An intelligent operation and maintenance control method and device of a core network

Abstract

This application discloses an intelligent operation and maintenance control method and apparatus for a core network, relating to the field of operation and maintenance technology. The method includes: in response to detecting an anomaly in the core network, identifying at least one target network element in the core network with the anomaly; obtaining target state data corresponding to the target network element, the target state data including at least one of: network fault level corresponding to the target network element, the value of at least one performance indicator, and service experience difference values; based on the target state data, using a reinforcement learning model, determining the target operation and maintenance action required by the target network element, the reinforcement learning model including the mapping relationship between different state data and operation and maintenance actions; determining the feasibility result of the target network element executing the target operation and maintenance action; and in response to the feasibility result indicating that the target network element is suitable for executing the target operation and maintenance action, controlling the target network element to execute the target operation and maintenance action. This application can perform operation and maintenance processing in a timely and effective manner when anomalies occur in the core network.

Description

Technical Field

[0001] This application relates to the field of operation and maintenance technology, and in particular to an intelligent operation and maintenance control method and device for a core network. Background Technology

[0002] The core network is the core component for control, management, and intelligent decision-making in mobile communication networks.

[0003] To ensure network communication quality, it is necessary to promptly detect and address faults in the core network. However, current core network operation and maintenance management mainly relies on manual intervention, which makes it impossible to implement maintenance actions in a timely and effective manner. Summary of the Invention

[0004] In view of the above problems, this application provides an intelligent operation and maintenance control method and apparatus for the core network, so as to perform operation and maintenance in a timely and effective manner when anomalies occur in the core network. The specific solution is as follows:

[0005] The first aspect of this application provides an intelligent operation and maintenance control method for a core network, comprising:

[0006] In response to the detection of an anomaly in the core network, at least one target network element in the core network is identified as having an anomaly.

[0007] For each target network element, obtain the target status data corresponding to the target network element. The target status data includes at least one of the following: the network fault level corresponding to the target network element, the index value of at least one performance indicator, and the service experience difference value of the service corresponding to the target network element.

[0008] Based on the target state data, the target operation and maintenance actions required for the target network element are determined using a trained reinforcement learning model. The reinforcement learning model includes the mapping relationship between different state data and operation and maintenance actions.

[0009] The feasibility result of the target network element performing the target operation and maintenance action is determined, and the feasibility result is used to indicate whether the target network element is suitable for performing the target operation and maintenance action;

[0010] In response to the feasibility results indicating that the target network element is suitable for performing the target operation and maintenance action, the target network element is controlled to perform the target operation and maintenance action.

[0011] In one possible implementation, the step of determining at least one target network element in the core network that exhibits an anomaly in response to detecting an anomaly in the core network includes:

[0012] Obtain data for at least one evaluation indicator corresponding to the core network;

[0013] Based on the data of at least one evaluation index, the evaluation result of the core network is determined using the target model, and the evaluation result is used to characterize whether there is an anomaly in the core network;

[0014] In response to the evaluation result indicating an anomaly in the core network, at least one target network element in the core network is identified as having an anomaly.

[0015] In another possible implementation, obtaining at least one evaluation index data corresponding to the core network includes:

[0016] Obtain the current network element characteristic data of each network element in the core network. The network element characteristic data includes at least one of the following: the index value of at least one performance indicator of the network element, alarm data, operation log data, and configuration data.

[0017] Obtain service indicator data for at least one user currently managed by the core network, wherein the user's service indicator data includes: the user's user identifier and the service identifier and experience indicator data corresponding to the service used by the user;

[0018] The determination of the evaluation result of the core network based on the at least one evaluation index data using the target model includes at least one of the following:

[0019] Based on the network element feature data, the anomaly prediction result of the core network is determined using the first prediction model. The anomaly prediction result includes: anomaly probability and anomaly level.

[0020] For each user, based on the user's service indicator data, the second prediction model is used to determine the service quality result of the core network. The service quality result is used to characterize whether the service quality corresponding to the service used by the user is qualified.

[0021] The evaluation results indicating anomalies in the core network include at least one of the following:

[0022] The anomaly probability exceeds a set probability value and the anomaly level exceeds a set level;

[0023] The service quality result indicates that the service used by the user is of substandard quality.

[0024] In another possible implementation, identifying at least one target network element in the core network that exhibits an anomaly includes at least one of the following:

[0025] Based on the index values, alarm data, and operation log data of at least one performance indicator corresponding to each network element in the core network, at least one target network element in the core network with an anomaly is identified.

[0026] In response to the user's service quality result indicating that the service quality corresponding to the service used by the user is unqualified, based on the user's user identifier, at least one network element responsible for managing the user is identified as at least one target network element with an anomaly.

[0027] In another possible implementation, the result of determining the feasibility of the target network element performing the target operation and maintenance action includes:

[0028] Based on the target operation and maintenance action, determine the operation and maintenance impact data of the target network element;

[0029] Based on the aforementioned operational impact data, the feasibility of the target network element performing the target operational action is determined.

[0030] The operational impact data includes at least one of the following:

[0031] The resource requirements information required for the target network element to perform the target operation and maintenance action;

[0032] The target network element has historically executed the target maintenance actions, and the maintenance execution results are as follows:

[0033] Information on the service impact on at least one target user that can be affected by the target network element performing the target operation and maintenance action.

[0034] In another possible implementation, where the operational impact data includes the resource demand information, determining the feasibility of the target network element performing the target operational action based on the operational impact data includes:

[0035] Obtain the current resource status information of the target network element;

[0036] Based on the resource status information, the resource usage of the target network element after the current moment is predicted using a trained third prediction model;

[0037] Based on the resource usage and resource demand information, the feasibility of the target network element performing the target operation and maintenance action is determined.

[0038] In another possible implementation, the operation and maintenance execution results of the target network element in the past execution of the target operation and maintenance actions are determined in the following way:

[0039] Obtain the current scene status information of the target network element, wherein the scene status information is at least used to characterize the current network status and resource status of the target network element;

[0040] Obtain at least one historical operation and maintenance data of the target network element. The historical operation and maintenance data includes: historical operation and maintenance actions performed by the target network element, historical scenario status information when the target network element performed the historical operation and maintenance actions, and operation and maintenance execution results of the historical operation and maintenance actions. The operation and maintenance execution results are used to characterize whether the historical operation and maintenance actions were successfully executed.

[0041] From the at least one historical operation and maintenance data, at least one target historical operation and maintenance data is determined, wherein the historical operation and maintenance actions in the target historical operation and maintenance data are the same as the target operation and maintenance actions and the historical scenario status information is similar to the scenario status information;

[0042] The operation and maintenance execution results in the historical operation and maintenance data of each target are determined as the operation and maintenance execution results of the target network element in the history of performing the target operation and maintenance actions.

[0043] In another possible implementation, determining the feasibility of the target network element performing the target maintenance action based on the maintenance impact data includes:

[0044] Based on the scenario state information and the historical scenario state information and operation and maintenance execution results in the target historical operation and maintenance data, the training fourth prediction model is used to predict the success rate of the target network element executing the target operation and maintenance action. The success rate exceeding a first set value indicates that the target network element is suitable for executing the target operation and maintenance action.

[0045] In another possible implementation, the service impact information of at least one target user affected by the target network element performing the target maintenance action is determined in the following way:

[0046] Identify at least one associated user that can be affected by the target network element performing the target operation and maintenance action;

[0047] The user level, user behavior data, and business indicator information of the associated user are obtained. The business indicator information is used to characterize the business performance of the target service currently used by the associated user. The target service is a service provided by the target network element.

[0048] The feasibility result of determining the target network element to perform the target operation and maintenance action based on the operation and maintenance impact data includes:

[0049] If the target user whose user level is higher than the set level is not included among the at least one associated user, it is determined that the target network element is suitable for performing the target operation and maintenance action.

[0050] In response to the fact that at least one of the at least associated users includes at least one target user, based on the user behavior data and business indicator information of the target user, a trained fifth prediction model is used to predict the business quality score of the target service after the target network element performs the target operation and maintenance action, wherein the business quality score exceeding a second set value indicates that the target network element is suitable for performing the target operation and maintenance action.

[0051] Furthermore, this application also provides an intelligent operation and maintenance control device for a core network, comprising:

[0052] A network element determination unit is used to determine at least one target network element in the core network that has an anomaly in response to the detection of an anomaly in the core network.

[0053] The status acquisition unit is used to acquire target status data corresponding to each target network element. The target status data includes at least one of the following: the network fault level corresponding to the target network element, the index value of at least one performance indicator, and the service experience difference value of the service.

[0054] An action determination unit is used to determine the appropriate target operation and maintenance action for the target network element based on the target state data and using a trained reinforcement learning model. The reinforcement learning model includes the mapping relationship between different state data and operation and maintenance actions.

[0055] The feasibility determination unit is used to determine the feasibility result of the target network element performing the target operation and maintenance action, and the feasibility result is used to indicate whether the target network element is suitable for performing the target operation and maintenance action;

[0056] An action execution unit is configured to control the target network element to perform the target operation and maintenance action in response to the feasibility result indicating that the target network element is suitable for performing the target operation and maintenance action.

[0057] In another aspect, this application also provides an electronic device that includes at least a processor and a memory;

[0058] This processor is used to execute any of the intelligent operation and maintenance control methods for the core network described above;

[0059] The memory is used to store the programs required by the processor to perform operations.

[0060] On another front, this application also provides a computer program product including computer-readable instructions, which, when executed on an electronic device, enable the electronic device to implement any of the above-mentioned intelligent operation and maintenance control methods for the core network.

[0061] In another aspect, this application also provides a computer-readable storage medium that carries one or more computer programs. When the one or more computer programs are executed by an electronic device, the electronic device can implement any of the above-mentioned intelligent operation and maintenance control methods for the core network.

[0062] By employing the above technical solution, in this application, after detecting an anomaly in the core network, the target network element with the anomaly is identified, and the target state data of the target network element is obtained. Based on the target state data and the corresponding operation and maintenance actions for different state data in the model obtained through reinforcement learning, this application can determine the target operation and maintenance actions required to resolve the anomaly of the target network element. This allows for timely and efficient determination of the operation and maintenance actions required to resolve the anomaly when it is detected in the core network. Moreover, before the target network element executes the target operation and maintenance action, this application also evaluates the feasibility of the target network element executing the target operation and maintenance action. Only when the feasibility result shows that the target network element is suitable for executing the target operation and maintenance action will the target network element be controlled to execute the target operation and maintenance action. This reduces anomaly situations such as failure to repair anomalies or decline in service quality caused by the target network element being unsuitable for executing the target operation and maintenance action. Consequently, it enables more timely and reasonable implementation of operation and maintenance actions when anomalies exist in the core network, improving the timeliness and reliability of core network operation and maintenance. Attached Figure Description

[0063] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and the originals and elements are not necessarily drawn to scale.

[0064] Figure 1 A flowchart illustrating the intelligent operation and maintenance control method for the core network provided in this application;

[0065] Figure 2 Another flowchart illustrating the intelligent operation and maintenance control method for the core network provided in this application;

[0066] Figure 3 A schematic diagram of an application scenario framework to which the solution of this application applies;

[0067] Figure 4 A schematic diagram of the composition structure of the intelligent operation and maintenance control device for the core network provided in this application;

[0068] Figure 5 A schematic diagram of the composition structure of an electronic device provided in this application. Detailed Implementation

[0069] The embodiments of this application are described below with reference to the accompanying drawings. The terminology used in the implementation section of this application is only for explaining specific embodiments and is not intended to limit the application. Those skilled in the art will recognize that, with technological advancements and the emergence of new scenarios, the technical solutions provided in the embodiments of this application are also applicable to similar technical problems.

[0070] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such terms are interchangeable where appropriate; this is merely a way of distinguishing objects with the same attributes in the embodiments of this application. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion, so that a process, method, system, product, or apparatus that comprises a series of elements is not necessarily limited to those elements, but may include other elements not explicitly listed or inherent to those processes, methods, products, or apparatuses.

[0071] like Figure 1 This illustration shows a schematic diagram of an implementation flow of the intelligent operation and maintenance control method for the core network provided in this application. The method of this embodiment can be applied to an electronic device, which can be a computer device independent of the core network, such as a server or other computer device outside the core network. In this application, the electronic device can run intelligent programs such as intelligent agents, and the running intelligent programs can perform the relevant operations of the embodiments of this application.

[0072] The method in this embodiment may include:

[0073] S101, in response to detecting an anomaly in the core network, determine at least one target network element in the core network that has an anomaly.

[0074] In this application, there are various possible implementations for detecting anomalies in the core network and identifying the target network element that is abnormal, and no restrictions are imposed on this.

[0075] S102, for each target network element, obtain the target status data corresponding to that target network element.

[0076] In this application, electronic devices can communicate with network elements of the core network based on at least one network protocol, or communicate indirectly with network elements of the core network through a base station, in order to obtain various related data such as status data of network elements in the core network, without any specific limitations.

[0077] The target status data of the target network element includes at least one of the following: the network fault level corresponding to the target network element, the value of at least one performance indicator, and the service experience difference value of the service corresponding to the target network element. The service corresponding to the target network element is the service that the target network element can provide or supports.

[0078] In this application, the network fault level of the target network element can be determined by collecting network anomaly information such as fault alarms from the target network element using a designated fault monitoring system (such as a base station or the target network element), and based on this network anomaly information. The network fault level of the target network element can characterize the degree of network fault corresponding to the target network element.

[0079] Among them, the performance indicators of the target network element refer to the indicators that can reflect the network performance or resource usage status of the target network element. The specific performance indicators to be monitored can be set according to actual needs, and there are no restrictions on this. For example, performance indicators may include, but are not limited to, the traffic load, latency, central processing unit (CPU) utilization, and memory utilization of the target network element.

[0080] The service corresponding to the target network element refers to the service that the target network element can provide to users. The service experience difference value for the service corresponding to the target network element is an indicator used to characterize the quality of the service. For example, this service experience difference value can be used to measure the difference between the service quality actually perceived by the user and the service quality measured at the network technology level. Specifically, the service experience difference value can be determined by a designated user experience testing system based on one or more of the following: network performance indicators such as network latency, packet loss rate, and throughput corresponding to the target network element, as well as user evaluation data regarding the service. There are no restrictions on the specific implementation method and process for determining the service experience difference value.

[0081] It is understandable that, in order to more accurately determine the maintenance actions required for the target network element, the target status data in this application may include all three types of data.

[0082] For example, the types of data included in the target status data of the target network element and the value range of each data type can be shown in Table 1 below:

[0083] Table 1

[0084]

[0085] Among them, KPI indicators are performance metrics for the target network elements. Of course, Table 1 is just an example, and there are no restrictions on the data source methods for each indicator.

[0086] S103. Based on the target state data of the target network element, the target operation and maintenance actions required for the target network element are determined using a trained reinforcement learning model.

[0087] In this application, the target maintenance action is the maintenance action determined by the reinforcement learning model that is required to resolve the anomaly of the target network element.

[0088] The reinforcement learning model is constructed using reinforcement learning and includes the mapping relationship between different state data and operation and maintenance actions. Based on this, this application can determine the operation and maintenance actions corresponding to the target state data of the target network element based on the reinforcement learning model, and the determined operation and maintenance actions are the target operation and maintenance actions to be executed by the target network element.

[0089] In this application, the mapping relationships in the trained reinforcement learning model can include appropriate operational actions for different state data (i.e., state spaces). Operational actions refer to the actions taken to maintain, optimize, or repair anomalies in target network elements.

[0090] For example, Table 2 shows some examples of selectable operational actions in the reinforcement learning model of this application:

[0091] Table 2

[0092]

[0093] Among these, adjusting load balancing can involve adjusting process allocation strategies; resource elastic scaling can involve dynamically adding or releasing computing resources; routing optimization can be achieved by changing data transmission paths; and fault isolation can involve disconnecting faulty nodes or starting backup nodes.

[0094] S104, Determine the feasibility of the target network element performing the target operation and maintenance action.

[0095] The feasibility result is used to indicate whether the target network element is suitable for performing the target operation and maintenance actions.

[0096] The suitability of a target network element for executing the target maintenance action indicates that the target network element possesses the capability to perform the target maintenance action, can repair anomalies in the target network element through the execution of the target maintenance action, and / or will not affect the service quality of the services supported by the target network element. Accordingly, whether a target network element is suitable for executing the target maintenance action can be determined by analyzing one or more of the following: whether the target network element has the capability to execute the target maintenance action; whether it can successfully execute the target maintenance action; and whether executing the target maintenance action will lead to a decrease in service quality, without any specific restrictions.

[0097] For example, to determine whether a target network element has the capability to perform the target maintenance action, we can analyze whether the available resources of the target network element support the execution of the target maintenance action. If the available resources of the target network element cannot support the execution of the target maintenance action, then the target network element is not suitable for performing the target maintenance action.

[0098] In this application, there are no restrictions on the specific implementation methods for determining the feasibility result.

[0099] S105, in response to the feasibility result indicating that the target network element is suitable for performing the target operation and maintenance action, control the target network element to perform the target operation and maintenance action.

[0100] It is understandable that if the feasibility result indicates that the target network element is suitable for performing the target operation and maintenance action, it means that the anomaly of the target network element can be resolved by performing the target operation and maintenance action. Therefore, in this case, controlling the target network element to perform the target operation and maintenance action can reduce the occurrence of abnormal states such as failure to repair anomalies or degradation of service quality of the target network element.

[0101] As described above, in this application, after detecting an anomaly in the core network, the target network element with the anomaly is identified, and its target state data is obtained. Based on the target state data and the corresponding maintenance actions for different state data in the model obtained through reinforcement learning, this application can determine the target maintenance actions required to resolve the anomaly of the target network element. This allows for timely and efficient determination of the necessary maintenance actions when an anomaly is detected in the core network. Furthermore, before the target network element executes the target maintenance action, this application also evaluates the feasibility of the target network element executing the target maintenance action. Only when the feasibility result indicates that the target network element is suitable for executing the target maintenance action will the target network element be controlled to execute the target maintenance action. This reduces anomaly situations such as failure to repair anomalies or degradation of service quality caused by the target network element being unsuitable for executing the target maintenance action. Consequently, it enables more timely and reasonable implementation of maintenance actions when anomalies exist in the core network, improving the timeliness and reliability of core network maintenance.

[0102] In this application, there are no restrictions on the specific implementation of the reinforcement learning model obtained through reinforcement learning. The following is a brief explanation using one implementation as an example:

[0103] This application can obtain multiple sample state data as training samples. This sample state data consists of the state data of network elements in the core network. The state data of each network element may include: the network fault level corresponding to the network element, the value of at least one performance indicator, and the service experience difference value. Based on the sample state data, after determining and executing maintenance actions, this application can use the reward function shown in the following formula to determine the reward value during the reinforcement learning process. :

[0104] ;

[0105] Among them, in Formula 1 As weight, The value can be any one of 1, 2, 3, and 4. Settings and adjustments can be made according to actual operational goals. When the values ​​of are different, The value of can also be different.

[0106] As can be seen, the reward value is related to the following four core metrics:

[0107] This indicates whether the network anomaly has been eliminated or mitigated. If the network anomaly is eliminated or mitigated after the operation and maintenance actions are performed, a positive reward will be generated.

[0108] This indicates whether the user's service quality has improved. If the user's service quality improves after the implementation of maintenance actions, a positive reward will be generated.

[0109] This indicates whether a KPI has been optimized (e.g., reduced latency and increased throughput are both examples of KPI optimization). If an KPI is optimized after performing maintenance actions, a positive reward will be generated.

[0110] This indicates whether a new fault or resource congestion has been triggered. If a new fault or resource congestion is triggered after the operation and maintenance action is performed, a negative reward will be generated if the result is true.

[0111] For example, the reward settings in this scheme could be as follows:

[0112] Successfully eliminated poor user service quality without causing new problems: Reward +10;

[0113] Ineffective maintenance actions or failure to improve KPIs: Reward 0;

[0114] Maintenance actions that lead to new faults or worsening of KPIs: Reward -10 or lower.

[0115] During reinforcement learning, the state of this application may include the state data of the network elements mentioned above, and the Q-value function is updated using the Bellman equation, which will not be elaborated further.

[0116] There are several possible implementations for detecting anomalies in the core network in this application.

[0117] In one possible implementation, to promptly detect anomalies in the core network, this application can obtain at least one evaluation indicator data corresponding to the core network. This evaluation indicator data can be data used to evaluate the operational status of the core network. For example, the evaluation indicator data may include the value of at least one performance indicator of each network element in the core network, as well as part or all of the service status information of the services provided by each network element in the core network, without limitation.

[0118] Accordingly, this application can determine the evaluation result of the core network based on the at least one evaluation index data using a target model. This evaluation result is used to characterize whether the core network has any anomalies. Based on this, in response to the evaluation result indicating that the core network has anomalies, this application will identify at least one target network element in the core network that has anomalies.

[0119] The target model can be any trained machine learning model or other artificial intelligence model, without any restrictions.

[0120] In this application, when an anomaly occurs in the core network, at least one target network element in the core network can be identified based on the metric values ​​of at least one performance indicator corresponding to each network element, alarm data, and operational log data. The performance indicators of the network elements can be found in the previous section on the performance indicators of target network elements, and will not be repeated here. Alarm data refers to alarm data generated during the operation of network elements to indicate network anomalies or faults.

[0121] For each network element, one or more of the following can indicate whether the network element is abnormal: the value of at least one performance indicator, alarm data, and operation log data. For example, if the value of a network element's performance indicator exceeds the set range corresponding to that performance indicator, or if alarm data indicates that the network element is abnormal, or if operation log data shows that the network element has network anomalies or other abnormal situations affecting the operation of the core network, then it is indicated that the network element is abnormal.

[0122] It is understandable that, considering the possibility that the core network may still operate normally even if a single network element is abnormal, for example, a certain type of network element in the core network has a corresponding backup network element, and even if that type of network element is abnormal, service can still be provided through the backup network element. Based on this, this application will only identify the abnormal network element after determining that the core network is abnormal. This can reduce the amount of data processing and locate the network element that caused the core network abnormality.

[0123] The specific implementation process for determining the evaluation results of the core network will also differ depending on the evaluation indicator data. The following explanation will be based on two different scenarios:

[0124] In the first possible scenario, obtaining at least one evaluation metric data corresponding to the core network can be achieved by obtaining the current network element characteristic data of each network element in the core network. This network element characteristic data includes at least one of the following: the metric value of at least one performance indicator of the network element, alarm data, operation log data, and configuration data. The configuration data can be the load balancing strategy and hardware / software parameter configuration information used in the network element, without specific limitations.

[0125] Accordingly, determining the evaluation results of the core network using the target model can include: based on the network element characteristic data of each network element, using a first prediction model to determine the anomaly prediction results of the core network. For example, after converting the network element characteristic data of each network element into feature vectors, the converted feature vectors are input into the first prediction model to obtain the anomaly prediction results output by the first prediction model.

[0126] The anomaly prediction result includes an anomaly probability and an anomaly level. For example, the anomaly level can be divided into mild, moderate, and severe. Of course, other level classifications can be used as needed, and there are no restrictions on this. In this application, if the anomaly probability exceeds a set probability value and the anomaly level exceeds a set level, then an anomaly is determined to exist in the core network. The setting of an anomaly level exceeding the set level can be configured as needed; for example, the anomaly level exceeding the set level could be either mild or moderate.

[0127] The first prediction model can be a trained machine learning model or other artificial intelligence model, without any restrictions.

[0128] For example, the first prediction model may include a Long Short-Term Memory (LSTM) network model and a fully connected layer. The feature vectors derived from the feature data of each network element can be input into the LSTM model, which learns key information from the feature data (such as whether an error or fault has occurred). The vector output by the LSTM model can be input into the fully connected layer, which outputs the anomaly probability and anomaly level of the core network.

[0129] In this application, the first prediction model can be trained using a supervised training method, and the loss function used to train the first prediction model can be the cross-entropy loss function. The specific training process is not restricted.

[0130] For details on identifying the abnormal target network element in this first possible scenario, please refer to the relevant introduction above, which will not be repeated here.

[0131] In the second possible scenario, obtaining at least one evaluation metric data corresponding to the core network may include: obtaining service metric data for at least one user currently managed by the core network. Here, "user" refers to a user accessing the core network through a user equipment or a user using services provided by the core network. The user's service metric data is used to characterize the services used by the user and the service quality information of those services. For example, the user's service metric data may include: the user's user identifier and the service identifier and experience metric data corresponding to the services used by the user.

[0132] Among them, the user identifier is a unique identifier for the user, which is used to track the user's business activities.

[0133] The service identifier is used to identify the type of service a user is currently using. For example, the service type can be divided into voice service, video service, or data transmission service.

[0134] Service experience metrics can be indicators that reflect service quality, such as network latency, packet loss rate, and throughput.

[0135] Accordingly, for each user, the service quality result of the core network can be determined using a second prediction model based on the user's service indicator data. This service quality result characterizes whether the service quality corresponding to the service used by the user is acceptable. Therefore, if the service quality result indicates that the user's service quality is unacceptable, it can be determined that there is an anomaly in the core network.

[0136] The second prediction model can be any classifier or other machine learning model, without any restrictions.

[0137] For example, the second prediction model can be a Support Vector Machine (SVM) model. Based on this, historical service indicator data can be used as training data samples, and the SVM model can be trained based on the sample service quality results labeled with the historical service indicator data. The historical service indicator data can include: user identifiers of historical users, historical service identifiers, and historical service experience indicators, etc. Historical users are users who historically accessed the core network through user equipment, and historical service identifiers are the service identifiers used by those historical users within the core network. The sample quality result labeled for each data sample is either "quality acceptable" or "quality unacceptable".

[0138] In this second possible scenario, the aforementioned method can be used to determine the target network element with anomalies. In an alternative approach, to more efficiently determine the target network element with anomalies, for each user, in response to the user's service quality result indicating that the service quality corresponding to the service used by the user is unqualified, the network element responsible for managing that user is determined as the target network element with anomalies based on the user's user identifier, thus obtaining at least one target network element with anomalies.

[0139] Understandably, in practical applications, the above two possible scenarios can be combined to determine whether there is an anomaly in the core network. For example, if the predicted anomaly probability exceeds a set probability value, the anomaly level exceeds a set level, and the service quality result indicates that the user's service quality is unqualified, then it is determined that there is an anomaly in the core network.

[0140] In any of the above embodiments of this application, there may be a variety of ways to determine the feasibility of the target network element performing the target operation and maintenance action.

[0141] In one possible implementation, to reliably and accurately determine the feasibility result, this application can first determine the operational impact data that affects the execution result of the target operational action. Specifically, based on the target operational action, the operational impact data of the target network element can be determined. Accordingly, based on the operational impact data, the feasibility result of the target network element executing the target operational action can be determined.

[0142] The operation and maintenance impact data can be any parameter data that can affect the execution result of the target network element's target operation and maintenance actions, without any specific restrictions.

[0143] For example, the operational impact data of the target network element may include at least one of the following:

[0144] Resource requirements needed by the target network element to perform the target operation and maintenance action;

[0145] The historical operation and maintenance results of the target network element when performing the target's operation and maintenance actions;

[0146] Information on the service impact on at least one target user that can be affected by the target network element performing the target operation and maintenance action.

[0147] There are several ways to determine the feasibility of the operation and maintenance impact data of the target network element. For example, based on the operation and maintenance impact data, a feasibility assessment model can be used to determine the feasibility of the target network element performing the target operation and maintenance action.

[0148] In particular, the method for determining the feasibility outcome will differ depending on the operational impact data. The following section will combine... Figure 2 Please provide an explanation.

[0149] like Figure 2 This illustrates another flowchart of the intelligent operation and maintenance control method for the core network provided in this application. The method in this embodiment may include:

[0150] S201, obtain at least one evaluation index data corresponding to the core network.

[0151] S202, Based on the data of at least one evaluation index, the evaluation result of the core network is determined using the target model.

[0152] The evaluation results are used to characterize whether there are any anomalies in the core network.

[0153] S203, in response to the evaluation result indicating an anomaly in the core network, identify at least one target network element in the core network that has an anomaly.

[0154] In this embodiment, for ease of understanding, an implementation method for detecting anomalies in the core network is used as an example. However, the other methods mentioned above for determining anomalies in the core network are also applicable to this embodiment.

[0155] S204. For each target network element, obtain the target status data corresponding to that target network element.

[0156] The target status data includes at least one of the following: the network fault level corresponding to the target network element, the value of at least one performance indicator, and the service experience difference value of the service. For details, please refer to the previous related introduction.

[0157] S205. Based on the target state data, the target operation and maintenance actions to be performed by the target network element are determined using a trained reinforcement learning model.

[0158] The reinforcement learning model includes the mapping relationship between different state data and operation and maintenance actions.

[0159] The above steps can be found in the relevant descriptions of the previous embodiments, and will not be repeated here.

[0160] S206, Obtain the resource requirements information needed for the target network element to perform the target operation and maintenance action.

[0161] Among them, the resource demand information is used to characterize the resource demand (such as the amount of resource required) of at least one resource required by the target network element to perform the target operation and maintenance action.

[0162] Since the resource requirements for different maintenance actions are relatively fixed, a resource requirement list can be pre-configured. This list can store the correspondence between different maintenance actions and resource requirements. Based on this, the resource requirement list can be queried to determine the resource requirements corresponding to the target maintenance action. This resource requirement list can be deployed on the core network side or in electronic devices, without specific restrictions.

[0163] S207, obtain the current resource status information of the target network element.

[0164] Among them, the current resource status information of the target network element represents the current available resources of the target network element.

[0165] For example, the current resource status information of the target network element may include at least one of the following:

[0166] The number of used and available Physical Resource Blocks (PRBs) in the cell corresponding to the target network element;

[0167] Network throughput of the cell corresponding to the target network element;

[0168] Network status indicators of the cell corresponding to the target network element.

[0169] Among them, the resource utilization rate of a cell can be reflected by detecting the number of PRBs used and available in the cell corresponding to the target network element.

[0170] The network throughput of the target network element in the corresponding cell can reflect the actual load and performance of the cell.

[0171] Among them, the network status indicators of the target network element corresponding to the cell are used to characterize the network status of the cell and can reflect whether there are network anomalies in the cell. For example, network status indicators may include, but are not limited to, network latency, packet loss rate, and number of user equipment accesses.

[0172] In practical applications, the resource status information corresponding to different network elements can be obtained in real time through the telecommunications business operation support system or specific network monitoring tools and stored in a designated database. When the resource status information of a certain network element is needed, the latest resource status information of that network element can be obtained from the designated database.

[0173] Of course, after obtaining resource status information, data cleaning and preprocessing can be used to remove noise and ensure the accuracy of the obtained resource status information.

[0174] S208. Based on the resource status information, use a trained third prediction model to predict the resource usage of the target network element after the current moment.

[0175] This resource usage information is a prediction of the target network element's resource usage over a specified time period after the current moment, based on the target network element's current resource status information. For example, this resource usage information could be the target network element's resource usage over the next 6 hours after the current moment.

[0176] Resource usage may include, but is not limited to, PRB utilization rate and available bandwidth of the target network element, as well as the usage of resources related to the services provided and maintenance actions performed by the target network element.

[0177] The type of the third prediction model can be varied and is not restricted. For example, the third prediction model can be a machine learning model such as an LSTM model, a Gated Recurrent Unit (GRU) model, or an Extreme Gradient Boosting (XGBoost) model.

[0178] The third prediction model can be trained using historical resource status information, with the goal of predicting the resource usage indicated by that historical status information. The specific training process for this third prediction model is not restricted.

[0179] S209, Based on the resource usage and resource demand information, determine the first feasible result for the target network element to perform the target operation and maintenance action.

[0180] For ease of distinction, the feasibility result determined based on resource usage and resource demand information is referred to as the first feasibility result.

[0181] The first feasibility result is used to characterize whether the target network element is suitable for performing the target operation and maintenance action. For example, the first feasibility result can characterize whether the resources of the target network element can support the successful operation and maintenance of the target operation and maintenance action.

[0182] Based on resource usage and resource demand information, it can be determined whether the target network element can provide the resources required to perform the target operation and maintenance action after the current time (such as within a set time period after the current time). For example, if the available resources of the target network element, as indicated by the resource usage information, are greater than the resource demand, as indicated by the resource demand information, then the target network element can provide the resources required to perform the target operation and maintenance action after the current time (such as within a set time period after the current time).

[0183] Correspondingly, if it is determined based on resource usage and resource demand information that the target network element cannot meet the resources required to perform the target operation and maintenance action after the current time, it means that the target network element cannot complete the execution of the target operation and maintenance action after the current time. Therefore, the first feasibility result indicates that the target network element is not suitable for performing the target operation and maintenance action.

[0184] S210, obtain the current scene status information of the target network element.

[0185] The scenario status information is used to characterize at least the current network and resource status of the target network element. For example, the current network status of the target network element could be the network bandwidth, throughput, and network latency of the cell corresponding to the target network element. The resource status of the target network element could be the number of PRBs currently available to the target network element and the PRB utilization rate.

[0186] Of course, the current scenario status information of the target network element may also include other performance indicators of the target network element, operating environment, and current time, etc., which can characterize the current scenario status of the target network element, without any restrictions.

[0187] S211, Obtain at least one historical operation and maintenance data of the target network element.

[0188] Among them, historical operation and maintenance data refers to the relevant data of the target network element's operation and maintenance actions performed before the current moment.

[0189] In this application, the historical operation and maintenance data includes: historical operation and maintenance actions performed by the target network element, historical scenario status information of the target network element when performing the historical operation and maintenance actions, and the operation and maintenance execution results of the target network element when performing the historical operation and maintenance actions. The operation and maintenance execution results are used to characterize whether the historical operation and maintenance actions were successfully executed.

[0190] For ease of distinction, the maintenance actions historically performed by the target network element are referred to as historical maintenance actions, and the scenario status information of the target network element when performing these historical maintenance actions is referred to as historical scenario status information. The types of information included in this historical scenario status information can be found in the previous introduction to scenario status information, and will not be repeated here.

[0191] Of course, the historical operation and maintenance data may also include information such as the historical time when the target network element performed the historical operation and maintenance action before the current moment (i.e., in the past).

[0192] S212, Based on the current scenario status information of the target network element, determine at least one target historical operation and maintenance data from at least one historical operation and maintenance data.

[0193] Among them, the historical operation and maintenance actions in the target's historical operation and maintenance data are the same as the target's operation and maintenance actions, and the historical scenario status information in the target's historical operation and maintenance data is similar to the current scenario status information of the target network element.

[0194] The similarity between historical scene state information and the current scene state information of the target network element can be defined as the similarity (matching degree) between the historical scene state information and the current scene state information of the target network element exceeding a set threshold. For example, the vector similarity between the first vector feature converted from the historical scene state information and the second vector feature converted from the current scene state information of the target network element can be determined as the similarity between the historical scene state information and the current scene state information of the target network element.

[0195] S213, based on the operation and maintenance execution results in the historical operation and maintenance data of the at least one target, determine the second feasibility result of the target network element performing the target operation and maintenance action.

[0196] In this embodiment, the at least one target historical operation and maintenance data is used as the operation and maintenance execution result of the target network element in the past when the target operation and maintenance action was performed.

[0197] The determination of the second feasibility result based on the operation and maintenance execution results in the historical operation and maintenance data of at least one target can be achieved in several ways. For example, the second feasibility result can be determined based on the percentage of successful operation and maintenance execution results in the historical operation and maintenance data of at least one target. For instance, if the percentage exceeds a set threshold, the second feasibility result indicates that the target network element is suitable for performing the target operation and maintenance action; if the percentage does not exceed the set threshold, the second feasibility result indicates that the target network element is not suitable for performing the target operation and maintenance action.

[0198] In one possible implementation, in order to more reliably determine the feasibility of the target network element performing the target operation and maintenance action and improve the accuracy of the determined second feasibility result, the success rate of the target network element performing the target operation and maintenance action can be predicted by using a trained fourth prediction model based on the scenario status information of the target network element and the historical scenario status information and operation and maintenance execution results in the historical operation and maintenance data of each target.

[0199] The execution success rate is used to characterize the probability that the target network element can successfully execute the target operation and maintenance action. For example, if the execution success rate exceeds a first preset value, it can indicate that the target network element is suitable for executing the target operation and maintenance action.

[0200] The fourth prediction model can be a neural network model or other machine learning model or other artificial intelligence model, without any restrictions.

[0201] S214, determine at least one associated user that can be affected by the target network element performing the target operation and maintenance action.

[0202] Among them, the associated users are users of the target services provided by the target network element, and the target services are the services that can be affected by the target network element performing the target operation and maintenance actions. The target service belongs to at least one type of service provided by the target network element.

[0203] S215, obtain the user level, user behavior data and business indicator information of the associated user.

[0204] This business metric information is used to characterize the performance of the target service currently being used by the associated user. For example, the business metric information may include, but is not limited to, collected user experience metrics for the target service and traffic data for that target service.

[0205] In this application, user level can represent user priority; the higher the user level, the greater the impact of the target business on the user.

[0206] S216, based on the user level, user behavior data and business indicator information of the associated users, determine the third feasibility result for the target network element to perform the target operation and maintenance action.

[0207] In this embodiment, for ease of distinction, the feasibility result determined based on the resource usage and resource demand information is referred to as the first feasibility result, the feasibility result determined based on the operation and maintenance execution results in the target's historical operation and maintenance data is referred to as the second feasibility result, and the feasibility result determined based on the user level, user behavior data, and business indicator information of the associated user is referred to as the third feasibility result.

[0208] In this application, it is determined that the specific implementation of the third feasible result may also have multiple possible implementations, and no limitation is imposed on this.

[0209] For example, in one possible implementation, if the at least one associated user does not include a target user with a user level higher than the set level, the third feasibility result indicates that the target network element is suitable for performing the target operation and maintenance action.

[0210] If at least one of the associated users includes at least one target user whose user level is higher than the set level, based on the target user's user behavior data and business indicator information, the trained fifth prediction model is used to predict the business quality score of the target service after the target network element performs the target operation and maintenance action.

[0211] Among them, if the service quality score exceeds the second set value, it indicates that the target network element is suitable for performing the target operation and maintenance action.

[0212] If the service quality score does not exceed the second set value, it means that the target network element performing the target maintenance action will have a significant impact on the target service used by the target user. Therefore, the target network element is not suitable for performing the target maintenance action.

[0213] The first and second settings can be set according to actual needs without restriction.

[0214] The fifth prediction model can have various structure and type, and can be any machine learning model or other artificial intelligence model without restriction.

[0215] S217 If at least one of the first feasibility result, the second feasibility result, and the third feasibility result indicates that the target network element is suitable for performing the target operation and maintenance action, control the target network element to perform the target operation and maintenance action.

[0216] It is understood that any one of the above-mentioned first feasibility result, second feasibility result, and third feasibility result can be obtained according to actual needs in this application. Therefore, this application can perform only the relevant operations to obtain the first feasibility result, second feasibility result, or third feasibility result as needed. Accordingly, if it is determined that the target network element is suitable for the target operation and maintenance action based on the first feasibility result, second feasibility result, or third feasibility result, then the target operation and maintenance action can be executed.

[0217] Furthermore, in order to improve the reliability of the target network element in executing the target operation and maintenance action, and reduce the resource consumption caused by the target network element's failure to successfully execute the target operation and maintenance action or the impact on the service use of some high-priority users, this application may also include: the target network element will only be controlled to execute the target operation and maintenance action when the first feasibility result, the second feasibility result and the third feasibility result all indicate that the target network element is suitable for executing the target operation and maintenance action.

[0218] For example, if the second feasibility result includes the success rate of the target network element executing the target maintenance action, and the third feasibility result includes the service quality score of the target service after the target network element executes the target maintenance action, then the first, second, and third feasibility results all indicate that a network element is suitable for executing the target maintenance action if it simultaneously meets the following conditions:

[0219] The first feasibility result indicates that the target network element is suitable for performing the target operation and maintenance actions;

[0220] The success rate of the target network element executing the target operation and maintenance action exceeds the first set value;

[0221] At least one of the associated users does not include the target user whose user level is higher than the set level; or, at least one of the associated users includes the target user and the business quality score exceeds the second set value.

[0222] Understandably, if the feasibility results indicate that the target network element is not suitable for performing the target maintenance action, this application may choose not to perform the target maintenance action but instead output a maintenance reminder to remind the user to promptly address any anomalies present in the target network element.

[0223] Of course, in order to improve the operational efficiency of the core network, if the feasibility results indicate that the target network element is not suitable for performing the target operational action, the intelligent programs such as the intelligent agents of electronic devices can also use a global optimization mechanism to select a solution with the least impact on resources, ensuring the stable operation of the core network and minimizing the impact on user services.

[0224] The global optimization mechanism can be:

[0225] First, the resource status information of the target network element is collected periodically, and the resource demand trend information of the target network element in the future target time interval is predicted based on the resource status information.

[0226] Secondly, based on resource demand trend information, digital twin technology or a state transition simulator based on reinforcement learning is used to simulate the quantitative indicators of operation and maintenance of target network elements performing target operation and maintenance actions.

[0227] Among them, the quantitative indicators of operation and maintenance may include, but are not limited to: changes in the balance of resource utilization, fluctuations in performance indicators (such as end-to-end latency, packet loss rate, and throughput), the probability of service experience degradation for high-priority users and the risk value of potential service level agreement (SLA) violations, and the fluctuation range of overall network KPIs.

[0228] Then, based on the quantitative indicators of the target operation and maintenance actions, the adjustment actions of the target operation and maintenance actions are determined using a model with reinforcement learning as the core, and the adjusted target operation and maintenance actions are obtained.

[0229] In addition to quantitative operational indicators, the model can also be used to comprehensively determine adjustment actions by combining feedback on the execution effects of historical operational actions for the target.

[0230] The action space for determining the adjustment action includes, but is not limited to: directly executing the original action, delaying execution, adjusting action parameters (reducing the scope of influence, reducing the adjustment magnitude), and replacing with alternative actions (such as changing from load migration to local expansion).

[0231] Finally, the feasibility of the adjusted target operation and maintenance actions can be evaluated using the methods described above. If feasible, the adjusted target operation and maintenance actions can be executed.

[0232] In this application, the electronic device can be located outside the core network, and can collect relevant data of network elements in the core network by means of some devices in the core network. For example, Figure 3 A framework example diagram of an application scenario to which the solution of this application is applicable is shown.

[0233] In a 5G core network based on the 3rd Generation Partnership Project (3GPP), since 3GPP introduces NWDAF (Network Data Analytics Function) network elements, in this application, electronic devices can utilize NWDAF network elements in the core network to collect the network element-related data required for this embodiment. This allows the electronic devices to perform intelligent operation and maintenance control of the core network based on the obtained data, as described above, and will not be repeated here. Of course, the core network may also include Session Management Function (SMF) network elements and Access and Mobility Management Function (AMF) network elements, etc., without limitation.

[0234] Corresponding to the intelligent operation and maintenance control method for a core network provided in this application, this application also provides an intelligent operation and maintenance control device for a core network. For example... Figure 4 This paper illustrates a schematic diagram of the architecture of an intelligent operation and maintenance control device for the core network provided in this application. The device includes:

[0235] Network element determination unit 401 is used to determine at least one target network element in the core network that has an anomaly in response to detecting an anomaly in the core network;

[0236] The status acquisition unit 402 is used to acquire target status data corresponding to each target network element. The target status data includes at least one of the following: the network fault level corresponding to the target network element, the index value of at least one performance indicator, and the service experience difference value of the service.

[0237] Action determination unit 403 is used to determine the appropriate target operation and maintenance action for the target network element based on the target state data and using a trained reinforcement learning model. The reinforcement learning model includes the mapping relationship between different state data and operation and maintenance actions.

[0238] The feasibility determination unit 404 is used to determine the feasibility result of the target network element performing the target operation and maintenance action. The feasibility result is used to indicate whether the target network element is suitable for performing the target operation and maintenance action.

[0239] Action execution unit 405 is used to control the target network element to perform the target operation and maintenance action in response to the feasibility result indicating that the target network element is suitable for performing the target operation and maintenance action.

[0240] In one possible implementation, the network element determination unit includes:

[0241] The indicator acquisition subunit is used to obtain data for at least one evaluation indicator corresponding to the core network.

[0242] The evaluation determines the sub-unit, which is used to determine the evaluation result of the core network based on the data of at least one evaluation index and using the target model. The evaluation result is used to characterize whether there are any anomalies in the core network.

[0243] The network element determination subunit is used to determine at least one target network element in the core network that has an anomaly in response to the evaluation result indicating that there is an anomaly in the core network.

[0244] In yet another possible implementation, the indicator obtains sub-units, including:

[0245] The network element feature acquisition subunit is used to obtain the current network element feature data of each network element in the core network. The network element feature data includes at least one of the following: the indicator value of at least one performance indicator of the network element, alarm data, operation log data, and configuration data.

[0246] The business metrics acquisition subunit is used to obtain business metrics data of at least one user currently managed by the core network. The user's business metrics data includes: the user's user identifier and the business identifier and experience metrics data corresponding to the services used by the user.

[0247] The evaluation identifies sub-units, including at least one of the following:

[0248] An anomaly prediction subunit is used to determine the anomaly prediction result of the core network based on the feature data of the network element and using the first prediction model. The anomaly prediction result includes: anomaly probability and anomaly level.

[0249] The quality determination subunit is used to determine the service quality result of the core network for each user based on the user's service indicator data using a second prediction model. The service quality result is used to characterize whether the service quality corresponding to the service used by the user is qualified.

[0250] The evaluation results indicate that the core network has at least one of the following anomalies:

[0251] The probability of this anomaly exceeds the set probability value, and the level of this anomaly exceeds the set level.

[0252] This service quality result indicates that the service used by this user is of substandard quality.

[0253] In another possible implementation, the network element determines the sub-unit, including at least one of the following:

[0254] The first network element determination subunit is used to respond to the evaluation result indicating that there is an anomaly in the core network, and to determine at least one target network element in the core network that has an anomaly based on the index value of at least one performance indicator corresponding to each network element in the core network, alarm data and operation log data.

[0255] The second network element determination subunit is used to determine at least one target network element that is abnormal in response to the evaluation result indicating that there is an anomaly in the core network, and in response to the service quality result indicating that the service quality corresponding to the service used by the user is unqualified, based on the user's user identifier.

[0256] In another possible implementation, the feasibility determination unit includes:

[0257] The impact determination subunit is used to determine the operation and maintenance impact data of the target network element based on the target operation and maintenance action;

[0258] The feasibility determination subunit is used to determine the feasibility of performing the target operation and maintenance action on the target network element based on the operation and maintenance impact data.

[0259] The operational impact data includes at least one of the following:

[0260] The resource requirements needed for the target network element to perform the target operation and maintenance action;

[0261] The historical operation and maintenance results of the target network element when performing the target's operation and maintenance actions;

[0262] Information on the service impact on at least one target user that can be affected by the target network element performing the target operation and maintenance action.

[0263] In another possible implementation, where the operational impact data determined by the impact determination subunit includes the resource requirement information, the feasibility determination subunit includes:

[0264] The resource acquisition subunit is used to obtain the current resource status information of the target network element;

[0265] The resource prediction subunit is used to predict the resource usage of the target network element after the current moment based on the resource status information and using a trained third prediction model.

[0266] The first result determination subunit is used to determine the feasibility of the target network element performing the target operation and maintenance action based on the resource usage and resource demand information.

[0267] In another possible implementation, the impact determination subunit includes: a first impact determination subunit, used to determine the operation and maintenance execution results of the target network element's historical execution of the target operation and maintenance action in the following manner:

[0268] Obtain the current scene status information of the target network element, which is used to characterize at least the current network status and resource status of the target network element;

[0269] Obtain at least one historical operation and maintenance data of the target network element. The historical operation and maintenance data includes: historical operation and maintenance actions performed by the target network element, historical scenario status information when the target network element performed the historical operation and maintenance actions, and operation and maintenance execution results of the historical operation and maintenance actions. The operation and maintenance execution results are used to characterize whether the historical operation and maintenance actions were successfully executed.

[0270] From the at least one historical operation and maintenance data, at least one target historical operation and maintenance data is determined, wherein the historical operation and maintenance actions in the target historical operation and maintenance data are the same as the target operation and maintenance actions and the historical scenario status information is similar to the scenario status information.

[0271] The operation and maintenance execution results in the historical operation and maintenance data of each target are determined as the operation and maintenance execution results of the target network element in history when the target operation and maintenance actions were performed.

[0272] In yet another possible implementation, the feasibility determination sub-unit includes:

[0273] The second result determination subunit is used to predict the success rate of the target network element performing the target operation and maintenance action based on the scenario state information and the historical scenario state information and operation and maintenance execution results in the historical operation and maintenance data of the target, using a trained fourth prediction model. If the success rate exceeds the first set value, it indicates that the target network element is suitable for performing the target operation and maintenance action.

[0274] In another possible implementation, the impact determination subunit includes: a second impact determination subunit, configured to determine the service impact information of at least one target user that can be affected by the target network element performing the target operation and maintenance action in the following manner:

[0275] Identify at least one associated user that will be affected by the target network element performing the target maintenance action;

[0276] Obtain the user level, user behavior data and business indicator information of the associated user. The business indicator information is used to characterize the business performance of the target service currently used by the associated user. The target service belongs to the service provided by the target network element.

[0277] This feasibility determination sub-result includes:

[0278] The third result determination subunit is used to determine that the target network element is suitable for performing the target operation and maintenance action in response to the fact that the at least one associated user does not include a target user with a user level higher than a set level.

[0279] The fourth result determination subunit is used to respond to the fact that at least one of the at least associated users includes at least one target user, and based on the user behavior data and business indicator information of the target user, use a trained fifth prediction model to predict the service quality score of the target service after the target network element performs the target operation and maintenance action, wherein the service quality score exceeds a second set value, indicating that the target network element is suitable to perform the target operation and maintenance action.

[0280] Furthermore, this application also provides an electronic device in its embodiments. For example... Figure 5 As shown, a schematic diagram of the composition structure of the electronic device is presented, which includes at least a processor 501 and a memory 502.

[0281] The processor 501 is used to execute the intelligent operation and maintenance control method of the core network as described in any of the above embodiments;

[0282] The memory 502 is used to store the programs required for the processor to perform operations.

[0283] It is understood that the electronic device may also include a display unit 503 and an input unit 504.

[0284] Of course, the electronic device can also have more than Figure 5 There are no restrictions on the number of components, whether more or fewer.

[0285] This application also provides a computer program product including computer-readable instructions, which, when executed on an electronic device, cause the electronic device to implement any of the intelligent operation and maintenance control methods for the core network provided in this application.

[0286] This application also provides a computer-readable storage medium that carries one or more computer programs. When the one or more computer programs are executed by an electronic device, the electronic device can implement any of the intelligent operation and maintenance control methods for the core network provided in this application.

[0287] It should also be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. In addition, in the device embodiment drawings provided in this application, the connection relationship between modules indicates that they have a communication connection, which can be implemented as one or more communication buses or signal lines.

[0288] Through the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware, or it can be implemented by special-purpose hardware including application-specific integrated circuits, special-purpose CPUs, special-purpose memory, special-purpose components, etc. Generally, any function performed by a computer program can be easily implemented by corresponding hardware, and the specific hardware structure used to implement the same function can also be diverse, such as analog circuits, digital circuits, or special-purpose circuits. However, for this application, software program implementation is more often the preferred implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a readable storage medium, such as a computer floppy disk, USB flash drive, mobile hard disk, ROM, RAM, magnetic disk, or optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, training equipment, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0289] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product.

[0290] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, training device, or data center to another website, computer, training device, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can store or a data storage device such as a training device or data center that integrates one or more available media. The available media may be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid-state drives (SSDs)).

Claims

1. A method for intelligent operation and maintenance control of a core network, characterized in that, include: In response to the detection of an anomaly in the core network, at least one target network element in the core network is identified as having an anomaly. For each target network element, obtain the target status data corresponding to the target network element. The target status data includes at least one of the following: the network fault level corresponding to the target network element, the index value of at least one performance indicator, and the service experience difference value of the service corresponding to the target network element. Based on the target state data, the target operation and maintenance actions required for the target network element are determined using a trained reinforcement learning model. The reinforcement learning model includes the mapping relationship between different state data and operation and maintenance actions. The feasibility result of the target network element performing the target operation and maintenance action is determined, and the feasibility result is used to indicate whether the target network element is suitable for performing the target operation and maintenance action; In response to the feasibility results indicating that the target network element is suitable for performing the target operation and maintenance action, the target network element is controlled to perform the target operation and maintenance action.

2. The intelligent operation and maintenance control method for the core network according to claim 1, characterized in that, The step of responding to the detection of an anomaly in the core network and determining at least one target network element in the core network that is anomaly includes: Obtain data for at least one evaluation indicator corresponding to the core network; Based on the data of at least one evaluation index, the evaluation result of the core network is determined using the target model, and the evaluation result is used to characterize whether there is an anomaly in the core network; In response to the evaluation result indicating an anomaly in the core network, at least one target network element in the core network is identified as having an anomaly.

3. The intelligent operation and maintenance control method for the core network according to claim 2, characterized in that, The acquisition of at least one evaluation index data corresponding to the core network includes: Obtain the current network element characteristic data of each network element in the core network. The network element characteristic data includes at least one of the following: the index value of at least one performance indicator of the network element, alarm data, operation log data, and configuration data. Obtain service indicator data for at least one user currently managed by the core network, wherein the user's service indicator data includes: the user's user identifier and the service identifier and experience indicator data corresponding to the service used by the user; The determination of the evaluation result of the core network based on the at least one evaluation index data using the target model includes at least one of the following: Based on the network element feature data, the anomaly prediction result of the core network is determined using the first prediction model. The anomaly prediction result includes: anomaly probability and anomaly level. For each user, based on the user's service indicator data, the second prediction model is used to determine the service quality result of the core network. The service quality result is used to characterize whether the service quality corresponding to the service used by the user is qualified. The evaluation results indicating anomalies in the core network include at least one of the following: The anomaly probability exceeds a set probability value and the anomaly level exceeds a set level; The service quality result indicates that the service used by the user is of substandard quality.

4. The intelligent operation and maintenance control method for the core network according to claim 3, characterized in that, Identifying at least one target network element in the core network that exhibits an anomaly, including at least one of the following: Based on the index values, alarm data, and operation log data of at least one performance indicator corresponding to each network element in the core network, at least one target network element in the core network with an anomaly is identified. In response to the user's service quality result indicating that the service quality corresponding to the service used by the user is unqualified, based on the user's user identifier, at least one network element responsible for managing the user is identified as at least one target network element with an anomaly.

5. The intelligent operation and maintenance control method for the core network according to claim 1, characterized in that, The feasibility results for determining the target network element to perform the target operation and maintenance action include: Based on the target operation and maintenance action, determine the operation and maintenance impact data of the target network element; Based on the aforementioned operational impact data, the feasibility of the target network element performing the target operational action is determined. The operational impact data includes at least one of the following: The resource requirements information required for the target network element to perform the target operation and maintenance action; The target network element has historically executed the target maintenance actions, and the maintenance execution results are as follows: Information on the service impact on at least one target user that can be affected by the target network element performing the target operation and maintenance action.

6. The intelligent operation and maintenance control method for the core network according to claim 5, characterized in that, When the operation and maintenance impact data includes the resource demand information, the step of determining the feasibility of the target network element performing the target operation and maintenance action based on the operation and maintenance impact data includes: Obtain the current resource status information of the target network element; Based on the resource status information, the resource usage of the target network element after the current moment is predicted using a trained third prediction model; Based on the resource usage and resource demand information, the feasibility of the target network element performing the target operation and maintenance action is determined.

7. The intelligent operation and maintenance control method for the core network according to claim 5, characterized in that, The operation and maintenance execution results of the target network element in the past when the target operation and maintenance actions were performed are determined in the following way: Obtain the current scene status information of the target network element, wherein the scene status information is at least used to characterize the current network status and resource status of the target network element; Obtain at least one historical operation and maintenance data of the target network element. The historical operation and maintenance data includes: historical operation and maintenance actions performed by the target network element, historical scenario status information when the target network element performed the historical operation and maintenance actions, and operation and maintenance execution results of the historical operation and maintenance actions. The operation and maintenance execution results are used to characterize whether the historical operation and maintenance actions were successfully executed. From the at least one historical operation and maintenance data, at least one target historical operation and maintenance data is determined, wherein the historical operation and maintenance actions in the target historical operation and maintenance data are the same as the target operation and maintenance actions and the historical scenario status information is similar to the scenario status information; The operation and maintenance execution results in the historical operation and maintenance data of each target are determined as the operation and maintenance execution results of the target network element in the history of performing the target operation and maintenance actions.

8. The intelligent operation and maintenance control method for the core network according to claim 7, characterized in that, The feasibility result of determining the target network element to perform the target operation and maintenance action based on the operation and maintenance impact data includes: Based on the scenario state information and the historical scenario state information and operation and maintenance execution results in the target historical operation and maintenance data, the training fourth prediction model is used to predict the success rate of the target network element executing the target operation and maintenance action. The success rate exceeding a first set value indicates that the target network element is suitable for executing the target operation and maintenance action.

9. The intelligent operation and maintenance control method for the core network according to claim 5, characterized in that, The service impact information of at least one target user that can be affected by the target network element performing the target operation and maintenance action is determined in the following manner: Identify at least one associated user that can be affected by the target network element performing the target operation and maintenance action; The user level, user behavior data, and business indicator information of the associated user are obtained. The business indicator information is used to characterize the business performance of the target service currently used by the associated user. The target service is a service provided by the target network element. The feasibility result of determining the target network element to perform the target operation and maintenance action based on the operation and maintenance impact data includes: If the target user whose user level is higher than the set level is not included among the at least one associated user, it is determined that the target network element is suitable for performing the target operation and maintenance action. In response to the fact that at least one of the at least associated users includes at least one target user, based on the user behavior data and business indicator information of the target user, a trained fifth prediction model is used to predict the business quality score of the target service after the target network element performs the target operation and maintenance action, wherein the business quality score exceeding a second set value indicates that the target network element is suitable for performing the target operation and maintenance action.

10. An intelligent operation and maintenance control device for a core network, characterized in that, include: A network element determination unit is used to determine at least one target network element in the core network that has an anomaly in response to the detection of an anomaly in the core network. The status acquisition unit is used to acquire target status data corresponding to each target network element. The target status data includes at least one of the following: network fault level corresponding to the target network element, indicator value of at least one performance indicator, and service experience difference value of the service. An action determination unit is used to determine the appropriate target operation and maintenance action for the target network element based on the target state data and using a trained reinforcement learning model. The reinforcement learning model includes the mapping relationship between different state data and operation and maintenance actions. The feasibility determination unit is used to determine the feasibility result of the target network element performing the target operation and maintenance action, and the feasibility result is used to indicate whether the target network element is suitable for performing the target operation and maintenance action; An action execution unit is configured to control the target network element to perform the target operation and maintenance action in response to the feasibility result indicating that the target network element is suitable for performing the target operation and maintenance action.

Images