model.

In the embodiment of the present invention, the server may obtain the second fault information that is predicted correctly according to the labeling instruction of the staff, and further, according to the predicted first fault information, predict the correct superior fault, and predict the correct second fault information to the evaluation index. The operation and maintenance model that does not meet the business requirements is updated to improve the accuracy of fault prediction, thereby improving the processing efficiency of the fault.

Optionally, the network data of the m types of services corresponds to the m operation and maintenance models, and the m operation and maintenance models are different from each other. Each operation and maintenance model is used to predict the network data of the corresponding service, and output fault information or non- The fault information, the server predicting the second fault information related to each of the upper faults according to the associated network data, may include: the server first inputs the associated network data to the associated operation and maintenance model to obtain the information output by the associated operation and maintenance model, and the associated operation and maintenance The model is an operation and maintenance model corresponding to the associated network data in the m operation and maintenance models. When the information output by the associated operation and maintenance model is fault information, the information output by the associated operation and maintenance model is determined to be related to each superior failure. The second fault information.

Optionally, the network data of the m types of services is in one-to-one correspondence with the m operation and maintenance models, and the m operation and maintenance models are different from each other. The server determines the n first fault information according to the network data of the m types of services, and may include: the server inputs the network data of the corresponding service to the m operation and maintenance models, to obtain the information output by the m operation and maintenance models, and the output of each operation and maintenance model. The information is fault information or non-fault information, and the information output by the m operation and maintenance models includes n fault information. Thereafter, the server determines the n pieces of failure information as n pieces of first failure information.

In a second aspect, a device for network operation and maintenance is provided. The device for network operation and maintenance includes at least one module, and at least one module is used to implement the network operation and maintenance method described in the first aspect.

In a third aspect, an apparatus for network operation and maintenance is provided, the apparatus comprising a processor, a memory, a network interface, and a bus. Among them, the bus is used to connect the processor, memory and network interface. The network interface is used to implement a communication connection between the server and the communication device. The processor is configured to execute a program stored in a memory to implement the method of network operation and maintenance described in the first aspect.

In a fourth aspect, a computer readable storage medium is provided, the computer readable storage medium storing instructions for causing a computer to perform the network operation and maintenance described in the first aspect when the computer readable storage medium is run on a computer Methods.

In a fifth aspect, a computer program product comprising instructions for causing a computer to perform the method of network operation and maintenance described in the first aspect when the computer program product is run on a computer is provided.

The technical effects obtained by the above second to fifth aspects are similar to those obtained by the corresponding technical means in the first aspect, and are not described herein again.

The beneficial effects brought by the technical solutions provided by the embodiments of the present invention are:

The server can determine n (1 ≤ n ≤ m) first fault information according to network data of m (m ≥ 2) services, and then divide part or all of the n first fault information into k (1 ≤ k ≤ n) The group failure information is the same as the upper fault of the network fault indicated by the first fault information in each set of fault information. After that, the server outputs k sets of fault information and k upper faults, k upper faults and k sets of fault information. Correspondingly, the staff can process faults and potential faults in the network in time. According to the embodiment of the present invention, various services can be comprehensively processed, and the operation and maintenance model whose evaluation index does not meet the service requirements can be automatically updated and improved. The accuracy of fault prediction improves the processing efficiency of faults.

DRAWINGS

1 is a schematic diagram of an implementation environment according to an embodiment of the present invention;

2 is a flowchart of a method for a network operation and maintenance method provided by an implementation of the present invention;

3 is a flowchart of a method for determining first fault information provided by an implementation of the present invention;

4 is a schematic diagram of a first fault information and a superior fault provided by the implementation of the present invention;

FIG. 5 is a flowchart of a method for predicting second fault information according to an embodiment of the present invention; FIG.

6 is a flowchart of a method for another network operation and maintenance method provided by the implementation of the present invention;

7 is a schematic diagram showing the marking of the upper fault and the first fault information diagram shown in FIG. 4 according to the implementation of the present invention;

8 is a schematic diagram showing the marking of the upper fault and the first fault information diagram shown in FIG. 4 according to the implementation of the present invention;

FIG. 9 is a schematic structural diagram of an apparatus for network operation and maintenance according to an embodiment of the present invention;

FIG. 10 is a schematic structural diagram of another apparatus for network operation and maintenance provided by an embodiment of the present invention;

11 is a schematic structural diagram of another apparatus for network operation and maintenance provided by an embodiment of the present invention;

FIG. 12 is a schematic structural diagram of an apparatus for network operation and maintenance according to an embodiment of the present invention.

Detailed ways

In order to make the objects, technical solutions and advantages of the present application more clear, the embodiments of the present application will be further described in detail below with reference to the accompanying drawings.

1 is a schematic diagram of an implementation environment according to an embodiment of the present invention. As shown in FIG. 1 , the implementation environment may include a server 001 and a communication device 002. For example, the communication device 002 may be a base station. The base station is configured to enable the terminal 10 in the cell to communicate, and the server 001 can acquire network data of multiple services from the base station. Server 001 can be a server, or a server cluster consisting of several servers, or a cloud computing service center.

In the embodiment of the present invention, the server 001 is configured to obtain network data of m (m≥2) services, and determine n (1≤n≤m) first fault information according to the network data of the multiple services, and then Part or all of the n first fault information is divided into k (1 ≤ k ≤ n) group fault information, and the upper fault of the network fault indicated by the first fault information in each set of fault information is the same, and then the k-group fault is output. The information and the k upper-level faults, the k upper-level faults and the k-group fault information are in one-to-one correspondence, thereby enabling the staff to perform fault processing. Further, in an achievable manner, in order to avoid potential failures affecting the network, the server may also predict potential faults according to the superior fault and the first fault information; in another achievable manner, in order to improve the accuracy of the fault prediction Sex, the server can also determine the correct superior fault and predict the correct first fault information according to the staff's labeling instructions, and then predict the potential fault based on predicting the correct superior fault and predicting the correct first fault information. The method for network operation and maintenance provided by the embodiment of the present invention is described below by taking the two implementations as an example.

In an implementation manner, the method for network operation and maintenance provided by the embodiment of the present invention is as shown in FIG. 2, and may include:

Step 201: The server acquires network data of m types of services, where m≥2.

Referring to FIG. 1, the server obtains network data of the service from the communication device. For example, the server may acquire network data of the service from the base station.

For example, the m services acquired by the server may include a predictive service, an alarm compression service, and an abnormality detection service. The predictive service may include a hardware failure prediction service, a performance prediction service, and a resource prediction service. The alarm compression service may include a single domain alarm compression service, an inter-area alarm compression service, and a root cause alarm analysis service. It can include key performance indicator (KPI) anomaly detection service and service degradation anomaly detection service. A brief description of each service is provided below.

The hardware failure prediction service is used to predict the hardware that is about to fail, and then replace or repair the hardware that is about to fail in time. For example, the hardware performance data and the hardware data collected by the sensor can be used for prediction, for example, prediction. The hardware can be a single board, a hard disk, or an optical module. Performance prediction services are used to predict network performance metrics such as bandwidth, throughput, and latency. The resource prediction service is used to predict network resources (such as the central processing unit (CPU) occupancy rate, etc.). The alarm compression service is used to compress a large amount of alarm data generated in the network to obtain important alarm data that affects the network. The single-domain alarm compression service in the alarm compression service is used to compress alarm data in the same product domain. For example, The network devices of the access layer can be regarded as communication devices of the same product domain. The inter-area alarm compression service is used to compress alarm data of different product domains. The root cause alarm analysis service is used to analyze the basic alarm data that affects the network. The anomaly detection service is used to monitor various indicators in the network and report abnormal information. The KPI anomaly detection service in the anomaly detection service is used to monitor KPIs (such as KPIs of packet loss rate and KPIs of call quality) in real time. The service degradation anomaly detection service is used to monitor key quality indicators (KQI) in real time. Among them, KPI is used to monitor the running status of the network, and KQI is used to measure the quality of the business.

For example, the network data of the hardware failure prediction service acquired by the server may include related performance indicators of the hardware and hardware data collected by the sensor, etc., and the network data of the obtained performance prediction service may include data such as network performance indicators, and the obtained resource prediction service The network data may include data such as network resources, and the network data of the obtained single-domain alarm compression service may include alarm data in the same product domain, and the acquired network data of the cross-domain alarm compression service may include alarm data of different product domains, and obtained. The network data of the KPI abnormality detecting service may include data such as KPI, and the acquired network data of the service degradation abnormality detecting service may include data such as KQI.

It should be noted that the period in which the server obtains the network data of each service may be determined according to the corresponding service requirement, for example, the period may be 20 minutes or 1 hour.

Step 202: The server determines, according to the network data of the m types of services, n first fault information, where each first fault information is used to indicate that a network fault occurs in the corresponding service, where 1≤n≤m.

Optionally, in the embodiment of the present invention, the network data of the m types of services are in one-to-one correspondence with the m operation and maintenance models, and the m operation and maintenance models are different from each other. Correspondingly, as shown in FIG. 3, the step 202 may include:

Step 2021: The server inputs network data of the corresponding service to the m operation and maintenance models to obtain information outputted by the m operation and maintenance models, and the information output by each operation and maintenance model is fault information or non-fault information, and m operation and maintenance model outputs. The information includes n fault information.

In the embodiment of the present invention, the server may use the operation and maintenance model to determine the first fault information according to the network data of the service, and assume that the types of services in the step 201 include the predictive service, the alarm compression service, and the abnormality detection service, and then The operation and maintenance model for determining the first failure information may include: a prediction class model, an alarm compression class model, and an anomaly detection class model. Assume that in step 201, the server obtains network data of eight types of services: hardware failure prediction service, performance prediction service, resource prediction service, single domain alarm compression service, cross-domain alarm compression service, and root cause. The alarm analysis service, the KPI anomaly detection service, and the service degradation anomaly detection service, the prediction class model may include a hardware failure prediction model, a performance prediction model, and a resource prediction model; the alarm compression model may include a single domain alarm compression model and an inter-domain alarm. The compression model and the root cause alarm analysis model; the anomaly detection class model may include a KPI anomaly detection model and a service degradation anomaly detection model, and the total number of operation and maintenance models is 8. The network data of the eight kinds of services corresponds to the eight operation and maintenance models one by one. The eight operation and maintenance models are different from each other.

The server inputs the network data of the corresponding service to the eight operation and maintenance models to obtain the information outputted by the eight operation and maintenance models. For example, the server inputs the network data of the hardware failure prediction service to the hardware failure prediction model, and obtains the output of the hardware failure prediction model. accident details. For another example, the server inputs the network data of the performance prediction service to the performance prediction model, and obtains the failure information output by the performance prediction model.

Step 2022: The server determines n pieces of fault information as n pieces of first fault information.

If the information output by each operation and maintenance model is fault information, the server can obtain m first fault information.

Step 203: The server divides part or all of the n first fault information into k group fault information, and the fault of the upper fault of the network fault indicated by the first fault information in each set of fault information is the same, and is indicated by any first fault information. The superior fault of the network fault is a fault causing the network fault indicated by any of the first fault information, 1 ≤ k ≤ n.

For example, when a certain first fault information is “cell 231 service degradation”, the upper fault of the network fault indicated by the first fault information may be a base station equipment fault. The cell managed by the base station includes a cell 231.

The network data and eight operation and maintenance models of the eight services in step 2021 are taken as an example. The server inputs the network data of the corresponding service to the eight operation and maintenance models, and assumes that the information output by the eight operation and maintenance models is fault information. In this way, the server got 8 first failure information. It is assumed that the server groups all of the eight first fault information, for example, the eight first fault information is divided into two sets of fault information, and the first group of fault information includes three first fault information, and the three first fault information. The superordinate fault of the indicated network fault is a fault of the base station equipment, and the second set of fault information includes five first fault information, and the fault of the upper fault of the network fault indicated by the five first fault information is another fault of the transport equipment.

FIG. 4 exemplarily shows a set of fault information and a schematic diagram of a superior fault corresponding to the set of fault information, the set of fault information including three first fault information: “cell 231 service degradation”, “Ethernet (ETH) The link connection is abnormal, and the CPU usage is high. The cell 231 service degradation is the network data that the server inputs the corresponding service to the service degradation abnormality detection model, and the service degradation abnormality detection model outputs the failure information. The ETH link connection abnormality is the network data that the server inputs the corresponding service to the KPI abnormality detection model, and the KPI abnormality detection model outputs the fault information. The "high CPU usage" is the network data that the server inputs to the resource prediction model, and the resource predicts the fault information output by the model. The superior fault of the network fault indicated by the three first fault information is a base station equipment fault.

Step 204: The server outputs k sets of fault information and k upper faults, and the k upper faults are in one-to-one correspondence with the k sets of fault information.

The server outputs k sets of fault information and k superior faults, so that the staff can perform fault processing according to the k sets of fault information and k superior faults. Further, the worker can also obtain potential faults in the network according to the superior fault and the first fault information output by the server, and process the potential faults.

Optionally, the server can display k sets of fault information and k superior faults. For example, the result of the 1 set of fault information and the corresponding superior fault displayed by the server may be as shown in FIG. 4 .

Step 205: The server acquires associated network data related to each superior fault according to the k upper faults and the first fault information corresponding to each upper fault.

Due to the strong correlation between network data, for example, a certain base station manages 3 cells, when the base station fails, the cell managed by the base station may be affected. Therefore, when the server obtains the superior fault and the first fault information, the server can further determine the potential fault in the network. In order to identify potential failures, the server may first obtain associated network data related to the superior failure.

It is assumed that, in step 202, the server determines eight first fault information according to the network data of the eight types of services. In step 203, the server divides the eight first fault information into two sets of fault information, and the first group of fault information. The first fault information includes: x1, x2, and x3. The upper fault of the network fault indicated by the three first fault information is A11; the second fault information includes five first fault information: y1, y2, and y3. , y4 and y5, the upper fault of the network fault indicated by the five first fault information is B11. Then the server obtains the associated network data related to A11 and the associated network data related to B11.

For example, the first fault information is: “cell 231 service degradation”, the upper fault of the network fault indicated by the first fault information is: the base station equipment fault, and the associated network data related to the upper fault obtained by the server may be: a cell. 232 KQI. The cell managed by the base station includes a cell 232 and a cell 231.

Step 206: The server predicts second fault information related to each superior fault according to the associated network data, where the second fault information is different from the first fault information.

The network failure indicated by the second failure information related to the superior failure refers to a network failure that can be caused by the superior failure.

The present step is described by taking the upper faults A11 and B11 in step 205 as an example. The server acquires the associated network data p1 related to A11, and then predicts the second fault information related to A11 according to the associated network data p1. Meanwhile, the server obtains The associated network data p2 associated with B11 then predicts the second failure information associated with B11 based on the associated network data p2.

For example, the first fault information is: "cell 231 service degradation", the upper fault of the network fault indicated by the first fault information is: the base station equipment fault, and the associated network data related to the superior fault acquired by the server is: the cell 232 KQI, then the second fault information related to the superior fault predicted by the server according to the associated network data may be: “Cell 232 service degradation”. The cell managed by the base station includes a cell 232 and a cell 231.

Optionally, the network data of the m types of services corresponds to the m operation and maintenance models, and the m operation and maintenance models are different from each other. Each operation and maintenance model is used to predict the network data of the corresponding service, and output fault information or non- accident details. Correspondingly, as shown in FIG. 5, step 206 may include:

Step 2061: The server inputs the associated network data to the associated operation and maintenance model to obtain information output by the associated operation and maintenance model, where the associated operation and maintenance model is an operation and maintenance model corresponding to the associated network data in the m operation and maintenance models.

Step 2062: When the information output by the associated operation and maintenance model is fault information, the server determines the information output by the associated operation and maintenance model as the second fault information related to each superior fault.

Assume that the eight services are: hardware failure prediction service, performance prediction service, resource prediction service, single domain alarm compression service, cross-domain alarm compression service, root cause alarm analysis service, KPI abnormality detection service, and service degradation abnormality detection service. There are 8 operation and maintenance models, which are: hardware failure prediction model, performance prediction model, resource prediction model, single domain alarm compression model, cross-domain alarm compression model, root cause alarm analysis model, KPI anomaly Detection model and business degradation anomaly detection model.

The upper-level fault and the first fault information shown in FIG. 4 are taken as an example. The upper-level fault is a fault of the base station equipment, and the three first fault information are: “cell 231 service degradation” and “ETH link connection abnormality”. And "high CPU usage." The associated network data obtained by the server related to the superior fault may be: KQI of the cell 232. Then, the server inputs the associated network data to the corresponding service degradation anomaly detection model, and obtains the fault information output by the service degradation anomaly detection model: “cell 232 service degradation”, after which the server determines “cell 232 service degradation” as the second. accident details.

Step 207: The server outputs k upper faults, k sets of fault information, and all predicted second fault information.

The server outputs k upper-level faults, k-group fault information, and all predicted second fault information, so that the worker performs fault processing according to k upper-level faults, k-group fault information, and all predicted second fault information.

Optionally, the server may display k upper faults, k sets of fault information, and all predicted second fault information.

In the embodiment of the present invention, the server may predict, according to the superior fault and the first fault information, the remaining network faults that may be caused by the superior fault. The superior diffusion labeling selection method provided by the embodiment of the invention enables the worker to process faults and potential faults in the network according to the superior fault, the first fault information and the second fault information, thereby improving the stability of the network and ensuring the normal network. run.

In summary, in the network operation and maintenance method provided by the embodiment of the present invention, the server can determine n (1 ≤ n ≤ m) first fault information according to network data of m (m ≥ 2) services, and then n Part or all of the first fault information is divided into k (1 ≤ k ≤ n) group fault information, and the fault of the network fault indicated by the first fault information in each set of fault information is the same, and then the server outputs the k group fault information. And k upper-level faults, k superordinate faults and k-group fault information are in one-to-one correspondence, so that the staff can timely deal with faults and potential faults in the network, and the method can comprehensively process multiple services.

In a second implementation manner, as shown in FIG. 6, the method for network operation and maintenance provided by the embodiment of the present invention may include:

Step 601: The server acquires network data of m types of services, where m≥2.

Step 601 can refer to step 201.

Step 602: The server determines, according to network data of the m types of services, n first fault information.

Each first fault information is used to indicate that a network fault occurs in the corresponding service, 1≤n≤m.

Optionally, the network data of the m types of services is in one-to-one correspondence with the m operation and maintenance models, and the m operation and maintenance models are different from each other. Correspondingly, the step 602 may include: the server inputs the network data of the corresponding service to the m operation and maintenance models. To obtain the information output by the m operation and maintenance models, the information output by each operation and maintenance model is fault information or non-fault information, and the information output by the m operation and maintenance models includes n fault information; the server determines the n fault information. For n first fault messages.

Step 602 can refer to step 202.

Step 603: The server divides part or all of the n first fault information into k sets of fault information.

The superior fault of the network fault indicated by the first fault information in each set of fault information is the same. The superior fault of the network fault indicated by any of the first fault information is a fault causing the network fault indicated by any of the first fault information, 1≤k≤n.

Step 603 can refer to step 203.

Step 604: The server outputs k sets of fault information and k upper faults.

The k upper level faults correspond to the k group fault information one by one.

The server outputs k sets of fault information and k superior faults, so that the staff can perform fault processing according to k upper faults and k sets of fault information. Further, the worker can also obtain potential faults in the network according to the superior fault and the first fault information output by the server, and process the potential faults.

Optionally, the server can display k sets of fault information and k superior faults.

Step 605: The server receives a first labeling instruction, where the first labeling instruction is used to indicate that the first fault information in the k group fault information is correctly predicted and the upper fault in the k upper fault faults are correctly predicted.

For example, after the server displays the k-group fault information and the k-level faults, the staff can mark the first fault information and the superior fault displayed by the server according to the actual fault condition of the network, and mark the first fault that the server predicts correctly. Information and superior failures. For example, the server may issue a prompt message for prompting the staff to use the first annotation symbol to mark the server to predict the correct first fault information and predict the correct superior fault, and use the second annotation symbol to mark the first fault of the server prediction error. Information and predicting the fault of the superior fault, after which the staff uses the first call symbol to mark the server to predict the correct first fault information and predict the correct superior fault, and uses the second call symbol to mark the first fault information of the server prediction error and Predict the wrong superior failure. Wherein the first label symbol and the second label symbol are different. For example, the first annotation symbol may be a checkmark "√", and the second annotation symbol may be a wrong identifier "×".

Taking the upper-level fault and the fault information shown in FIG. 4 as an example, it is assumed that the staff determines that the server is faulty about the base station equipment, the ETH link connection is abnormal, and the prediction of the high CPU occupancy rate is correct, and the service about the cell 231 is correct. The prediction of degradation is wrong, then the staff can use "√" to mark the three prediction results of "base station equipment failure", "ETH link connection abnormality", and "high CPU occupancy rate", and use "x" The prediction result of "cell 231 business deterioration" is marked, and the result is shown in FIG. 7.

Suppose the staff determines that the server's prediction about the ETH link connection anomaly is correct. If the other three predictions are wrong, the staff can use "√" to mark the prediction result of "ETH link connection anomaly". And use "X" to mark the other three prediction results, the result of which is shown in Figure 8.

Step 606: The server acquires a first sample set according to the first annotation instruction, where the first sample set includes information indicated by the first annotation instruction.

The server acquires a first sample set based on the first annotation instruction in step 605, the first sample set including the first fault information that is correctly predicted within the k sets of fault information and the upper fault that is correctly predicted by the k upper faults.

For example, k is equal to 2, and the first group of fault information includes three first fault information: x1, x2, and x3, and the upper fault of the network fault indicated by the three first fault information is A11; the second group of fault information includes The five first fault information: y1, y2, y3, y4, and y5, and the upper fault of the network fault indicated by the five first fault information is B11. Assuming that the first annotation instruction is used to indicate x1 and x2 in the first set of fault information, y4 and y5 in the second set of fault information, and the prediction of the superior fault A11 is correct, then the information included in the first sample set is :x1, x2, y4, y5, and A11.

Step 607: The server acquires, according to the first sample set, associated network data related to each superior fault in the first sample set.

It is assumed that the information included in the first sample set in step 605 is: x1, x2, y4, y5, and A11, and the server may acquire associated network data related to the superior fault A11 according to the first sample set, for example, A11 is a base station. If the device fails, then the associated network data associated with A11 may be: KQI of cell 232. The cell managed by the base station includes a cell 232.

Step 608: The server predicts second fault information related to each superior fault according to the associated network data, where the second fault information is different from the first fault information.

The network failure indicated by the second failure information related to the superior failure refers to a network failure that can be caused by the superior failure. In the embodiment of the present invention, since the server obtains the second fault information according to the first fault information that is correctly predicted in the k group fault information and the correct fault fault in the k upper faults, the accuracy of the second fault information is higher. .

Optionally, the network data of the m types of services corresponds to the m operation and maintenance models, and the m operation and maintenance models are different from each other. Each operation and maintenance model is used to predict the network data of the corresponding service, and output fault information or non- The fault information, correspondingly, step 608 may include: the server inputs the associated network data to the associated operation and maintenance model to obtain the information output by the associated operation and maintenance model, wherein the associated operation and maintenance model is associated with the associated network data in the m operation and maintenance models. Corresponding operation and maintenance model; when the information output by the associated operation and maintenance model is fault information, the server determines the information output by the associated operation and maintenance model as the second failure information related to each superior fault.

Step 608 can refer to step 206.

Step 609: The server outputs the first sample set and all the predicted second fault information.

The first sample set includes the first fault information that is correctly predicted within the k sets of fault information and the upper fault that is correctly predicted by the k upper faults. The server outputs the first sample set and all the predicted second fault information, so that the worker performs fault processing according to the server predicting the correct first fault information, predicting the correct superior fault, and predicting all the second fault information.

Optionally, the server may display the first sample set and all the second fault information predicted.

In the embodiment of the present invention, through steps 605 to 609, the server may predict a network fault that may be caused by a correct superior fault according to the staffing instruction, so that the staff can correctly predict the first fault information according to the prediction. All of the superior faults and predicted second fault information are processed in time for faults and potential faults in the network. Moreover, since the accuracy of the second fault information is high, the processing efficiency of the fault is also improved.

Step 610: The server determines all the predicted second fault information as a sample set to be labeled.

Optionally, the network data of the m types of services corresponds to the m operation and maintenance models, and the m operation and maintenance models are different from each other. Each operation and maintenance model is used to predict the network data of the corresponding service, and output fault information or non- accident details. In the embodiment of the present invention, in order to further update the operation and maintenance model whose evaluation index does not meet the service requirement, and further improve the accuracy of the fault prediction, the server may use the second fault information predicted by the operation and maintenance model in step 608. The sample set to be labeled is determined, so that the worker labels the sample set to be labeled, and obtains the second fault information that is predicted correctly.

Step 611: The server receives a second labeling instruction, where the second labeling instruction is used to indicate that the second fault information is correctly predicted in the sample set to be labeled.

For example, after the server displays all the predicted second fault information, the staff can mark the second fault information displayed by the server according to the actual fault condition of the network, and mark the second fault information that the server predicts correctly. Refer to Figure 7 and Figure 8 in step 605 for the labeling method.

Step 612: The server acquires a second sample set according to the second annotation instruction, where the second sample set includes information indicated by the second annotation instruction.

The server acquires a second sample set based on the second annotation instruction in step 611, the second sample set includes second fault information that is correctly predicted within the sample set to be labeled.

For example, all of the second fault information predicted in step 608 includes z1, z2, z3, and z4. Assuming that the second annotation instruction is used to indicate that the predictions of z1 and z2 are correct, then the information included in the second sample set is: z1 and z2.

Step 613: The server determines the first sample set and the second sample set as the target sample set.

The first sample set includes the first fault information that is correctly predicted within the k sets of fault information and the upper fault that is correctly predicted by the k upper faults, and the second sample set includes the second fault information that is correctly predicted within the sample set to be labeled, and the server will The first sample set and the second sample set are determined as a target sample set, and the target sample set is used to update an operation and maintenance model in which the evaluation index does not satisfy the business requirement.

Step 614: The server determines, according to the target sample set, an evaluation index of the first operation and maintenance model, where the first operation and maintenance model is any one of the m operation and maintenance models.

The server determines an evaluation index of the first operation and maintenance model according to the predicted first failure information, the predicted correct superior failure, and the predicted correct second failure information.

Optionally, the evaluation index of the first operation and maintenance model may be the accuracy of the first operation and maintenance model. The accuracy of the model is the ratio of the number of correct results predicted by the model to the total number of predicted results. The higher the accuracy of the model, the better the prediction effect of the model.

Step 615: When the evaluation index of the first operation and maintenance model does not belong to the specified evaluation index range, the server updates the first operation and maintenance model by using the target sample set.

When the evaluation index of the first operation and maintenance model is the accuracy of the first operation and maintenance model, the corresponding specified evaluation index range may be [f, 1], for example, f may be equal to 0.4, and the server may be in the first operation and maintenance model. When the evaluation index is less than 0.4, the first operation and maintenance model is updated by using the target sample set. For example, the supervised learning algorithm in the machine learning algorithm can be used to train the first operation and maintenance model. The model training process can refer to related technologies, and details are not described herein.

Optionally, the evaluation index of the first operation and maintenance model may also be the precision of the first operation and maintenance model, and the higher the precision of the model, the better the prediction effect of the model. The evaluation index of the first operation dimension model can also be the false discovery rate, and the smaller the error detection rate of the model, the better the prediction effect of the model. The evaluation index of the first operation and maintenance model may also be an error omission rate, etc. The embodiment of the present invention does not limit the evaluation index of the first operation and maintenance model, and the specified evaluation index range may be based on the determined evaluation index of the first operation and maintenance model. determine.

Optionally, each operation and maintenance model in the m operation and maintenance models is managed by a pair of application units and a model trainer, and the application unit is configured to determine an evaluation index of the first operation and maintenance model according to the target sample set, and in the first operation When the evaluation index of the dimensional model does not belong to the specified evaluation index range, the model training device sends a model update request, and the model training device is configured to update the first operation and maintenance model by using the target sample set according to the model update request sent by the application unit.

In the embodiment of the present invention, through the steps 610 to 615, the server may obtain the second fault information that is predicted correctly according to the labeling instruction of the staff, and then predict the correct first fault information and predict the correct superior fault and the prediction according to the prediction. The second fault information updates the operation and maintenance model that the evaluation index does not meet the business requirements, improves the accuracy of the fault prediction, and further improves the fault processing efficiency.

The embodiment of the present invention effectively predicts faults and potential faults in the network by using the operation and maintenance experience of the staff. In the embodiment of the present invention, the server can update the operation and maintenance model in time, and achieve the purpose of timely prediction and accurate prediction. Reduce labor costs and improve the processing efficiency of faults. Through the method for actively preventing passive processing of network operation and maintenance provided by the embodiment of the present invention, the staff can quickly know the running state of the network, timely deal with faults and potential faults in the network, improve the stability of the network, and ensure the network. normal operation.

In summary, in the network operation and maintenance method provided by the embodiment of the present invention, the server can determine n (1 ≤ n ≤ m) first fault information according to network data of m (m ≥ 2) services, and then n Part or all of the first fault information is divided into k (1 ≤ k ≤ n) group fault information, and the fault of the network fault indicated by the first fault information in each set of fault information is the same, and then the server outputs the k group fault information. And k upper-level faults, k upper-level faults and k-group fault information are in one-to-one correspondence, so that the staff can timely deal with faults and potential faults in the network, and through this method, comprehensive processing of multiple services can be performed, and The operation and maintenance model whose evaluation index does not meet the business requirements is automatically updated, which improves the accuracy of fault prediction and improves the processing efficiency of faults.

It should be noted that the sequence of the steps of the network operation and maintenance method provided by the embodiment of the present invention may be appropriately adjusted, and the steps may also be correspondingly increased or decreased according to the situation, and any technology familiar to those skilled in the art may be disclosed in the present application. The methods that can be easily conceived within the scope of the present invention are covered by the scope of the present application and therefore will not be described again.

The embodiment of the present invention provides a device for network operation and maintenance. The network operation and maintenance device can be used for the server shown in FIG. 1. As shown in FIG. 9, the network operation and maintenance device 900 includes:

The first obtaining module 910 is configured to perform step 201 or step 601 in the foregoing embodiment.

The first determining module 920 is configured to perform step 202 or step 602 in the foregoing embodiment.

The dividing module 930 is configured to perform step 203 or step 603 in the foregoing embodiment.

The first output module 940 is configured to perform step 204 or step 604 in the foregoing embodiment.

Optionally, the network data of the m types of services is in one-to-one correspondence with the m operation and maintenance models, and the m operation and maintenance models are different from each other. The first determining module 920 is configured to perform step 2021 or step 2022 in the foregoing embodiment.

Further, as shown in FIG. 10, the network operation and maintenance apparatus 900 may further include:

The second obtaining module 950 is configured to perform step 205 in the foregoing embodiment.

The first prediction module 960 is configured to perform step 206 in the foregoing embodiment.

The second output module 970 is configured to perform step 207 in the foregoing embodiment.

The meaning of other marks in FIG. 10 can be referred to FIG.

Further, as shown in FIG. 11, the device 900 of the network operation and maintenance may further include:

The first receiving module 980 is configured to perform step 605 in the foregoing embodiment.

The third obtaining module 990 is configured to perform step 606 in the foregoing embodiment.

The fourth obtaining module 991 is configured to perform step 607 in the foregoing embodiment.

The second prediction module 992 is configured to perform step 608 in the foregoing embodiment.

The third output module 993 is configured to perform step 609 in the foregoing embodiment.

Optionally, the network data of the m types of services corresponds to the m operation and maintenance models, and the m operation and maintenance models are different from each other. Each operation and maintenance model is used to predict the network data of the corresponding service, and output fault information or non- accident details. Further, as shown in FIG. 11, the device 900 of the network operation and maintenance may further include:

The second determining module 994 is configured to perform step 610 in the foregoing embodiment.

The second receiving module 995 is configured to perform step 611 in the foregoing embodiment.

The fifth obtaining module 996 is configured to perform step 612 in the foregoing embodiment.

The third determining module 997 is configured to perform step 613 in the foregoing embodiment.

The fourth determining module 998 is configured to perform step 614 in the foregoing embodiment.

The update module 999 is configured to perform step 615 in the above embodiment.

The meaning of other marks in Fig. 11 can be referred to Fig. 9.

Optionally, the network data of the m types of services corresponds to the m operation and maintenance models, and the m operation and maintenance models are different from each other. Each operation and maintenance model is used to predict the network data of the corresponding service, and output fault information or non- The fault information, the first prediction module 960 in FIG. 10 or the second prediction module 992 in FIG. 11 is configured to perform step 2061 and step 2062 in the foregoing embodiment, including:

Inputting associated network data to the associated operation and maintenance model to obtain information outputted by the associated operation and maintenance model, wherein the associated operation and maintenance model is an operation and maintenance model corresponding to the associated network data in the m operation and maintenance models;

When the information output by the associated operation and maintenance model is fault information, the information output by the associated operation and maintenance model is determined as the second fault information related to each superior fault.

In summary, the network operation and maintenance device provided by the embodiment of the present invention can determine n (1 ≤ n ≤ m) first fault information according to network data of m (m ≥ 2) services, and then n Part or all of the first fault information is divided into k (1 ≤ k ≤ n) group fault information, and the fault of the network fault indicated by the first fault information in each set of fault information is the same, and then the server outputs the k group fault information. And k upper-level faults, k upper-level faults and k-group fault information are in one-to-one correspondence, so that the staff can timely deal with faults and potential faults in the network, and the device can comprehensively process multiple services, and can also The operation and maintenance model whose evaluation index does not meet the business requirements is updated, which improves the accuracy of fault prediction and improves the processing efficiency of faults.

FIG. 12 is a schematic structural diagram of an apparatus for network operation and maintenance provided by an embodiment of the present invention, and the apparatus may be used in the server shown in FIG. 1. As shown in FIG. 12, the apparatus includes a processor 1201 (such as a CPU), a memory 1202, a network interface 1203, and a bus 1204. The bus 1204 is used to connect the processor 1201, the memory 1202, and the network interface 1203. The memory 1202 may include a random access memory (RAM), and may also include a non-volatile memory, such as at least one disk storage. The communication connection between the server and the communication device is implemented through a network interface 1203, which may be wired or wireless. The program 12021 is stored in the memory 1202. The program 12021 is used to implement various application functions. The processor 1201 is configured to execute the program 12021 stored in the memory 1202 to implement the network operation and maintenance method shown in FIG. 2 or FIG. 6.

A person skilled in the art can clearly understand that, for the convenience and brevity of the description, the specific working process of the foregoing apparatus and module can be referred to the corresponding process in the foregoing method embodiment, and details are not described herein again.

In the above embodiments, it may be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented in software, it may be implemented in whole or in part in the form of a computer program product comprising one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions described in accordance with embodiments of the present invention are generated in whole or in part. The computer can be a general purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a readable storage medium of a computer or transferred from one computer readable storage medium to another computer readable storage medium, for example, the computer instructions can be from a website site, computer, server or data The center transmits to another website site, computer, server, or data center by wire (eg, coaxial cable, fiber optic, digital subscriber line) or wireless (eg, infrared, wireless, microwave, etc.). The computer readable storage medium can be any available media that can be accessed by a computer or a data storage device such as a server, data center, or the like that includes one or more available media. The usable medium may be a magnetic medium (eg, a floppy disk, a hard disk, a magnetic tape), an optical medium, or a semiconductor medium (eg, a solid state hard disk) or the like.

In the several embodiments provided by the present application, it should be understood that the disclosed apparatus and method may be implemented in other manners. For example, the device embodiments described above are merely illustrative. For example, the division of the modules is only a logical function division. In actual implementation, there may be another division manner, for example, multiple modules or components may be combined or Can be integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interface, device or unit, and may be in an electrical, mechanical or other form.

A person skilled in the art may understand that all or part of the steps of implementing the above embodiments may be completed by hardware, or may be instructed by a program to execute related hardware, and the program may be stored in a computer readable storage medium. The storage medium mentioned may be a read only memory, a magnetic disk or an optical disk or the like.

The above description is only an optional embodiment of the present application, and is not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application are included in the protection of the present application. Within the scope.

Claims (14)

A method for network operation and maintenance, characterized in that the method comprises: Obtain network data of m kinds of services, m≥2; Determining, according to the network data of the m types of services, n first fault information, where each of the first fault information is used to indicate that a corresponding service has a network fault, 1≤n≤m; And dividing part or all of the n first fault information into k sets of fault information, where the first fault of the network fault indicated by the first fault information in each set of fault information is the same, and the network indicated by any first fault information The fault of the fault is a fault that causes the network fault indicated by any of the first fault information, 1≤k≤n; The k sets of fault information and k upper faults are output, and the k upper faults are in one-to-one correspondence with the k sets of fault information. The method according to claim 1, wherein after the outputting the k sets of fault information and the k upper faults, the method further comprises: Obtaining associated network data related to each of the upper faults according to the k upper faults and the first fault information corresponding to each of the upper faults; And predicting, according to the associated network data, second fault information related to each of the upper faults, where the second fault information is different from the first fault information; The k upper level faults, the k sets of fault information, and all of the predicted second fault information are output. The method according to claim 1, wherein after the outputting the k sets of fault information and the k upper faults, the method further comprises: Receiving a first labeling instruction, where the first labeling instruction is used to indicate that the first fault information that is correctly predicted in the k group fault information and the upper fault that is correctly predicted in the k upper faults; Acquiring a first sample set based on the first annotation instruction, where the first sample set includes information indicated by the first annotation instruction; Acquiring associated network data related to each superior fault in the first sample set according to the first sample set; And predicting, according to the associated network data, second fault information related to each of the upper faults, where the second fault information is different from the first fault information; The first sample set and all of the predicted second fault information are output. The method according to claim 3, wherein the network data of the m types of services are in one-to-one correspondence with the m operation and maintenance models, and the m operation and maintenance models are different from each other, and each operation and maintenance model is used for Forecasting network data corresponding to the service, and outputting fault information or non-fault information; After the outputting the first sample set and the predicted second fault information, the method further includes: Determining all predicted second fault information as a sample set to be labeled; Receiving a second labeling instruction, where the second labeling instruction is used to indicate that the second fault information is correctly predicted in the sample set to be labeled; Acquiring a second sample set based on the second annotation instruction, the second sample set including information indicated by the second annotation instruction; Determining the first sample set and the second sample set as a target sample set; Determining, according to the target sample set, an evaluation index of the first operation and maintenance model, where the first operation and maintenance model is any one of the m operation and maintenance models; When the evaluation index of the first operation and maintenance model does not belong to the specified evaluation index range, the first operation and maintenance model is updated by using the target sample set. The method according to claim 2 or 3, wherein the network data of the m types of services are in one-to-one correspondence with the m operation and maintenance models, and the m operation and maintenance models are different from each other, and each operation and maintenance model is used. Predicting network data of the corresponding service, and outputting fault information or non-fault information; The predicting the second fault information related to each of the upper faults according to the associated network data includes: Inputting the associated network data to the associated operation and maintenance model to obtain information output by the associated operation and maintenance model, where the associated operation and maintenance model is the operation and maintenance corresponding to the associated network data in the m operation and maintenance models model; And determining, when the information output by the associated operation and maintenance model is fault information, the information output by the associated operation and maintenance model as the second fault information related to each of the upper faults. The method according to claim 1, wherein the network data of the m types of services are in one-to-one correspondence with the m operation and maintenance models, and the m operation and maintenance models are different from each other; Determining, according to the network data of the m types of services, the n first fault information, including: And inputting network data of the corresponding service to the m operation and maintenance models to obtain information output by the m operation and maintenance models, where the information output by each operation and maintenance model is fault information or non-fault information, the m The information output by the operation and maintenance model includes n fault information; The n pieces of fault information are determined as the n first fault information. A device for network operation and maintenance, characterized in that the device comprises: a first acquiring module, configured to acquire network data of m types of services, where m≥2; a first determining module, configured to determine n first fault information according to the network data of the m types of services, where each of the first fault information is used to indicate that a corresponding service has a network fault, 1≤n≤m; a dividing module, configured to divide part or all of the n first fault information into k group fault information, where the first fault of the network fault indicated by the first fault information in each set of fault information is the same, any first fault The superior fault of the network fault indicated by the information is a fault causing the network fault indicated by any of the first fault information, 1≤k≤n; The first output module is configured to output the k group fault information and the k upper faults, where the k upper faults are in one-to-one correspondence with the k group fault information. The device according to claim 7, wherein the device further comprises: a second acquiring module, configured to acquire, according to the k upper faults and the first fault information corresponding to each upper fault, associated network data related to each of the upper faults; a first prediction module, configured to predict, according to the associated network data, second fault information related to each of the upper faults, where the second fault information is different from the first fault information; And a second output module, configured to output the k upper faults, the k sets of fault information, and all predicted second fault information. The device according to claim 7, wherein the device further comprises: a first receiving module, configured to receive a first labeling instruction, where the first labeling instruction is used to indicate that the first fault information that is correctly predicted in the k group fault information and the upper fault that is correctly predicted in the k upper faults; a third acquiring module, configured to acquire a first sample set based on the first annotation instruction, where the first sample set includes information indicated by the first annotation instruction; a fourth acquiring module, configured to acquire, according to the first sample set, associated network data related to each upper-level fault in the first sample set; a second prediction module, configured to predict, according to the associated network data, second fault information related to each of the upper faults, where the second fault information is different from the first fault information; And a third output module, configured to output the first sample set and all second fault information predicted. The device according to claim 9, wherein the network data of the m types of services are in one-to-one correspondence with the m operation and maintenance models, and the m operation and maintenance models are different from each other, and each operation and maintenance model is used for Forecasting the network data corresponding to the service, and outputting fault information or non-fault information, The device also includes: a second determining module, configured to determine all the predicted second fault information as a sample set to be labeled; a second receiving module, configured to receive a second labeling instruction, where the second labeling instruction is used to indicate that the second fault information is correctly predicted in the sample set to be labeled; a fifth acquiring module, configured to acquire a second sample set based on the second annotation instruction, where the second sample set includes information indicated by the second annotation instruction; a third determining module, configured to determine the first sample set and the second sample set as a target sample set; a fourth determining module, configured to determine, according to the target sample set, an evaluation index of the first operation and maintenance model, where the first operation and maintenance model is any one of the m operation and maintenance models; And an update module, configured to update the first operation and maintenance model by using the target sample set when an evaluation index of the first operation and maintenance model does not belong to a specified evaluation index range. The device according to claim 8 or 9, wherein the network data of the m types of services are in one-to-one correspondence with the m operation and maintenance models, and the m operation and maintenance models are different from each other, and each operation and maintenance model is used. Predicting the network data of the corresponding service, and outputting fault information or non-fault information, The first prediction module or the second prediction module is configured to: Inputting the associated network data to the associated operation and maintenance model to obtain information output by the associated operation and maintenance model, where the associated operation and maintenance model is the operation and maintenance corresponding to the associated network data in the m operation and maintenance models model; And determining, when the information output by the associated operation and maintenance model is fault information, the information output by the associated operation and maintenance model as the second fault information related to each of the upper faults. The device according to claim 7, wherein the network data of the m types of services are in one-to-one correspondence with the m operation and maintenance models, and the m operation and maintenance models are different from each other. The first determining module is configured to: And inputting network data of the corresponding service to the m operation and maintenance models to obtain information output by the m operation and maintenance models, where the information output by each operation and maintenance model is fault information or non-fault information, the m The information output by the operation and maintenance model includes n fault information; The n pieces of fault information are determined as the n first fault information. A computer readable storage medium, wherein the computer readable storage medium stores instructions for causing a computer to perform any of claims 1 to 6 when the computer readable storage medium is run on a computer The method of network operation and maintenance. A device for network operation and maintenance, characterized in that the device comprises: a processor, a memory, a network interface and a bus, The bus is configured to connect the processor, the memory, and the network interface, and the processor is configured to execute a program stored in the memory to implement the network operation and maintenance method according to any one of claims 1 to 6. .

Images