5G core network strategy control method and system based on energy consumption perception

Abstract

The invention relates to a 5G core network strategy control method and system based on energy consumption perception. The method comprises the following steps: acquiring network function entity static energy consumption, user terminal energy consumption and service flow dynamic energy consumption data through an energy information function EIF; calculating static energy consumption PenC and dynamic energy consumption DenC based on the collected data, establishing a model of total energy consumption EC = PenC + DenC, and generating an energy consumption analysis report; inputting the energy consumption index, the QoS demand and the network prediction information into a multi-objective optimization engine, and generating a candidate strategy set containing a green path; pDU session QoS reconfiguration is executed based on an optimization strategy, and a renewable energy power supply path is preferentially selected in UPF forwarding; monitoring the energy consumption data in real time through the EIF, comparing the energy consumption data with a preset threshold value, and triggering an abnormal alarm; strategy re-optimization is triggered according to energy consumption abnormity, strategy rollback is executed after energy consumption is recovered, and AI model prediction is combined for active optimization.

Description

Technical Field

[0001] This application relates to the field of information and data technology, specifically to a 5G core network strategy control method and system based on energy consumption awareness. Background Technology

[0002] With the large-scale deployment and application of 5G networks, their high energy consumption problem has become increasingly prominent, and has become a major challenge for operators to control costs and achieve "dual carbon" goals. As the brain and traffic hub of the network, the energy consumption management of the 5G core network is crucial. The existing 5G core network policy control mechanism mainly focuses on quality of service (QoS) indicators, such as bandwidth, latency and packet loss rate. Energy consumption is not considered as a key factor in the decision-making process. Although the traditional operation and maintenance management (OAM) system can collect total power consumption data at the device level, the data granularity is coarse and cannot be refined to the specific network function (NF), user terminal (UE) or service flow level, resulting in a lack of precision in energy efficiency management. Existing technologies lack a mechanism to jointly optimize energy consumption indicators and QoS requirements, and fail to dynamically adjust policies and perform closed-loop optimization based on actual energy consumption status. Although some existing technologies involve energy consumption optimization, they are not embedded in the architecture and policy control process unique to the 5G core network. In particular, they lack dedicated energy consumption sensing entities and deep collaboration with QoS policies, making it difficult to meet the urgent need for 5G networks to achieve green and low-carbon operation while ensuring service performance. Summary of the Invention

[0003] In order to solve the problems existing in the prior art, the purpose of this application is to provide a 5G core network policy control method and system based on energy consumption awareness.

[0004] The energy consumption-aware 5G core network policy control method described in this application includes: S101. Collect static energy consumption data of network functional entities, user terminal energy consumption data, and dynamic energy consumption data of service flows through the Energy Information Function (EIF). S102. Calculate static energy consumption (PenC) and dynamic energy consumption (DenC) based on the collected data, establish the total energy consumption EC=PenC+DenC model and generate an energy consumption analysis report; S103. Input the energy consumption indicators, QoS requirements, and network prediction information into the multi-objective optimization engine to generate a candidate strategy set containing green paths. S104. Perform PDU session QoS reconfiguration based on optimization strategy, and prioritize renewable energy power supply path in UPF forwarding; S105. Monitor energy consumption data in real time through EIF, compare it with preset thresholds and trigger abnormal alarms. S106. Based on the energy consumption anomaly triggering strategy re-optimization, execute strategy rollback after energy consumption recovers, and combine AI model prediction for proactive optimization.

[0005] Preferably, in step S101, static energy consumption data obtained from network functional entities is classified and stored through a preset collection mechanism to form a preliminary static energy consumption record. User terminal energy consumption data is collected synchronously, and after unified formatting, its real-time status is determined. Combined with dynamic energy consumption data of business flows, a filtering mechanism is triggered for data exceeding a threshold to locate the source of abnormal energy consumption. The support vector machine algorithm is used to classify the dynamic energy consumption data, identify its fluctuation pattern, and compare and analyze it with the static energy consumption record. Nodes with significant differences are marked as potential risk points, and the key monitoring scope is clarified. Based on this scope, real-time energy consumption data is continuously tracked through a periodic update mechanism to analyze the changing trend. The dynamic data of business flows and terminal energy consumption are aggregated and analyzed to finally form a comprehensive energy consumption distribution status, completing the closed-loop processing from collection to judgment.

[0006] Preferably, in step S102, based on data collected from network functional entities and business processes, static and dynamic energy consumption are classified separately to form hierarchically stored energy consumption records. After the static data is calibrated using a standardized template, it is combined with the dynamic energy consumption and the total energy consumption value is calculated by summing. This is used to construct an energy consumption distribution model. The support vector machine algorithm is used to analyze its correlation with network functions, identify key energy consumption areas, extract the corresponding network function operation logs for key areas, deeply mark energy consumption peaks exceeding the threshold, locate high-energy-consuming modules, and then combine real-time data from business processes to continuously track energy consumption changes using a periodic scanning mechanism to form a dynamically adjusted distribution state. Static and dynamic data are aggregated to generate structured records and stored hierarchically as a complete basis for energy consumption analysis.

[0007] Preferably, in S103, energy consumption indicators, service quality data and network prediction information are obtained from network operation records, and after being classified and stored by the data acquisition framework, a structured initial dataset is formed. Then, the data integration tool unifies the format, generates a multi-dimensional integrated comprehensive data group, and inputs it into the optimization engine for correlation analysis to obtain the priority ranking of each dimension. Based on the ranking results, candidate solutions are screened using the green path standard to form a preliminary strategy set. For strategies that exceed the energy consumption limit, a secondary filtering is performed through the correlation analysis module to obtain an optimized strategy combination. Then, the balance between service quality and green path is evaluated by the target optimization tool to determine the final applicable solution. The solution is then bound to network prediction information through an automated storage mechanism to build a callable strategy library.

[0008] Preferably, in S104, based on network operation logs and real-time monitoring of the user plane, a PDU session service quality dataset is obtained and processed in layers to determine the optimization priority of each session. For sessions below the threshold, the forwarding path is adjusted through a reconfiguration mechanism to form an updated scheme. Combined with renewable energy power supply information, the path selection module matches feasible green forwarding paths. When unstable energy supply is detected, a backup path switching mechanism is activated to dynamically optimize and form a final set of forwarding paths. After the set is automatically stored and bound to policies, a path policy library that can be called by services is generated.

[0009] Preferably, in step S105, energy consumption data is collected in real time through an energy information framework. The monitoring module continuously tracks the data to form a preliminary energy consumption record. The data is then compared with a preset threshold, and abnormal points that exceed the range are marked to form an abnormal dataset. Based on the timestamp and device identifier of the abnormal data, the abnormal situation is classified by a filtering tool. Then, an alarm mechanism is used to prioritize the abnormal by category. High-priority abnormalities are immediately triggered to locate the specific object. A structured log is generated and archived for the alarm object. At the same time, the persistence of the abnormality is analyzed in conjunction with the device operating status. If the abnormality persists for a timeout, a backup process is started for processing. The data after processing is tracked by the monitoring module, and the historical records are updated to form a complete monitoring file.

[0010] Preferably, in S106, energy consumption data is continuously collected through a monitoring mechanism, classified and organized to form an energy consumption dataset, anomalies deviating from the normal range are identified and an anomaly list is generated, and then time and equipment information are extracted. The anomaly priority is determined through a classification algorithm, and strategy optimization processing is implemented for high-priority anomalies. If the data is recovered, a rollback mechanism is triggered, and an adjustment plan is generated by combining an artificial intelligence model to predict future energy consumption trends. The optimized data is continuously tracked through the monitoring module, and historical records are associated to form a complete energy consumption monitoring archive.

[0011] The energy consumption-aware 5G core network policy control system described in this application includes: The energy consumption data acquisition module is used to collect static and dynamic energy consumption data from network functional entities, user terminals and business flows through energy information functions, and to classify and format the data to obtain structured energy consumption records. An energy consumption analysis and modeling module, connected to the energy consumption data acquisition module, is used to construct a total energy consumption model based on the energy consumption records, calculate static and dynamic energy consumption, identify high energy consumption areas and fluctuation patterns, and generate an energy consumption analysis report. The multi-objective strategy generation module is connected to the energy consumption analysis and modeling module. It is used to input energy consumption indicators, service quality requirements, and network prediction information into the optimization engine, perform multi-dimensional correlation analysis, and obtain a candidate strategy set containing green paths. The policy execution and path control module is connected to the multi-objective policy generation module and is used to perform QoS reconfiguration of PDU sessions, prioritize renewable energy power supply paths in user plane functions, and dynamically adjust forwarding paths according to energy status. The energy consumption monitoring and alarm module is connected to the energy consumption data acquisition module and the strategy execution and path control module. It is used to monitor energy consumption data in real time, compare it with preset thresholds, identify anomalies and trigger alarms, generate structured alarm records and start backup processing procedures. The strategy optimization and prediction module is connected to the energy consumption monitoring and alarm module and the multi-objective strategy generation module. It is used to trigger strategy re-optimization based on energy consumption anomalies, execute strategy rollback after energy consumption recovers, and combine AI models to predict energy consumption trends.

[0012] The 5G core network policy control method and system based on energy consumption awareness described in this application has the advantage of achieving joint optimization of energy efficiency and service quality: by elevating energy consumption indicators to independent policy parameters at the same level as QoS, and performing multi-objective optimization in the policy decision-making process, it fundamentally changes the traditional core network policy mode of "emphasizing performance and neglecting energy efficiency". While ensuring a good service experience, it significantly reduces overall network energy consumption and improves the precision of energy management: By introducing a dedicated Energy Information Function (EIF) entity, a separate calculation model of "Total Energy Consumption (EC) = Static Energy Consumption (PenC) + Dynamic Energy Consumption (DenC)" was established, which can accurately perceive and model the energy consumption of network functions, user terminals and individual service flows, enabling energy efficiency management to move from equipment-level extensive management to a new stage of flow-level refined management. Based on the closed-loop optimization mechanism triggered by energy consumption threshold, the strategy can be adjusted in real time and automatically when abnormal energy consumption is detected, and a rollback is performed after energy consumption is restored. This ensures that the network always maintains the best balance between energy efficiency and performance in a dynamically changing environment, overcoming the drawbacks of slow response of traditional static configuration. The innovative green path selection is introduced: In the forwarding path decision of User Plane Function (UPF), renewable energy priority is introduced for the first time, and network nodes powered by green energy are given priority, which directly supports the network's carbon emission reduction target and gives the 5G core network a clear "green" attribute. It has the ability to predict energy consumption risks in a forward-looking manner: By combining AI or ML models and using historical energy consumption and strategy data, it can predict future energy consumption risks and trigger proactive pre-scheduling strategies, realizing the transformation from "passive response" to "proactive defense" and further improving the intelligence and robustness of the system. Attached Figure Description

[0013] Figure 1This is the flow chart of a 5G core network policy control method based on energy consumption awareness as described in this application. Figure 1 ; Figure 2 This is the flow chart of a 5G core network policy control method based on energy consumption awareness as described in this application. Figure 2 . Detailed Implementation

[0014] like Figures 1-2 As shown, this application describes a 5G core network policy control method based on energy consumption awareness.

[0015] like Figures 1-2 As shown, S101 collects static energy consumption data of network function entities, energy consumption data of user terminals, and dynamic energy consumption data of service flows through the Energy Information Function (EIF).

[0016] Furthermore, in step S101, static energy consumption data is obtained from network functional entities through the energy information function, and the data is classified and stored using a pre-established acquisition mechanism to obtain a preliminary static energy consumption record. Based on the preliminary static energy consumption records, energy consumption data of user terminals is collected synchronously and processed using a unified formatting method to determine the real-time status of user terminal energy consumption. By monitoring the real-time energy consumption status of user terminals, dynamic energy consumption data of business flows can be obtained. If the dynamic energy consumption data exceeds a preset threshold, a data filtering mechanism is triggered to determine the source of abnormal energy consumption. Based on the source of abnormal energy consumption, the support vector machine algorithm is used to classify and analyze the dynamic energy consumption data of the business flow to obtain the fluctuation pattern of dynamic energy consumption. By comparing and analyzing the static energy consumption data of network functional entities through the fluctuation pattern of dynamic energy consumption, if there is a significant difference between the static energy consumption data and the dynamic energy consumption fluctuation pattern, it is marked as a potential risk point, and the scope of entities that need to be monitored in key areas is determined. Based on the key monitored entities, real-time energy information data is obtained, and a periodic update mechanism is used to continuously track the collected data and determine the changing trend of energy consumption data. By analyzing the changing trends of energy consumption data, we can perform a linked analysis of business dynamics and terminal energy consumption, and use data aggregation to obtain the overall distribution of energy consumption.

[0017] Specifically, in step S101, the specific implementation method of collecting static energy consumption of network functional entities, energy consumption of user terminals and dynamic energy consumption of service flows through the Energy Information Function (EIF) can be integrated into a complete technical process. For collecting static energy consumption data of network functional entities, assuming a server cluster in a data center contains 100 servers, each with a rated power of 500 watts, the system automatically reads the power management interface data of the servers every hour through the EIF module, calculates the static energy consumption as 100 units × 500 watts = 50,000 watt-hours (i.e., 50 kilowatt-hours), and stores the data in the energy consumption database for subsequent analysis. The analysis results show that static energy consumption accounts for 40% of the total energy consumption, indicating that the standby mode of the equipment needs to be optimized. For collecting user terminal energy consumption data, taking an enterprise network with 5,000 user terminals as an example, EIF collects CPU and screen power consumption data of the terminals every minute through the terminal device energy consumption monitoring agent program. Assuming that the average power consumption of each terminal is 50 watts, the total energy consumption is 5,000 units × 50 watts = 250,000 watt-hours (i.e. 250 kilowatt-hours). Through algorithm analysis, it is found that energy consumption surges by 20% during peak hours. The system automatically generates energy-saving strategy suggestions, such as reducing screen brightness to reduce energy consumption by 10%. For dynamic energy consumption data collection of business flows, taking video streaming services as an example, EIF monitors the bandwidth usage and processing time of each business flow in real time. Assuming that a certain business flow has a bandwidth of 10Mbps and a processing time of 2 hours, the energy consumption model calculates that the dynamic energy consumption is 10Mbps × 2 hours × 0.5W / Mbps·hour = 10Wh. Through analysis, it was found that the energy consumption fluctuation of this business flow is positively correlated with the user access volume. The system automatically adjusts resource allocation to reduce redundant energy consumption by 5%. These data collection and analysis processes are implemented through EIF's automated interface, forming a closed-loop optimization logic. Static energy consumption analysis provides a benchmark for terminal energy consumption optimization, and the terminal energy consumption strategy in turn affects the dynamic adjustment of business flow, ultimately achieving the goal of reducing overall energy consumption by about 15%.

[0018] like Figures 1-2 As shown in step S102, calculate static energy consumption (PenC) and dynamic energy consumption (DenC) based on the collected data, establish the total energy consumption model EC=PenC+DenC, and generate an energy consumption analysis report.

[0019] Furthermore, in step S102, by acquiring collected data from network functional entities and business processes, preliminary classification is performed on static energy consumption and dynamic energy consumption respectively, resulting in hierarchically stored energy consumption records; Based on the energy consumption records stored in layers, the static energy consumption data is checked item by item, and the data is calibrated using a pre-established standardized template to determine the calibrated static energy consumption dataset. By combining the calibrated static energy consumption dataset with the dynamic energy consumption data in the business process, the comprehensive total energy consumption value is obtained by summing the data. For the overall total energy consumption value, an energy consumption distribution model is constructed, and the support vector machine algorithm is used to classify the correlation between the total energy consumption value and network function to identify key areas of energy consumption distribution; Based on the key areas of energy consumption distribution, obtain the operation logs of relevant network functions. If the energy consumption peak in the operation log exceeds the preset threshold, then the data in that area is deeply marked to identify the high-energy-consuming functional modules. By using high-energy-consuming functional modules and combining them with real-time data streams from business processes, a periodic scanning mechanism is used to continuously track energy consumption changes and obtain the dynamically adjusted energy consumption distribution status. Based on the dynamically adjusted energy consumption distribution, a structured comprehensive data record is generated. Data aggregation tools are used to store static and dynamic energy consumption in layers to determine the basis for the final energy consumption analysis.

[0020] Specifically, in step S102, during the process of building the total energy consumption model and generating the analysis report, the system uses automation technology to accurately calculate and comprehensively analyze static energy consumption and dynamic energy consumption, forming a complete technical process. Regarding the calculation of static energy consumption (PenC), assuming a communication base station has 50 core devices, each with a base power of 800 watts, the system collects power data of the devices every 4 hours through its built-in energy consumption monitoring module. The static energy consumption is calculated to be 50 units × 800 watts = 40,000 watt-hours (i.e., 40 kilowatt-hours). The data is automatically entered into the energy consumption analysis system. Subsequently, through historical data comparison algorithms, it is found that static energy consumption is stable during off-peak hours, accounting for approximately 35% of the total energy consumption, providing a basic reference for subsequent analysis. Regarding the calculation of dynamic energy consumption (DenC), taking the task processing of a cloud computing platform as an example, the system monitors the resource consumption of the task flow in real time. Assuming that a task flow occupies a bandwidth of 20Mbps and lasts for 3 hours, the energy consumption calculation model is set to consume 0.8 watts per Mbps·hour, resulting in dynamic energy consumption of 20Mbps × 3 hours × 0.8 watts / Mbps·hour = 48 watt-hours. The analysis algorithm further identifies that dynamic energy consumption fluctuates greatly during peak task periods and is related to task scheduling density. The system combines this data with static energy consumption to form a model result of total energy consumption EC = PenC + DenC = 40000 watt-hours + 48 watt-hours = 40048 watt-hours. The system automatically generates analysis reports based on the total energy consumption model. Through the built-in energy consumption distribution algorithm, it calculates that the proportions of static energy consumption and dynamic energy consumption are 99.88% and 0.12%, respectively. Combined with historical trend analysis, it identifies that the fluctuation of dynamic energy consumption is mainly affected by the suddenness of task flow. The system automatically associates the analysis results with the base station equipment management database to generate time distribution suggestions for optimizing task scheduling, thus completing the closed-loop logic from data collection to report generation.

[0021] like Figures 1-2 As shown in step S103, energy consumption indicators, QoS requirements, and network prediction information are input into the multi-objective optimization engine to generate a candidate policy set containing green paths.

[0022] Furthermore, in step S103, energy consumption indicators, service quality data, and network prediction information are obtained from network operation records, and classified and stored through a pre-established data acquisition framework to obtain a structured initial dataset; For the structured initial dataset, a data integration tool was used to process the energy consumption indicators, service quality data and network prediction information in a unified format to determine the comprehensive data group after multi-dimensional fusion. The integrated data set after multi-dimensional fusion is input into the optimization engine. The dimensional analysis module analyzes the data correlation layer by layer to obtain the priority ranking results between each dimension. Based on the priority ranking results, candidate solutions that meet the green path standard are selected using preset path generation rules to determine the initial strategy set. For the initial set of strategies, if the energy consumption index of a certain strategy exceeds the preset threshold, it will be filtered a second time through the data correlation analysis module to obtain an optimized strategy combination. Based on the optimized strategy combination, the balance between service quality and green path is evaluated using target optimization tools to determine the final applicable strategy. For the final applicable strategy, an automated storage mechanism is used to bind it with network prediction information to generate a strategy library that can be called.

[0023] Specifically, in step S103, within the field of communication networks, the system integrates energy consumption indicators with service quality requirements and network prediction information through automation technology, inputs them into a multi-objective optimization engine, generates a candidate strategy set containing green paths, and forms a complete business process. The system extracts energy consumption data from the database. For example, it assumes that the average energy consumption of a data center over the past 24 hours is 5000 kWh. At the same time, it collects service quality requirement data, such as requiring network latency to be no more than 50 milliseconds and packet loss rate to be less than 0.1%. The system calls the network prediction module, which, based on historical traffic data and machine learning algorithms, predicts that the peak traffic period in the next 12 hours may reach 100Gbps per second, and inputs this data along with energy consumption indicators into the multi-objective optimization engine. The engine uses a genetic algorithm for calculation, setting the energy consumption weight to 0.4 and the service quality weight to 0.6. After 1000 iterations, a set of candidate strategies is generated, one of which is a green path strategy: By dynamically adjusting the routing path and directing traffic to a backup node with lower energy consumption, it is expected to reduce energy consumption to 4500 kWh, while keeping latency at 48 milliseconds and packet loss rate at 0.08%. The system analyzes the candidate strategies and uses simulation algorithms to evaluate the performance of each strategy under predicted traffic. It finds that the green path strategy can maintain service quality stability during peak periods and is highly compatible with business needs. The system automatically stores the analysis results in the policy library and works in conjunction with the network topology management module to generate executable routing adjustment schemes, forming a complete logical chain from data integration to policy output, ensuring that subsequent network scheduling can prioritize energy saving and performance balance.

[0024] like Figures 1-2 As shown, S104 performs PDU session QoS reconfiguration based on the optimization strategy and prioritizes renewable energy power supply paths in UPF forwarding.

[0025] Further, in step S104, the current quality of service parameters of the protocol data unit session are obtained from the network operation log, and the forwarding status of the user plane is collected in real time through a pre-established monitoring framework to obtain the initial quality of service dataset. For the initial quality of service dataset, a data classification tool is used to perform hierarchical processing on the relevant parameters of the protocol data unit session to determine the priority ranking of each session under the optimization strategy. Based on the priority ranking results, if the service quality parameter of a certain session is lower than the preset threshold, the forwarding path of the session will be adjusted through the reconfiguration process to obtain the updated path allocation scheme. For the updated path allocation scheme, obtain the path information of renewable energy power supply, and match the user plane function forwarding through the path selection module to determine the suitable feasible path combination; Based on the adapted feasible path combination, if the power supply status of a certain path is detected to be unstable, it will be dynamically adjusted through the backup path switching mechanism to determine the final set of forwarding paths. For the final set of forwarding paths, an automated storage tool is used to bind it with optimization strategies to obtain a path strategy library that can be invoked by business operations.

[0026] Specifically, in step S104, within the communication network domain, the system implements the optimization strategy through automated processes, prioritizing renewable energy power supply paths in PDU session QoS reconfiguration and UPF forwarding, and constructs a complete business logic chain. First, the system extracts the QoS parameters of the current PDU session from the network management database. Assuming that the initial bandwidth requirement of a session is 50Mbps and the latency tolerance is 30 milliseconds, the real-time monitoring module detects that network congestion has caused the latency to rise to 35 milliseconds, triggering the QoS reconfiguration requirement. The system automatically invokes the optimization algorithm, adopts a traffic allocation model based on deep learning, analyzes session data from the past 6 hours, predicts that bandwidth demand may increase to 70Mbps in the next 2 hours, and calculates a new QoS configuration scheme accordingly, adjusting the bandwidth allocation to 75Mbps to reserve redundancy, while resetting the latency target to 28 milliseconds. In the selection of UPF forwarding paths, the system prioritizes nodes powered by renewable energy. Assuming there are three forwarding paths available in the current network, path A has an energy source of 80% renewable energy, path B has 50%, and path C has 30%. The system uses a weighted calculation model, setting the weight of energy source to 0.7 and the weight of transmission efficiency to 0.3. The analysis shows that path A has the highest total score, with a transmission delay of 25 milliseconds and a bandwidth support of 80 Mbps, which meets the adjusted QoS requirements. The system further conducted load tests on path A using network simulation tools to simulate its performance under peak traffic conditions in the next hour, confirming that its packet loss rate was only 0.05%, which meets business requirements. The system automatically distributes reconfiguration parameters and path selection schemes to the UPF node, and dynamically updates forwarding rules through the SDN controller to ensure that session data flow passes through path A first. At the same time, the execution logs are stored in the monitoring platform, forming a closed-loop processing flow from demand detection to path optimization. To enhance logical coherence, the system also links with the billing module to record data on the impact of path selection on energy consumption. It is assumed that the use of path A reduces energy consumption per unit flow to 0.2 kWh per GB, providing a reference for subsequent strategy optimization.

[0027] like Figures 1-2 As shown, S105 monitors energy consumption data in real time through EIF, compares it with preset thresholds, and triggers an abnormal alarm.

[0028] Furthermore, in step S105, energy consumption data is collected in real time through the energy information framework, and a pre-established monitoring module is used to continuously track data changes to obtain preliminary energy consumption records. Based on the preliminary energy consumption records, each set of data is compared with a preset threshold. If a data point exceeds the preset threshold range, it is marked as an anomaly, and the set of abnormal data is determined. For the abnormal data set, obtain its corresponding timestamp and device identifier, and classify the abnormal situation through data filtering tools to obtain the classified abnormal categories; Based on the classified anomaly categories, an alarm mechanism is used to prioritize the anomalies of different categories. If the priority of a certain category is higher than the set standard, an alarm signal is immediately triggered to determine the specific object that triggered the alarm. For the specific object that triggers the alarm, the corresponding alarm log is generated through the information framework, and the automated storage tool is used to archive it to obtain structured alarm records; Based on the structured alarm records, obtain the operating status data of the relevant equipment, analyze the persistence of the abnormal situation through comparison, and if the persistence exceeds the preset time limit, start the backup processing procedure and determine the subsequent processing plan. For subsequent processing, the processed energy consumption data is tracked through the data monitoring module, and the data is linked with historical data using a record update tool to obtain a complete monitoring file.

[0029] Specifically, in step S105, within the field of communication networks, the system achieves real-time monitoring and anomaly alarm of energy consumption data through automated processes, and constructs an efficient energy consumption management logic chain. The system first collects energy consumption data from network devices every 5 minutes through the EIF (Energy Information Framework) interface. Assuming that the current energy consumption of a certain node is 3.5 kWh per hour, while the preset normal threshold is 2.8 kWh per hour, the system automatically compares the collected data with the threshold and finds that it exceeds the threshold by 0.7 kWh, triggering the anomaly detection mechanism. The system calls a time-series-based anomaly analysis algorithm to extract energy consumption data from the past 24 hours. The average value is calculated to be 2.6 kWh per hour with a standard deviation of 0.3. The system determines that the current energy consumption value exceeds the normal fluctuation range (i.e., the average plus twice the standard deviation is 3.2 kWh), thus confirming the abnormal state. The system combines energy consumption anomalies with traffic load data through the correlation analysis module. Assuming the current node traffic is 800 Mbps, historical data shows that traffic and energy consumption are linearly correlated. The regression model calculates that for every 100 Mbps increase in traffic, energy consumption increases by about 0.4 kWh. It is inferred that the current anomaly may be caused by a surge in traffic. The system further generates alarm information and automatically pushes the abnormal details (including energy consumption of 3.5 kWh, threshold of 2.8 kWh, and traffic load of 800 Mbps) to the network management platform, and triggers the energy-saving strategy evaluation module to analyze the node operating parameters and derive temporary solutions to reduce device power consumption, such as dynamically adjusting the processing priority of non-critical tasks, which is expected to reduce energy consumption to 3.0 kWh per hour. The system stores alarm logs and analysis results in the database and links with the maintenance scheduling module to automatically generate optimization suggestion reports, ensuring continuous improvement of subsequent energy consumption management strategies and forming a closed-loop process from monitoring to alarm to optimization.

[0030] like Figures 1-2 As shown, in step S106, the strategy is re-optimized based on the energy consumption anomaly trigger. After the energy consumption is restored, the strategy is rolled back and combined with AI model prediction for proactive optimization.

[0031] Furthermore, in step S106, energy consumption data is continuously collected through a pre-established monitoring mechanism, and changes in the collected data are tracked in real time. Data processing tools are used to classify and organize each set of data to obtain a classified energy consumption data set. For the categorized energy consumption data set, information on outliers is obtained, and the data is matched with a preset normal range using a comparison tool. If a set of data deviates from the normal range, it is marked as outlier data, and a list of outlier data is determined. For the list of abnormal data, information extraction tools are used to obtain the corresponding time identifiers and device information. The abnormal data are then prioritized using a classification algorithm to obtain a sorted list of abnormal data priorities. Based on the sorted list of abnormal priorities, obtain the abnormal data with higher priority, process it through the preset strategy optimization module, and if the data is restored to the normal range after processing, trigger the strategy rollback process to determine the rollback execution plan. For rollback execution plans, artificial intelligence models are used to predict and analyze future energy consumption trends, obtain information on potential anomalies, generate optimization and adjustment plans through forward-looking adjustment tools, and determine the direction of the adjusted strategy. Based on the adjusted strategy, the optimized energy consumption data is continuously tracked through the data monitoring module, and a recording tool is used to correlate it with historical data to obtain a complete energy consumption monitoring profile.

[0032] Specifically, in step S106, within the field of communication networks, the system optimizes the energy consumption anomaly trigger strategy through an automated process, and performs strategy rollback after energy consumption is restored, while combining artificial intelligence models for predictive optimization. The system detects an abnormal energy consumption at a node, for example, the current energy consumption reaches 4.2 kWh per hour, exceeding the preset threshold of 3.0 kWh by 1.2 kWh. It automatically triggers an optimization strategy, calling a machine learning-based energy consumption adjustment algorithm. By analyzing the energy consumption data of the past 48 hours, it calculates that the peak energy consumption is 3.8 kWh per hour and the average is 2.9 kWh per hour. Combined with the current ambient temperature of 28 degrees Celsius, it infers that the abnormality may be related to the decrease in the heat dissipation efficiency of the equipment. The system then generates a temporary load reduction plan, transferring some low-priority data streams to the backup node, and it is expected that the energy consumption can be reduced to 3.3 kWh per hour. When the system detects that the energy consumption has recovered to 2.7 kWh per hour, which is lower than the threshold of 3.0 kWh, the policy rollback mechanism is automatically triggered. By gradually restoring the original data flow allocation, the calculated energy consumption after recovery is stabilized at 2.9 kWh per hour, ensuring that network performance is not affected. The system calls a prediction model based on long short-term memory networks, inputs energy consumption data and traffic load data from the past 7 days, such as an average daily traffic of 600 Mbps, and predicts that the peak energy consumption may reach 3.6 kWh per hour in the next 24 hours. The system generates a pre-adjustment plan in advance, optimizes the equipment operating parameters, and controls the predicted peak energy consumption within 3.1 kWh per hour. To ensure a closed-loop logic, the system automatically uploads all adjustment records and prediction results to the energy management database, and works in conjunction with the network load balancing module to dynamically update traffic allocation strategies, forming a complete process from anomaly optimization to recovery and then to prediction adjustment.

[0033] The energy consumption-aware 5G core network policy control system described in this application includes: The energy consumption data acquisition module is used to collect static and dynamic energy consumption data from network functional entities, user terminals and business flows through energy information functions, and to classify and format the data to obtain structured energy consumption records. An energy consumption analysis and modeling module, connected to the energy consumption data acquisition module, is used to construct a total energy consumption model based on the energy consumption records, calculate static and dynamic energy consumption, identify high energy consumption areas and fluctuation patterns, and generate an energy consumption analysis report. The multi-objective strategy generation module is connected to the energy consumption analysis and modeling module. It is used to input energy consumption indicators, service quality requirements, and network prediction information into the optimization engine, perform multi-dimensional correlation analysis, and obtain a candidate strategy set containing green paths. The policy execution and path control module is connected to the multi-objective policy generation module and is used to perform QoS reconfiguration of PDU sessions, prioritize renewable energy power supply paths in user plane functions, and dynamically adjust forwarding paths according to energy status. The energy consumption monitoring and alarm module is connected to the energy consumption data acquisition module and the strategy execution and path control module. It is used to monitor energy consumption data in real time, compare it with preset thresholds, identify anomalies and trigger alarms, generate structured alarm records and start backup processing procedures. The strategy optimization and prediction module is connected to the energy consumption monitoring and alarm module and the multi-objective strategy generation module. It is used to trigger strategy re-optimization based on energy consumption anomalies, execute strategy rollback after energy consumption recovers, and combine AI models to predict energy consumption trends.

[0034] For those skilled in the art, various other corresponding changes and modifications can be made based on the technical solutions and concepts described above, and all such changes and modifications should fall within the protection scope of the claims of this application.

Claims

1. A 5G core network policy control method based on energy consumption awareness, characterized in that, include: S101. Collect static energy consumption data of network functional entities, energy consumption data of user terminals, and dynamic energy consumption data of service flows through the Energy Information Function (EIF). S102. Calculate static energy consumption PenC and dynamic energy consumption DenC based on the collected data, establish the total energy consumption EC=PenC+DenC model and generate an energy consumption analysis report. S103. Input the energy consumption indicators, QoS requirements, and network prediction information into the multi-objective optimization engine to generate a candidate strategy set containing green paths. S104. Perform PDU session QoS reconfiguration based on optimization strategy, and prioritize renewable energy power supply path in UPF forwarding; S105. Monitor energy consumption data in real time through EIF, compare it with preset thresholds and trigger abnormal alarms. S106. Based on the energy consumption anomaly triggering strategy re-optimization, execute strategy rollback after energy consumption recovers, and combine AI model prediction for proactive optimization.

2. The 5G core network policy control method based on energy consumption awareness according to claim 1, characterized in that, The collection of static energy consumption data of network function entities, user terminal energy consumption data, and dynamic energy consumption data of service flows through the energy information function includes: The static energy consumption data of network functional entities are classified and stored to obtain static energy consumption records; Synchronously collect energy consumption data from user terminals, and determine the real-time status after uniform formatting processing; Acquire dynamic energy consumption data of business flows, and trigger a filtering mechanism for data exceeding the threshold to locate the source of anomalies; The support vector machine algorithm is used to classify and analyze dynamic energy consumption data to obtain fluctuation patterns; By comparing and analyzing static energy consumption data with dynamic energy consumption fluctuation patterns, nodes with significant differences are marked as potential risk points. Based on continuous tracking of real-time energy consumption data within key monitoring areas, the overall energy consumption distribution status is obtained.

3. The 5G core network policy control method based on energy consumption awareness according to claim 1, characterized in that, The process of calculating static and dynamic energy consumption based on collected data, establishing a total energy consumption model, and generating an energy consumption analysis report includes: Static energy consumption and dynamic energy consumption are classified and stored in layers respectively; The static energy consumption data is calibrated, and the total energy consumption value is obtained by summing the dynamic energy consumption data. Construct an energy consumption distribution model, use the support vector machine algorithm to analyze the correlation between total energy consumption and network function, and identify key energy consumption areas; Energy consumption peaks exceeding the threshold are deeply marked to locate high-energy-consuming functional modules; By continuously tracking energy consumption changes in real-time business process data, a dynamically adjusted distribution status can be obtained; By aggregating static and dynamic data, a structured record is obtained.

4. The 5G core network policy control method based on energy consumption awareness according to claim 1, characterized in that, The step of inputting energy consumption indicators, service quality requirements, and network prediction information into a multi-objective optimization engine to generate a candidate strategy set containing green paths includes: Energy consumption indicators, service quality data, and network prediction information are obtained from network operation records to obtain a structured initial dataset; The initial dataset is processed in a unified format to obtain a comprehensive dataset that integrates multiple dimensions. The comprehensive data set is input into the optimization engine for correlation analysis to obtain the priority ranking of each dimension; Based on the priority ranking results, candidate solutions that meet the green path criteria are selected to obtain a preliminary strategy set; Strategies that exceed the energy consumption threshold are filtered a second time to obtain an optimized strategy combination; Assess the balance between service quality and green pathways to determine the appropriate solutions; The final solution is bound to the network prediction information to obtain a callable strategy library.

5. The 5G core network policy control method based on energy consumption awareness according to claim 1, characterized in that, The step of performing PDU session service quality reconfiguration based on optimization strategies and prioritizing renewable energy power supply paths in user plane function forwarding includes: Obtain the current quality of service parameters of the protocol data unit session from the network operation log, collect the user plane forwarding status in real time, and obtain the quality of service dataset; The service quality dataset is processed in layers to determine the optimization priority for each session; For sessions with service quality parameters below the threshold, the forwarding path is reconfigured to obtain an updated path allocation scheme. Obtain path information for renewable energy power supply, match user plane function forwarding, and determine feasible path combinations; For paths with unstable power supply, a backup path switching mechanism is activated to determine the final set of forwarding paths; The final set of paths is bound to optimization strategies to obtain a path strategy library that can be called by business applications.

6. The 5G core network policy control method based on energy consumption awareness according to claim 1, characterized in that, The method of monitoring energy consumption data in real time through energy information functions, comparing it with preset thresholds, and triggering abnormal alarms includes: Real-time energy consumption data is collected and continuously tracked to obtain preliminary energy consumption records; The data is compared with a preset threshold, and outliers that exceed the range are marked to obtain an abnormal data set. Based on the timestamps and device identifiers of the abnormal data, the abnormal situations are classified to obtain the abnormal categories; An alarm mechanism is used to prioritize anomalies, and alarms are triggered immediately for high-priority anomalies. Generate alarm logs and archive them to obtain structured alarm records; Analyze the persistence of the anomaly; if it exceeds the preset time limit, initiate the backup processing procedure. After tracking and processing, the data is linked to historical records to form a complete monitoring archive.

7. The 5G core network policy control method based on energy consumption awareness according to claim 1, characterized in that, The step of triggering strategy re-optimization based on energy consumption anomalies, rolling back the strategy after energy consumption recovers, and combining artificial intelligence model predictions for proactive optimization includes: Continuously collect and categorize energy consumption data to obtain an energy consumption dataset; Identify outliers that deviate from the normal range and obtain an anomaly list; Extract the timestamps and device information from the abnormal data, and determine the anomaly priority using a classification algorithm; For high-priority anomalies, implement strategy optimization processing; if the data is recovered, trigger strategy rollback. Artificial intelligence models are used to predict future energy consumption trends and generate forward-looking adjustment plans. Continuously track and optimize the data, and link it with historical records to obtain energy consumption monitoring files.

8. A 5G core network strategy control system based on energy consumption awareness, characterized in that, include: The energy consumption data acquisition module is used to collect static and dynamic energy consumption data from network functional entities, user terminals and business flows through energy information functions, and to classify and format the data to obtain structured energy consumption records. An energy consumption analysis and modeling module, connected to the energy consumption data acquisition module, is used to construct a total energy consumption model based on the energy consumption records, calculate static and dynamic energy consumption, identify high energy consumption areas and fluctuation patterns, and generate an energy consumption analysis report. The multi-objective strategy generation module is connected to the energy consumption analysis and modeling module. It is used to input energy consumption indicators, service quality requirements, and network prediction information into the optimization engine, perform multi-dimensional correlation analysis, and obtain a candidate strategy set containing green paths. The policy execution and path control module is connected to the multi-objective policy generation module and is used to perform service quality reconfiguration of PDU sessions, prioritize renewable energy power supply paths in user plane functions, and dynamically adjust forwarding paths according to energy status. The energy consumption monitoring and alarm module is connected to the energy consumption data acquisition module and the strategy execution and path control module. It is used to monitor energy consumption data in real time, compare it with preset thresholds, identify anomalies and trigger alarms, generate structured alarm records and start backup processing procedures. The strategy optimization and prediction module is connected to the energy consumption monitoring and alarm module and the multi-objective strategy generation module. It is used to trigger strategy re-optimization based on energy consumption anomalies, execute strategy rollback after energy consumption recovers, and combine artificial intelligence models to predict energy consumption trends.

9. The 5G core network strategy control system based on energy consumption awareness according to claim 8, characterized in that, The energy consumption data acquisition module is also used to filter and locate anomalies in the acquired energy consumption data in real time, and to classify and analyze the dynamic energy consumption data using the support vector machine algorithm.

10. The 5G core network strategy control system based on energy consumption awareness according to claim 8, characterized in that, The strategy execution and path control module is also used to dynamically switch backup paths according to the renewable energy power supply status, and bind the final path set with the strategy and store it as a callable strategy library.

Images