Energy storage method and system for base station management

Abstract

The invention discloses an energy storage method and system for base station management, and the method comprises the steps: carrying out the multi-modal data fusion processing according to the historical business volume data, real-time user movement mode data and meteorological data of each virtual base station in a target region, and obtaining a dynamic load prediction result of each virtual base station in a target time period; based on a dynamic load prediction result, adopting a distributed constraint optimization algorithm to dynamically allocate energy storage resources, and generating an optimal allocation strategy containing a cross-base-station energy scheduling path; and according to the wireless network topology structure characteristics of the target area, constructing an energy scheduling decision model based on a graph theory, optimizing an energy storage scheduling time sequence, and generating a dynamic energy storage scheduling scheme with a fault-tolerant mechanism so as to realize energy storage for base station management. By utilizing the embodiment of the invention, the energy efficiency of the base station can be improved, energy waste is reduced, and efficient configuration and real-time scheduling of energy storage resources are realized.

Description

Technical Field

[0001] The present invention belongs to the technical field of energy storage, and particularly relates to an energy storage method and system for base station management. Background Art

[0002] In modern communication networks, base stations are an important part of supporting mobile communication, and their operating efficiency directly affects the quality of network services and the user experience. With the rapid development of 5G and future communication technologies, the traffic volume of base stations has increased rapidly. Especially during peak traffic periods and in dense user environments, base stations are facing huge pressure on energy demand. To address this challenge, energy storage technology has been gradually introduced into base station management to balance supply and demand, improve energy efficiency, and reduce operating costs.

[0003] Currently, traditional base station management methods mostly rely on static energy scheduling strategies, which are difficult to effectively cope with changing traffic volume, user movement patterns, and meteorological conditions. These methods often rely on historical data for energy demand prediction and lack full utilization of real-time data, resulting in energy shortages or waste during peak periods. In addition, factors such as the charge and discharge efficiency of energy storage units, the state of charge, and the energy transmission loss between adjacent base stations have not been effectively considered in energy storage scheduling, which makes the configuration and use of energy storage resources unreasonable and affects the overall energy management efficiency. Summary of the Invention

[0004] The purpose of the present invention is to provide an energy storage method and system for base station management to solve the deficiencies in the prior art, which can improve the energy efficiency of base stations, reduce energy waste, and achieve efficient allocation and real-time scheduling of energy storage resources.

[0005] An embodiment of the present application provides an energy storage method for base station management, the method comprising: Performing multi-modal data fusion processing on historical traffic volume data, real-time user movement pattern data, and meteorological data of each virtual base station in the target area to obtain the dynamic load prediction results of each virtual base station during the target period; Based on the dynamic load prediction results, combining the current state of charge, charge and discharge efficiency curve of the energy storage unit associated with each virtual base station, and the energy transmission path loss parameters between adjacent base stations, using a distributed constraint optimization algorithm to perform dynamic allocation of energy storage resources, and generating an optimized allocation strategy including cross-base station energy scheduling paths; According to the characteristics of the wireless network topology structure of the target area, constructing a graph theory-based energy scheduling decision model, performing delay sensitivity analysis and energy efficiency balance processing on the optimized allocation strategy, optimizing the energy storage scheduling timing, and generating a dynamic energy storage scheduling scheme with a fault tolerance mechanism to achieve energy storage for base station management.

[0006] Optionally, performing multi-modal data fusion processing based on the historical traffic data, real-time user movement pattern data, and meteorological data of each virtual base station in the target area to obtain the dynamic load prediction results of each virtual base station during the target period, including: According to the historical traffic data, real-time user movement pattern data, and meteorological data of each virtual base station in the target area, using a multi-source data alignment algorithm based on dynamic time warping to eliminate the problem of inconsistent data timestamps and perform interpolation filling on missing values to obtain a time-synchronized multi-modal data set; For the time-synchronized multi-modal data set, using a feature extraction network based on a multi-head attention mechanism to extract the temporal features of historical traffic, the spatial features of user movement patterns, and the environmental features of meteorological data respectively, and through a cross-modal attention mechanism, weighted fusion of features of different modalities is performed to generate a fused multi-modal feature representation; Input the fused multi-modal feature representation into a pre-trained prediction model based on a spatio-temporal graph convolutional network, combine the base station topology structure of the target area, capture the spatio-temporal dependence relationship between base stations, predict the dynamic load change trend of each virtual base station during the target period, and obtain the dynamic load prediction results of each virtual base station during the target period.

[0007] Optionally, based on the dynamic load prediction results, combining the current state of charge, charge and discharge efficiency curve of the energy storage unit associated with each virtual base station, and the energy transmission path loss parameters between adjacent base stations, using a distributed constraint optimization algorithm to perform dynamic allocation of energy storage resources, and generating an optimized allocation strategy including cross-base station energy scheduling paths, including: According to the current state of charge, charge and discharge efficiency curve of the energy storage unit associated with each virtual base station, and the energy transmission path loss parameters between adjacent base stations, constructing a multi-dimensional constraint model of the energy storage state, and defining a feasible solution space for energy storage resource allocation through energy conservation constraints, charge and discharge rate constraints, and transmission loss constraints; Combining the dynamic load prediction results with the multi-dimensional constraint model of the energy storage state, using a distributed optimization framework based on the Lagrangian relaxation method to decompose the global optimization problem of energy storage resource allocation into multiple sub-problems, each sub-problem corresponding to a local optimization objective of a virtual base station, and the local optimization objectives including load balancing, energy transmission efficiency, and energy storage life; For the decomposed sub-problems, using a path optimization method based on a genetic algorithm, combining the energy transmission path loss parameters between adjacent base stations, dynamically searching for the optimal energy scheduling path, and generating a preliminary cross-base station energy scheduling plan; For the preliminary cross-base station energy scheduling plan, using a global optimization algorithm based on a consensus mechanism, through information exchange and iterative update between distributed nodes, coordinating the solutions of each sub-problem, and generating an optimized allocation strategy that satisfies the global constraint conditions.

[0008] Optionally, based on the characteristics of the wireless network topology in the target area, an energy scheduling decision model based on graph theory is constructed to perform delay sensitivity analysis and energy efficiency balancing processing on the optimized allocation strategy, optimize the energy storage scheduling timing, and generate a dynamic energy storage scheduling scheme with a fault tolerance mechanism to achieve energy storage for base station management, including: Based on the characteristics of the wireless network topology in the target area and the optimized allocation strategy, an energy scheduling decision model based on graph theory is constructed, where nodes represent base stations or energy storage units, edges represent energy scheduling paths, and edge weight assignment is performed through path loss parameters and transmission delay data to generate a weighted network topology graph; For the energy scheduling paths in the network topology graph, a delay sensitivity analysis method based on the shortest path algorithm is used to calculate the transmission delay of each path, and critical paths with delay sensitivity are screened out through a delay tolerance threshold; For the critical paths with delay sensitivity, an energy efficiency optimization method based on load balancing is used. Combining the dynamic load prediction results of base stations and the charge-discharge efficiency curves of energy storage units, the allocation ratio of energy scheduling paths is adjusted to ensure that high-load base stations obtain energy support first, and a scheduling scheme with balanced energy efficiency is generated; For the scheduling scheme with balanced energy efficiency, a fault tolerance mechanism based on redundant paths is introduced. By constructing backup energy scheduling paths and dynamic switching strategies, it is ensured that energy transmission can still be maintained when some paths fail, and finally a dynamic energy storage scheduling scheme with a fault tolerance mechanism is generated.

[0009] Optionally, based on the characteristics of the wireless network topology in the target area and the optimized allocation strategy, an energy scheduling decision model based on graph theory is constructed, where nodes represent base stations or energy storage units, edges represent energy scheduling paths, and edge weight assignment is performed through path loss parameters and transmission delay data to generate a weighted network topology graph, including: Based on the characteristics of the wireless network topology in the target area, base stations and energy storage units are abstracted as nodes, and the energy scheduling paths in the optimized allocation strategy are abstracted as edges. Among them, through the physical connection relationship and energy transmission requirements between base stations, the sets of nodes and edges are initialized to generate a preliminary network topology graph; For each energy scheduling path, a path loss calculation model based on environmental perception is used. Combining the distance between base stations, transmission medium characteristics, and environmental interference factors, the path loss parameters are calculated. Among them, through the energy transmission efficiency target in the optimized allocation strategy, the path loss parameters are normalized to generate path loss weights; According to the real-time scheduling requirements in the optimized allocation strategy, combining historical transmission delay data and the current network state, a delay prediction method based on time series analysis is used to dynamically update the transmission delay data of each energy scheduling path and use it as the delay weight; Perform multi-objective optimization fusion on the path loss weight and transmission delay weight, adopt a comprehensive weight assignment method based on the weighted summation method, combine the priority and fault tolerance requirements of the energy scheduling path, and assign comprehensive weights to the edges in the preliminary network topology graph to generate a weighted network topology graph.

[0010] Another embodiment of the present application provides an energy storage system for base station management, and the system includes: A fusion module, configured to perform multi-modal data fusion processing according to the historical traffic data, real-time user movement pattern data, and meteorological data of each virtual base station in the target area, and obtain the dynamic load prediction results of each virtual base station in the target period; An allocation module, configured to perform dynamic allocation of energy storage resources based on the dynamic load prediction results, combine the current state of charge, charge and discharge efficiency curve of the energy storage unit associated with each virtual base station, and the path loss parameters of energy transmission between adjacent base stations, and use a distributed constraint optimization algorithm to generate an optimized allocation strategy including cross-base station energy scheduling paths; A construction module, configured to construct a graph theory-based energy scheduling decision model according to the wireless network topology structure characteristics of the target area, perform delay sensitivity analysis and energy efficiency balance processing on the optimized allocation strategy, optimize the energy storage scheduling timing, and generate a dynamic energy storage scheduling scheme with a fault tolerance mechanism to achieve energy storage for base station management.

[0011] Another embodiment of the present application provides a storage medium, in which a computer program is stored, and the computer program is set to execute the method described in any one of the above when running.

[0012] Another embodiment of the present application provides an electronic device, including a memory and a processor, a computer program is stored in the memory, and the processor is set to run the computer program to execute the method described in any one of the above.

[0013] Compared with the prior art, an energy storage method for base station management provided by the present invention performs multi-modal data fusion processing according to the historical traffic data, real-time user movement pattern data, and meteorological data of each virtual base station in the target area, and obtains the dynamic load prediction results of each virtual base station in the target period; based on the dynamic load prediction results, a distributed constraint optimization algorithm is used to perform dynamic allocation of energy storage resources, and an optimized allocation strategy including cross-base station energy scheduling paths is generated; according to the wireless network topology structure characteristics of the target area, a graph theory-based energy scheduling decision model is constructed, the energy storage scheduling timing is optimized, and a dynamic energy storage scheduling scheme with a fault tolerance mechanism is generated to achieve energy storage for base station management, thereby being able to improve the energy efficiency of the base station, reduce energy waste, and achieve efficient configuration and real-time scheduling of energy storage resources. Description of the Drawings

[0014] Figure 1 Hardware structure block diagram of a computer terminal for an energy storage method for base station management provided by an embodiment of the present invention; Figure 2 Flow schematic diagram of an energy storage method for base station management provided by an embodiment of the present invention; Figure 3 Structure schematic diagram of an energy storage system for base station management provided by an embodiment of the present invention. Detailed implementation manners

[0015] The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be construed as a limitation to the present invention.

[0016] An embodiment of the present invention first provides an energy storage method for base station management. This method can be applied to electronic devices, such as computer terminals, specifically ordinary computers, etc.

[0017] The following takes running on a computer terminal as an example to describe it in detail. Figure 1 Hardware structure block diagram of a computer terminal for an energy storage method for base station management provided by an embodiment of the present invention. As Figure 1 shown, this computer device includes a processor, a memory, and a network interface connected through a system bus. Among them, the memory can include a non-volatile storage medium and an internal memory.

[0018] The non-volatile storage medium can store an operating system and a computer program. This computer program includes program instructions. When the program instructions are executed, the processor can execute any energy storage method for base station management.

[0019] The processor is used to provide computing and control capabilities to support the operation of the entire computer device.

[0020] The internal memory provides an environment for the operation of the computer program in the non-volatile storage medium. When this computer program is executed by the processor, the processor can execute any energy storage method for base station management.

[0021] This network interface is used for network communication, such as sending assigned tasks, etc. Those skilled in the art can understand that Figure 1 the structure shown in is only a block diagram of some structures related to the solution of the present application, and does not constitute a limitation to the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements.

[0022] It should be understood that the processor can be a Central Processing Unit (CPU), and the processor can also be other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc.

[0023] See Figure 2 , an embodiment of the present invention provides an energy storage method for base station management, which may include the following steps: S201, perform multi-modal data fusion processing based on the historical traffic data, real-time user movement pattern data, and meteorological data of each virtual base station in the target area to obtain the dynamic load prediction results of each virtual base station during the target period; This method aims to obtain the dynamic load prediction results of each virtual base station during a specific period by performing multi-modal data fusion processing on the historical traffic data, real-time user movement pattern data, and meteorological data of each virtual base station in the target area. Specifically, the historical traffic data reflects the user request volume and service requirements of the base station at different times; the real-time user movement pattern data provides the real-time behavior and movement trends of users in the network; the meteorological data may affect the operating environment of the base station and user behavior, such as changes in user travel patterns caused by weather changes. By comprehensively analyzing these multi-dimensional data, more accurate load prediction can be obtained, helping base station managers to formulate energy allocation and scheduling strategies in advance.

[0024] The realization of this step is of far-reaching significance. It not only improves the accurate prediction of the load conditions of each virtual base station but also effectively avoids the situation of insufficient energy or excessive consumption due to sudden load increases. This dynamic load prediction method makes the energy management of the base station more intelligent and forward-looking, helping to improve the stability of the network and the user experience. Especially during peak hours, accurate prediction can ensure sufficient energy supply, avoid service interruption or speed reduction phenomena, and thus greatly improve user satisfaction and network reliability.

[0025] Specifically, based on the historical traffic data, real-time user movement patterns data, and meteorological data of each virtual base station in the target area, a multi-source data alignment algorithm based on dynamic time warping can be used to eliminate the problem of inconsistent data timestamps and fill in the missing values by interpolation to obtain a time-synchronized multi-modal dataset; In the process of multi-modal data fusion, it is first necessary to ensure the consistency of timestamps of different data sources, which is very important for cross-analysis. The dynamic time warping (DTW) algorithm can flexibly align different time series even if they have variations or delays in time. At the same time, for the possible missing values in the data source, the missing parts are reasonably filled by interpolation methods to ensure that the generated multi-modal dataset is complete and continuous.

[0026] This step ensures the accuracy and consistency of the data during the fusion process, thus enhancing the reliability of subsequent analysis and prediction. By eliminating the inconsistency of timestamps and filling in the missing values, a time-synchronized multi-modal dataset is formed, providing a solid foundation for dynamic load prediction and accurately reflecting the network status and user behavior in the target area.

[0027] In the process of data fusion, it is first necessary to ensure the consistency of timestamps of each data source. Data often comes from different sources and devices. For example, the historical traffic data of a base station may be recorded hourly, while the user movement patterns and meteorological data may be recorded per minute. To solve this problem, a method based on dynamic time warping (DTW) can be used to flexibly compare different time series even if they have variations or delays in time. DTW arranges the time series in a non-linear way to maximize the similarity between two time series, thus achieving accurate time alignment.

[0028] While dealing with inconsistent timestamps, there may also be cases of missing data. For example, within a certain time period, a base station may fail to successfully record user activity data. To solve this problem, interpolation methods can be used to fill in the missing values. Common interpolation methods include linear interpolation, spline interpolation, etc. Through linear interpolation, we can linearly calculate based on the data points before and after the missing value to calculate the missing value. This process not only helps to maintain the integrity of the time series but also provides an accurate data basis for subsequent analysis.

[0029] Finally, through the DTW algorithm and interpolation processing, a time-synchronized multi-modal dataset is formed. This dataset will include information in multiple dimensions such as historical traffic, real-time user movement patterns, and meteorological data, so as to be fully utilized in subsequent analysis. Effective data alignment and missing value processing can significantly improve the accuracy of subsequent load prediction and provide a reliable basis for the energy scheduling and optimization of base stations.

[0030] For the time-synchronized multi-modal dataset, a feature extraction network based on the multi-head attention mechanism is adopted to extract the temporal features of historical traffic volume, the spatial features of user movement patterns, and the environmental features of meteorological data respectively. Through the cross-modal attention mechanism, the features of different modalities are weighted and fused to generate a fused multi-modal feature representation; The feature extraction network based on the multi-head attention mechanism can effectively capture the key features in various types of data. For historical traffic volume data, extracting its temporal features is to understand the change trend of traffic flow; the spatial features of user movement pattern data help to analyze the distribution of users in the network; the environmental features of meteorological data can reveal the impact of the external environment on user behavior. The introduction of the cross-modal attention mechanism allows the model to dynamically adjust the weights according to the importance of different modalities, thus forming a comprehensive feature representation, laying a foundation for subsequent load prediction.

[0031] The ability to extract and fuse features in this process greatly enhances the processing effect of complex data, enabling the model to capture richer context information, and thus improving the accuracy of dynamic load prediction. The weighted fusion of features of different modalities ensures the prominent display of important information, while unimportant information is weakened, which is particularly important in the prediction process.

[0032] In the process of feature extraction, adopting a network structure based on the multi-head attention mechanism can effectively extract important features from the time-synchronized multi-modal dataset. For historical traffic volume data, the network needs to focus on the temporal changes of the data and extract the peak and trough situations of traffic flow. For example, in a specific period, the historical traffic volume data may show a peak phenomenon, clearly reflecting an increase in user access demand. Through the multi-head attention mechanism, the network can process multiple attention heads in parallel and learn different temporal features, enabling the model to adapt to complex traffic fluctuations.

[0033] At the same time, when processing user movement pattern data, it is particularly important to extract spatial features. For example, the distribution of users in different geographical locations may affect the energy demand of the base station. In this process, based on the extraction of spatial features, the aggregation degree of users in a certain area can be analyzed, and then the load situation of the base station in this area can be predicted. Through the feature extraction of user movement pattern data, high-population density areas can be identified, providing an important basis for subsequent energy allocation.

[0034] The extraction of environmental features from meteorological data is also equally important because changes in the external environment may directly affect the user's behavior patterns. Meteorological factors such as temperature, precipitation, and wind speed can all influence the user's travel plans and the demand for using the network. By extracting environmental features, we can better understand the impact of weather changes on the traffic volume, and then consider the influence of meteorological factors in energy scheduling. Finally, through the cross-modal attention mechanism, these features are weighted and fused to generate a multi-modal feature representation containing comprehensive information. This feature representation can provide a solid foundation for the subsequent dynamic load prediction model.

[0035] The fused multi-modal feature representation is input into a pre-trained prediction model based on the spatio-temporal graph convolutional network. Combining with the base station topology structure of the target area, it captures the spatio-temporal dependence relationship between base stations, predicts the dynamic load change trend of each virtual base station within the target time period, and obtains the dynamic load prediction results of each virtual base station within the target time period.

[0036] In this step, the fused multi-modal feature representation will be input into a pre-trained spatio-temporal graph convolutional network (ST-GCN). In this network, the spatial dependence between base stations and the changing patterns over time can be captured. By combining the topology structure of the base stations, the model can simulate the mutual influence between base stations, thereby predicting the load changes within a specific time period.

[0037] The core value of this step lies in being able to utilize the established spatio-temporal dependence relationship to provide more accurate dynamic load prediction. The mutual influence between base stations is complex, and through the modeling of the graph convolutional network, this can be better reflected. Eventually, the prediction results are more in line with the actual situation, improving the response ability of the system.

[0038] In this step, the fused multi-modal feature representation will be input into a pre-trained spatio-temporal graph convolutional network (ST-GCN). In the spatio-temporal graph convolutional network, the model can not only handle the changes in time series data but also consider the topological structure relationship between base stations. For example, the load of a certain base station may be affected by the load conditions of surrounding base stations. For instance, the user request volume of adjacent base stations may be transferred to this base station to a certain extent. Through the graph convolutional layer, the network can effectively capture these spatial correlations, thereby improving the accuracy of load prediction.

[0039] Combining the topology structure of the base stations, the spatio-temporal graph convolutional network regards each base station as a node in the graph, and the energy transmission path between them as an edge. By representing the relationship between nodes through the adjacency matrix, the model can simultaneously learn the features between nodes and the information of the graph structure. In this way, when a certain base station has a traffic peak at a certain time period, adjacent base stations can also capture this change through graph convolution and influence each other, thus better predicting the future load situation.

[0040] Finally, the prediction results output by the spatio-temporal graph convolutional network will reflect the dynamic load change trend of each virtual base station during the target time period. For example, the model may predict that during a large-scale event, the traffic of a specific base station will increase sharply. Based on these predictions, base station managers can make preparations in advance and rationally allocate energy storage resources. This prediction mechanism not only improves the intelligent level of resource scheduling but also provides guarantee for the stable operation of base stations under fluctuating user demands.

[0041] S202, based on the dynamic load prediction results, combined with the current state of charge, charge-discharge efficiency curve of the energy storage unit associated with each virtual base station, and the energy transmission path loss parameters between adjacent base stations, use the distributed constraint optimization algorithm to dynamically allocate energy storage resources and generate an optimized allocation strategy including cross-base station energy scheduling paths; Based on the dynamic load prediction results, this method combines the actual service demands of different virtual base stations with the current state of charge, charge-discharge efficiency curve of the associated energy storage units, and the energy transmission path loss parameters between adjacent base stations, and uses the distributed constraint optimization algorithm to dynamically allocate energy storage resources. Specifically, the state of charge of the energy storage unit determines the available amount of energy storage, the charge-discharge efficiency reflects the actual energy loss during the charge-discharge process, and the path loss parameter indicates the energy loss encountered when transmitting energy between different base stations. These factors are taken into account through the optimization algorithm, so that the allocation of energy storage resources is not only based on the current service demands but also considers the economy and efficiency of energy transmission, thereby generating an optimized allocation strategy including cross-base station energy scheduling paths to ensure that each base station can obtain an appropriate amount of energy support during high-demand periods.

[0042] This optimized allocation strategy can effectively improve the energy management efficiency of base stations, especially in the case of large business fluctuations. For example, when a base station expects a sharp increase in traffic during a specific period, the energy storage unit can provide the necessary energy support during the demand peak, reducing the dependence on the power grid and thus lowering the operating cost. At the same time, considering the energy transmission loss between adjacent base stations, the optimized allocation strategy can ensure that the optimal path is selected during energy scheduling to achieve efficient energy allocation. This energy storage management method based on dynamic load prediction and real-time optimization not only improves the response ability of base stations but also enhances the overall network stability and service quality, providing a better experience for users.

[0043] Specifically, a multi-dimensional constraint model of the energy storage state can be constructed according to the current state of charge, charge-discharge efficiency curve of the energy storage unit associated with each virtual base station, and the energy transmission path loss parameters between adjacent base stations, and the feasible solution space of energy storage resource allocation can be defined through energy conservation constraints, charge-discharge rate constraints, and transmission loss constraints; In this step, it is first necessary to establish a multi-dimensional constraint model to comprehensively understand the state of the energy storage unit. The current state of charge of the energy storage unit associated with each virtual base station reflects the upper limit of its available energy, and the charge and discharge efficiency curve provides the efficiency changes during the charging and discharging processes, which may be affected by factors such as temperature and aging. The energy transmission path loss parameters between adjacent base stations can help the model calculate the possible losses encountered when transferring energy between different base stations. Through this information, a comprehensive model including energy conservation constraints, charge and discharge rate constraints, and transmission loss constraints is constructed to define the feasible solution space for energy storage resource allocation.

[0044] The construction of this model lays the foundation for subsequent optimization, ensuring the rationality and feasibility of resource allocation. By clarifying the feasible solution space, it avoids system instability or energy shortage caused by resource allocation under extreme conditions, ensuring sufficient energy supply during peak periods. This constraint model not only makes the management of energy storage resources more systematic but also provides clear guidance for subsequent decision-making.

[0045] In this stage, it is first necessary to collect various data related to the virtual base stations, including the current state of charge of the energy storage unit of each base station, the charge and discharge efficiency curve, and the energy transmission path loss parameters between adjacent base stations. Through comprehensive analysis of these data, a multi-dimensional energy storage state model can be constructed. Taking base stations A, B, and C as examples, assume that the current state of charge of the energy storage unit of base station A is 60%, and the charge and discharge efficiency is 85%; the state of charge of base station B is 75%, while that of base station C is 100%. On this basis, analyze the energy transmission path loss between adjacent base stations. For example, the loss from base station A to base station B is 10%, and the loss from base station B to C is 5%.

[0046] Next, by establishing energy conservation constraints, ensure that the energy released from the energy storage unit must meet the actual energy requirements of each base station, and calculate the loss during energy transfer. Taking base station B as an example, if it is predicted that the energy required during peak periods is 100 kWh, then when transferring energy from base station A to base station B, a 10% loss needs to be considered, which means that base station A needs to release 111.11 kWh of energy in its state of charge. The charge and discharge rate constraints need to be set according to the physical characteristics of the batteries of each base station. For example, base station A can only release at 80% of its maximum charge and discharge rate to prevent damage.

[0047] Through the above steps, the feasible solution space is finally defined, providing clear guidance for subsequent optimization. For example, assume that the constraint conditions set in the model are: the maximum release capacity of each base station, the threshold of energy loss, and the unit energy release time of each energy storage unit, etc. This feasible solution space will lay the foundation for the energy storage resource allocation between each base station, ensuring that the subsequent optimization process can be carried out within a reasonable range.

[0048] Combine the dynamic load prediction results with the multi-dimensional constraint model of the energy storage state, and adopt a distributed optimization framework based on the Lagrangian relaxation method to decompose the global optimization problem of energy storage resource allocation into multiple sub-problems. Each sub-problem corresponds to the local optimization objective of a virtual base station, and the local optimization objectives include load balancing, energy transfer efficiency, and energy storage life. In this step, the combination of the dynamic load prediction results and the multi-dimensional constraint model provides clear goals and constraints for the optimal allocation of energy storage resources. By applying the Lagrangian relaxation method, the global optimization problem is decomposed into multiple local sub-problems, and each sub-problem corresponds to the optimization objective of a virtual base station. These local optimization objectives include load balancing, which means that the load between base stations should be distributed as evenly as possible; energy transfer efficiency, that is, while meeting the energy demand, minimizing energy loss; and energy storage life, that is, extending the service life of the energy storage device as much as possible during the charge and discharge process.

[0049] This decomposed optimization strategy can effectively reduce the computational complexity and improve the execution efficiency of the algorithm. By refining the global problem into multiple local problems, each base station can perform optimization calculations independently and then coordinate to ensure the optimal solution of the overall problem. This method not only improves the flexibility of resource allocation but also enhances the base station's ability to quickly respond to dynamic changes, ensuring efficient energy scheduling even under extreme conditions.

[0050] At this stage, combine the previously constructed multi-dimensional constraint model of the energy storage state with the dynamic load prediction results to form an efficient optimization framework. Specifically, the dynamic load prediction results provide an important basis for the energy demand of each base station in the future time period. Suppose during a high-load period, base station A is expected to require 150 kWh, base station B requires 120 kWh, and base station C actually needs 80 kWh during this time period. Based on these load predictions, to achieve load balancing, the system needs to dynamically adjust the allocation of energy storage resources to ensure that the demands of all base stations can be met.

[0051] Next, the distributed optimization strategy based on the Lagrangian relaxation method decomposes the global optimization problem into multiple sub-problems. Each sub-problem corresponds to a virtual base station, so that each can independently handle its own optimization objective. For example, the local optimization objective of base station A may be to minimize the released energy cost and maximize load balancing; base station B may need to focus on energy transfer efficiency to ensure receiving energy through the optimal path; while the local optimization objective of base station C focuses on energy storage life to extend the battery's usage cycle as much as possible. Through this method, each virtual base station can achieve efficient cooperation while meeting independent goals.

[0052] Finally, the solutions to each sub-problem will be fed back into the global optimization framework. Through iterative calculations, the results of local optimization will be aggregated to guide subsequent global optimization decisions. The decision-making process needs to consider the energy transfer efficiency, state of charge, and transmission losses between base stations in order to generate a reasonable energy allocation plan to ensure balanced and reasonable energy supply between base stations during high-demand periods.

[0053] For the decomposed sub-problems, a path optimization method based on genetic algorithm is adopted. Combining the energy transfer path loss parameters between adjacent base stations, it dynamically searches for the optimal energy scheduling path to generate a preliminary cross-base station energy scheduling plan. In this step, for the previously decomposed sub-problems, a genetic algorithm is applied for path optimization, taking into account the energy transfer path loss parameters between adjacent base stations. The genetic algorithm is an optimization algorithm that simulates the processes of natural selection and genetic mechanisms. Through steps such as selection, crossover, and mutation, it can effectively explore the optimization solution space. Combining the energy transfer path loss parameters, the genetic algorithm evaluates and selects energy scheduling paths in each generation of iteration, minimizing the loss when energy is transferred from one base station to another.

[0054] This step ensures the efficiency of the energy scheduling plan, optimizes the energy transfer between base stations, and minimizes losses to the greatest extent. This efficient energy scheduling strategy not only guarantees the energy supply of each base station during high-load periods but also improves the overall system operation efficiency, reduces energy costs, and provides a sustainable solution for future energy scheduling.

[0055] In this step, for the decomposed sub-problems, a path optimization method based on genetic algorithm is adopted to ensure the efficiency of energy scheduling. The genetic algorithm can search for the optimal path in a large solution space by simulating the process of natural selection, including operations such as selection, crossover, and mutation. In the initial stage of this process, the system will randomly generate multiple candidate energy scheduling paths, regarding these paths as "individuals". For example, base station A can choose to transfer energy through base station B or base station C, and these transfer paths can be diverse in the initial stage.

[0056] In each generation of the algorithm's operation, the system will evaluate the fitness of each candidate path according to the total loss of the energy transfer path. The higher the fitness of a path, the lower the energy loss during transmission, and thus it will be preferentially selected. For example, if the transmission loss from base station A to base station B is 10% while the transmission loss from base station A to base station C is 5%, the latter will be considered more advantageous during evaluation. Then, the system will select paths with higher fitness for crossover and mutation operations to generate new candidate scheduling plans.

[0057] After multiple generations of iteration and optimization, the genetic algorithm will gradually converge to the optimal solution. The finally generated preliminary cross-base station energy scheduling scheme will enable each base station to optimize the energy transmission efficiency and minimize energy loss while meeting the load demand. For example, if the path for base station A to transmit energy to base station B through base station C is considered optimal, then the system will preferentially adopt this path for energy scheduling to ensure rapid response and sufficient energy support during high-demand periods.

[0058] For the preliminary cross-base station energy scheduling scheme, a global optimization algorithm based on a consensus mechanism is adopted. Through information exchange and iterative update among distributed nodes, the solutions of each sub-problem are coordinated to generate an optimized allocation strategy that meets the global constraint conditions.

[0059] In this step, the preliminary cross-base station energy scheduling scheme needs to be further optimized by a global optimization algorithm based on a consensus mechanism. The consensus mechanism can ensure that different nodes in the distributed system reach an agreement during information exchange. This means that each virtual base station exchanges information by sharing their energy demands, current loads, and available energy storage data. During this process, each node can influence each other and continuously adjust their energy allocation through iterative update to achieve a more optimized overall allocation strategy.

[0060] By adopting the consensus mechanism, the allocation strategies of different base stations and energy storage units can be coordinated to ensure that the overall system reaches the optimal while meeting the local optimization goals. In this way, not only the efficiency of energy allocation is improved, but also the loss during the energy transmission process can be effectively reduced, thus enhancing the stability and reliability of the entire network. In addition, the real-time update ability provided by the consensus mechanism enables the system to dynamically adapt to load changes, thus achieving a more efficient energy scheduling.

[0061] In the last step, for the preliminary energy scheduling scheme, a global optimization algorithm based on a consensus mechanism is used for further coordination and optimization. As a decentralized method, the consensus mechanism can help different virtual base stations achieve information sharing and consistency during resource allocation. Among the base stations, information such as energy demand, energy storage status, and load prediction can be collected in real time through information exchange. This process can ensure that each base station understands the situation of each other, thus providing the necessary data support for global optimization.

[0062] During this process, each node will perform iterative updates and adjust its own energy allocation strategy based on the previous energy scheduling scheme. Suppose Base Station A initially allocates 80 kWh to Base Station B. However, after information exchange, it is found that the demand of Base Station B is 100 kWh, while Base Station C has a lighter load and only requires 50 kWh. Through the consensus mechanism, each base station can negotiate to reallocate resources to more reasonably meet the load demand. For example, Base Station A can decide to increase the energy allocated to Base Station B to 100 kWh and correspondingly reduce the allocation to Base Station C to enhance the support for Base Station B.

[0063] Finally, through the continuous exchange and collaboration of this information, the system will generate an optimized allocation strategy that meets the global constraint conditions. This strategy not only needs to ensure that the energy requirements of each base station are met, but also takes into account the state of charge of the energy storage unit, the charge and discharge efficiency, and the path loss, so as to achieve efficient cross-base station energy scheduling. For example, the final scheduling scheme may take Base Station A as the main energy output point, while Base Stations B and C serve as load adjustment points, forming an overall system that can operate efficiently.

[0064] This global optimization strategy based on the consensus mechanism can greatly improve energy use efficiency, reduce operating costs, and ensure the stability of the network under extreme conditions. Through distributed decision-making, each base station can quickly respond to changing demands and implement flexible energy allocation, providing strong support for the deployment of future smart grids and renewable energy systems.

[0065] S203, according to the characteristics of the wireless network topology structure in the target area, construct a graph theory-based energy scheduling decision model, conduct delay sensitivity analysis and energy efficiency balance processing on the optimized allocation strategy, optimize the energy storage scheduling timing, and generate a dynamic energy storage scheduling scheme with a fault tolerance mechanism to realize the energy storage for base station management.

[0066] According to the characteristics of the wireless network topology structure in the target area, construct a graph theory-based energy scheduling decision model to achieve efficient energy storage resource management. This model regards the base stations and energy storage units in the target area as nodes in the graph, and the energy scheduling paths between them as edges. By analyzing the path loss parameters and transmission delay data, weights are assigned to the edges, taking into account the losses and delays that may be encountered in the actual energy transmission process. The generated weighted network topology graph will provide a basis for subsequent scheduling optimization to ensure that the transmission delay and loss can be minimized while outputting energy.

[0067] By constructing such a graph theory model, precise control of the energy scheduling process can be achieved, ensuring that high-load base stations can obtain the necessary energy support preferentially. This not only improves the overall energy efficiency of the network but also realizes the intelligent and dynamic energy management, enabling it to quickly adapt to changes in user demands. At the same time, this strategy has a certain fault tolerance mechanism, which can automatically adjust when some paths fail, achieving high reliability and stability of the system and providing guarantee for the smooth operation of the base stations.

[0068] Specifically, based on the characteristics of the wireless network topology structure and the optimization allocation strategy in the target area, an energy scheduling decision-making model based on graph theory can be constructed. Among them, nodes represent base stations or energy storage units, and edges represent energy scheduling paths. By using path loss parameters and transmission delay data, weights are assigned to the edges to generate a weighted network topology graph. Constructing a weighted network topology graph can provide a comprehensive basis for subsequent energy scheduling decisions, helping to ensure quick and efficient decision-making in actual energy transmission. By considering the weights of path loss and transmission delay, it can assist the decision-making system in comprehensively evaluating the economy and timeliness of different paths when selecting energy scheduling paths. This multi-dimensional weight assignment mechanism lays the foundation for optimizing the energy scheduling scheme and provides important support for achieving efficient and flexible resource allocation.

[0069] When constructing an energy scheduling decision-making model based on graph theory, it is first necessary to identify all base stations and their associated energy storage units in the target area. Each base station and energy storage unit can be represented as a node in the graph, and the energy scheduling path between them is represented as an edge. In this process, a geographical information system (GIS) tool can be used to draw the physical layout of the wireless network. For example, assume there are base stations A, B, C and energy storage unit D in a certain city. When constructing, A, B, and C can be regarded as nodes, and D as another node to form a preliminary network topology graph.

[0070] Subsequently, it is necessary to assign weights to each generated edge. This work is crucial because the weights will directly affect subsequent scheduling decisions. The path loss parameter is an important indicator for evaluating energy transmission efficiency and can usually be calculated by combining the physical distance between base stations, the characteristics of the transmission medium (such as radio waves, optical fibers, cables), and environmental interference (such as building blockages, weather conditions, etc.). Assume the distance between base station A and base station B is 1 kilometer, and due to the urban environment and other interferences, the path loss is 8 dB, which means that there will be a certain proportion of energy loss during energy transmission.

[0071] Finally, during the process of assigning weights to each edge, the delay data of the edges also needs to be considered. These delay data can be obtained based on the analysis of historical transmission records and the current network status. Combining the two metrics of path loss parameters and transmission delay, the finally generated weighted network topology map will provide a comprehensive basis for subsequent energy scheduling decisions, ensuring rapid and efficient decision-making in actual energy transmission.

[0072] Specifically, according to the characteristics of the wireless network topology structure in the target area, the base stations and energy storage units can be abstracted as nodes, and the energy scheduling paths in the optimization allocation strategy can be abstracted as edges. Among them, through the physical connection relationship and energy transmission requirements between base stations, the sets of nodes and edges are initialized to generate a preliminary network topology map. In this stage, it is first necessary to comprehensively analyze the wireless network structure in the target area to identify the locations of all base stations and energy storage units. Defining each base station and energy storage unit as a node in the graph can simplify the complex network structure and make it suitable for further algorithm processing. For example, if there are base stations A, B, and C, and energy storage unit D in the target area, A, B, C, and D can be abstracted as four nodes in the graph respectively. Each node represents an entity, and their relationships need to be represented through physical connections.

[0073] Next, the physical connection relationships between base stations are used to construct edges. The forms of connection can be optical fibers, wireless signals, etc. For example, assume that base station A is connected to base station B through an optical fiber, and base station B is connected to base station C through a wireless signal. Then the connection between A and B in the graph will be represented as edge 1, and the connection between B and C will be represented as edge 2. The system initializes the sets of nodes and edges according to the energy transmission requirements and connection characteristics between base stations. At this time, the initial weight of each edge can be assigned according to the type and distance of the connection. For example, an edge with an optical fiber connection may be assigned a lower weight, while an edge with a wireless signal connection may be assigned a higher weight due to greater environmental interference.

[0074] Finally, through the above analysis and processing, the generated preliminary network topology map provides a basic framework for subsequent energy scheduling decisions, ensuring that subsequent steps can be optimized and analyzed based on this map. This graph theory model not only clearly presents the network structure but also provides the necessary data support for subsequent calculations.

[0075] Constructing such a network topology diagram is the basis of the entire energy scheduling decision-making process, which can effectively reflect the connection relationship and energy transmission requirements between base stations and energy storage units in the target area. This model will provide the necessary data support and decision-making basis for subsequent energy allocation optimization. Through the abstract representation of nodes and edges, the mutual relationship between various parts of the system can be quickly understood and analyzed, laying a solid foundation for real-time decision-making. This structure not only improves the scheduling efficiency but also makes it possible to monitor and dynamically adjust complex network conditions in real time.

[0076] For each energy scheduling path, a path loss calculation model based on environmental perception is adopted. Combining the distance between base stations, the characteristics of the transmission medium, and environmental interference factors, the path loss parameters are calculated. Among them, by optimizing the energy transmission efficiency target in the allocation strategy, the path loss parameters are normalized to generate path loss weights. In this step, the system will calculate the loss parameters of each energy scheduling path using environmental perception data. First, obtain the physical distance data between base stations, which can usually be measured through GIS or the existing network layout. For example, the distance between base station A and base station B is 1 kilometer, while the distance between A and C is 2 kilometers. Next, it is necessary to identify the transmission medium used for each path, such as optical fiber or radio wave, because different media have different signal transmission effects. With this information, the system can evaluate the expected loss. For example, assume that the loss of a wireless signal due to building blockage in an urban environment is 8 dB, while the loss of a fiber optic connection is only 2 dB.

[0077] After calculating the path loss, the system will normalize these loss parameters. The purpose of normalization is to convert different loss values into relative values so that they can be compared with other key indicators (such as transmission delay). For example, if the path loss values are 5 dB and 10 dB, they will be adjusted to values between 0 and 1 after normalization, which is convenient for subsequent processing. Finally, the generated path loss weights will provide key information for subsequent energy scheduling decisions. In this way, the loss performance of different paths can be compared with a unified standard, ensuring the rationality and effectiveness in the decision-making process. For example, in the decision-making process, if the program finds that the path loss weight from base station A to base station B is significantly lower than that from A to C, then the system will give priority to considering energy scheduling through B.

[0078] Through this calculation process, we can not only accurately understand the performance of each energy scheduling path in the actual transmission process, but also optimize the efficiency of different paths. This path loss calculation model based on environment perception provides a scientific basis for energy scheduling, allowing operators to formulate reasonable energy scheduling strategies according to specific environments and conditions. By normalizing the path loss weights, it helps to improve the flexibility and adaptability of energy scheduling and ensure efficient energy transmission in complex environments.

[0079] According to the real-time scheduling requirements in the optimization allocation strategy, combined with the historical transmission delay data and the current network status, a delay prediction method based on time series analysis is used to dynamically update the transmission delay data of each energy scheduling path and use it as the delay weight; At this stage, the system will combine historical transmission delay data and current network status to implement delay prediction. First, it is necessary to collect and store past transmission delay data, including delay records of each path in different time periods. This data is crucial for discovering delay patterns and trends. Assume that during past peak hours, the transmission delay from base station A to base station B is usually between 300 milliseconds and 400 milliseconds, while during non-peak hours it is 80 milliseconds to 150 milliseconds. Through these historical data, the system can identify delay characteristics under high load conditions.

[0080] Next, a method based on time series analysis (such as ARIMA or LSTM model) is used to predict the transmission delay data of each path. By inputting historical delay data into the model, the system can predict the delay changes of the path in the future. For example, when the system recognizes that a weekend is approaching and historical data shows that the traffic volume during this period is usually large, the system will adjust the predicted value of the transmission delay accordingly.

[0081] Finally, the transmission delay data of each energy scheduling path is dynamically updated, and the updated delay data is incorporated into the subsequent energy scheduling decision as a new delay weight. This ensures that the latest network status is fully considered during energy scheduling, thereby effectively coping with transmission delay issues under high load conditions and ensuring the timeliness and effectiveness of responding to demand.

[0082] This dynamically updated delay prediction mechanism can significantly enhance the system's ability to respond to changes in network status. This approach ensures that the network can maintain efficient and stable energy delivery even when demand fluctuates dramatically, reducing the degradation of service quality that may be caused by delays. In addition, timely delay data updates provide decision makers with real-time and accurate information, making energy scheduling plans more flexible and dynamic, and adapting to complex and changing network environments.

[0083] Perform multi-objective optimization and fusion of path loss weight and transmission delay weight, adopt a comprehensive weight assignment method based on the weighted summation method, and combine the priority and fault tolerance requirements of the energy scheduling path to assign comprehensive weights to the edges in the preliminary network topology diagram to generate a weighted network topology diagram.

[0084] In the last step, the system performs multi-objective optimization and fusion of path loss weight and transmission delay weight. First, set the relative importance of each component weight. For example, in some service scenarios, ensuring low latency may be more important than reducing loss, so a higher proportion can be given to the delay weight. Suppose the weight factor of the loss weight is set to 0.4, and the factor of the delay weight is 0.6. Then, in the comprehensive calculation, the final weight of each path will be determined by the combination of these two factors.

[0085] Next, combining the priority and fault tolerance requirements of the energy scheduling path, the system will assign comprehensive weights to each edge, which can ensure that the optimal path is selected in a complex scheduling environment. For example, although an edge performs well in terms of transmission loss, it may be given a lower priority due to its weak fault tolerance mechanism, while another edge with slightly higher loss but strong fault tolerance may be given a higher priority.

[0086] Finally, the generated weighted network topology diagram will become the basis for determining the energy scheduling path. The way of assigning comprehensive weights ensures that information in multiple dimensions is effectively integrated, making the decision-making process more scientific and reasonable. This optimized network topology diagram can effectively support energy scheduling decisions, improve the overall efficiency and reliability of the system, and ensure that reliable energy support can be provided in a timely manner to meet the needs of the base station in the face of complex situations.

[0087] By performing multi-objective optimization and fusion of path loss weight and transmission delay weight, the system can not only improve the intelligence and flexibility of selecting the scheduling path, but also better meet the energy delivery requirements under high load conditions. This method of assigning comprehensive weights enables the characteristics of different paths to be fully utilized, thereby improving the overall efficiency and reliability of the system. In addition, this process provides strong data support for real-time decision-making, ensuring that the network can be flexibly adjusted in the face of dynamic changes and guaranteeing the service quality of base station users.

[0088] For the energy scheduling paths in the network topology diagram, adopt a delay sensitivity analysis method based on the shortest path algorithm to calculate the transmission delay of each path, and screen out the delay-sensitive critical paths through the delay tolerance threshold; At this stage, through the constructed weighted network topology graph, the shortest path algorithm is applied to conduct delay sensitivity analysis on each energy scheduling path. Specifically, starting from the source node, according to the transmission delay weights of each path, the shortest transmission delay to reach the target node is calculated. For example, assuming that base station A needs to send energy to base station B, the paths can be A - B or A - C - B. Then, the calculation process needs to consider the total delay of each path to determine which path is the most superior under the current state. After calculating the transmission delays of all paths, the system will use the set delay tolerance threshold to screen out those delay-sensitive critical paths. These critical paths refer to the paths that have high delay sensitivity while meeting the energy transmission requirements. For example, if the transmission delay of a certain path exceeds the tolerance threshold, then this path will be marked as not suitable for use and should be replaced by a path with a lower delay. Therefore, this analysis not only ensures the response speed under high load conditions but also avoids the problem of low energy distribution efficiency caused by delays.

[0089] By implementing delay sensitivity analysis, the accuracy of energy scheduling path selection can be improved, ensuring that the selected scheduling path has good timeliness during actual operation. This is of great significance for ensuring the energy requirements of high-load base stations and helps the entire network to quickly respond and conduct energy allocation in the face of sudden user growth. Through this method, a good balance can be achieved between energy efficiency and stability, thus better serving the user needs in the target area.

[0090] At this stage, according to the constructed weighted network topology graph, the shortest path algorithm is applied for delay sensitivity analysis. The shortest path algorithm, especially Dijkstra's algorithm, can effectively help us find the optimal path from the source node to the target node. For example, assuming that base station A needs to transmit energy to base station B, the system will calculate the total delay of each path by analyzing all available paths from A to B to determine the fastest energy transmission method. This process not only involves the length of the path but also the weights of each edge in the selected path (i.e., transmission loss and delay).

[0091] After the calculation is completed, the system will obtain the transmission delay results of all possible paths. Next, screening needs to be carried out according to the set delay tolerance threshold. For example, if the set threshold is 500 milliseconds and the transmission delay of a certain path is greater than this value, the system will mark it as unqualified. Through this screening, it can be ensured that the selected path meets the energy transmission while avoiding the degradation of network performance caused by excessive delay. For example, assuming that the delay of the main path from A to B is 450 milliseconds, while the delay of an alternative path reaches 600 milliseconds, then it is more reasonable to use the first path.

[0092] Through this step, the successfully screened key paths sensitive to time delay will provide important references for subsequent energy scheduling optimization. Capturing these sensitive paths can ensure that the system can respond faster when selecting energy transmission paths during peak loads or emergencies, and can meet the load requirements of the base stations to the greatest extent.

[0093] For the key paths sensitive to time delay, an energy efficiency optimization method based on load balancing is adopted. Combining the dynamic load prediction results of the base stations and the charge-discharge efficiency curves of the energy storage units, the allocation ratio of the energy scheduling paths is adjusted to ensure that high-load base stations obtain energy support first, and an energy efficiency balanced scheduling scheme is generated. In this process, for the identified key paths sensitive to time delay, the system will apply an energy efficiency optimization method based on load balancing to adjust the energy scheduling scheme. The first step is to determine which base stations have energy requirements during peak hours according to the dynamic load prediction results of the base stations. Suppose the predicted load of base station A is 150 kWh, while that of base station B is 80 kWh. To optimize energy allocation, the system will first ensure that base station A can obtain sufficient energy in a timely manner. Next, combined with the charge-discharge efficiency curves of the energy storage units, the available energy scheduling paths are further analyzed, and the allocation ratio of each path is adjusted accordingly. For example, if the energy storage unit of base station C has a high charging efficiency and is close to base station A, then the system will tend to allocate more energy from base station C to base station A to reduce transmission losses. Through such allocation, it is ensured that the needs of high-load base stations can be maximally met, and the overall energy efficiency is optimized. Finally, the generated energy efficiency balanced scheduling scheme will fully consider the load conditions of each base station and ensure the effective utilization of the energy scheduling paths at the same time. This dynamic adjustment strategy ensures flexibility and efficiency in responding to sudden demands and helps the stable operation of the overall network.

[0094] Through the energy efficiency optimization method based on load balancing, not only can it ensure that high-load base stations obtain the required energy support, but also can improve the energy use efficiency of the entire system. This optimized scheduling scheme helps to reduce waste in energy scheduling and ensures that the system can maintain dynamic balance in an environment with frequent load fluctuations. Ultimately, a flexible and efficient energy scheduling mechanism is achieved, which can better meet complex network requirements.

[0095] In this step, for the identified key paths sensitive to time delay, an energy efficiency optimization method based on load balancing is adopted. Before performing energy allocation, the system first needs to obtain the dynamic load prediction results of the base stations. Suppose during a certain period, the predicted load of base station A is 200 kWh, while those of base stations B and C are 100 kWh and 150 kWh respectively. Since the load demand of base station A is the largest, it must be ensured that it can obtain the required energy in a timely manner during energy allocation.

[0096] Next, the system analyzes the current state of each energy storage unit in combination with the charge-discharge efficiency curve of the energy storage unit to perform more efficient energy scheduling. For example, if the charging efficiency of energy storage unit D is 90%, then during the energy scheduling process from D to A, B, and C, the system will consider a relatively high scheduling ratio from D to A while ensuring that the demands of B and C can also be met. In this way, the system will dynamically adjust the allocation ratios of each energy scheduling path to ensure that high-load base stations obtain energy support first and minimize energy losses.

[0097] Finally, considering the above factors, the generated energy efficiency balanced scheduling scheme will ensure the energy support for base stations under high-load conditions and play a positive role in optimizing the overall energy efficiency of the system. This flexible scheduling strategy not only improves the network's response ability but also provides an efficient solution for future possible load changes.

[0098] For the energy efficiency balanced scheduling scheme, introduce a fault tolerance mechanism based on redundant paths. By constructing backup energy scheduling paths and dynamic switching strategies, ensure that energy transmission can still be maintained when some paths fail, and finally generate a dynamic energy storage scheduling scheme with a fault tolerance mechanism.

[0099] In this step, for the previously generated energy efficiency balanced scheduling scheme, introduce a fault tolerance mechanism for redundant paths. First, by analyzing the current network topology, identify feasible redundant paths and use them as backup scheduling paths. For example, if the main scheduling path from base station A to base station B fails, a backup path from base station A through base station C and finally to base station B can be set up. The key to establishing these backup paths is to be able to quickly identify the status of the main path and seamlessly switch to the backup path when a failure occurs. Then, corresponding dynamic switching strategies need to be formulated. This strategy should be able to monitor the status of the transmission path in real time and determine whether to continue using the main path or switch to the backup path. For example, when it is detected that the transmission delay of the main path exceeds the set threshold, the system will automatically switch to the backup path to ensure that energy can reach the destination in time. Through this dynamic switching mechanism, the scheduling scheme can be quickly adjusted when the network is in an abnormal state to ensure the uninterruptedness and stability of the overall energy transmission. Finally, a dynamic energy storage scheduling scheme with a perfect fault tolerance mechanism is constructed, which can not only ensure energy transmission when the main path fails but also improve the overall reliability of the system and provide guarantee for the continuous operation of the base station. The introduction of this mechanism provides stronger flexibility and security for energy allocation in complex networks.

[0100] Introducing a fault-tolerant mechanism with redundant paths can significantly enhance the robustness and flexibility of the entire energy scheduling system. Even in the face of equipment failures or network anomalies, the system can still ensure the effective transmission of energy and reduce the risk of service interruption caused by single-point failures. In addition, the implementation of dynamic switching strategies ensures the real-time nature of energy policies, contributing to improving the resource utilization rate and user experience of the entire network. This solution not only enhances the resilience of the system but also provides a solid technical foundation for future large-scale network deployments.

[0101] In this step, by introducing a fault-tolerant mechanism based on redundant paths, the reliability of the energy efficiency balanced scheduling scheme is increased. First, it is necessary to analyze the existing network topology and identify available alternative scheduling paths. This analysis aims to equip each main scheduling path with one or more corresponding alternative paths. For example, if the main scheduling path from base station A to base station B fails, the system should be able to quickly switch to the alternative path from A through base station C to B. This process ensures the continuity of energy transmission and reduces the risk of service interruption caused by a single path failure.

[0102] Next, it is necessary to formulate a dynamic switching strategy to monitor the status of the current transmission path in real time. If the system detects that the delay of the main path exceeds the set threshold or a certain path fails, the alternative path is immediately activated. For example, when there is a signal packet loss in the transmission path of base station A, the system will automatically switch to the alternative path. This mechanism ensures the flexibility and adaptability of the network and helps to continuously maintain efficient energy transmission in a complex network environment.

[0103] Finally, the dynamic energy storage scheduling scheme with a fault-tolerant mechanism generated in this stage can not only effectively handle sudden failures of the main path but also improve the stability and reliability of the overall system. In actual operation, it ensures that energy can be flexibly scheduled between different paths, ensuring the smooth operation of base stations under high load conditions, thereby enhancing the service quality and user experience of the entire wireless network.

[0104] It can be seen that, based on the historical traffic data, real-time user movement pattern data, and meteorological data of each virtual base station in the target area, multi-modal data fusion processing is carried out to obtain the dynamic load prediction results of each virtual base station during the target period; based on the dynamic load prediction results, a distributed constraint optimization algorithm is used for dynamic allocation of energy storage resources to generate an optimized allocation strategy including cross-base station energy scheduling paths; according to the characteristics of the wireless network topology in the target area, a graph theory-based energy scheduling decision model is constructed to optimize the energy storage scheduling timing and generate a dynamic energy storage scheduling scheme with a fault-tolerant mechanism to realize energy storage for base station management, thereby being able to improve the energy efficiency of base stations, reduce energy waste, and achieve efficient configuration and real-time scheduling of energy storage resources.

[0105] Another embodiment of the present invention provides an energy storage system for base station management. Refer to Figure 3 , the system may include: A fusion module 301, configured to perform multi-modal data fusion processing based on the historical traffic data, real-time user movement pattern data, and meteorological data of each virtual base station in the target area, and obtain the dynamic load prediction results of each virtual base station in the target period; An allocation module 302, configured to perform dynamic allocation of energy storage resources based on the dynamic load prediction results, combine the current state of charge, charge and discharge efficiency curve of the energy storage unit associated with each virtual base station, and the energy transfer path loss parameters between adjacent base stations, and use a distributed constraint optimization algorithm to generate an optimized allocation strategy including the cross-base station energy scheduling path; A construction module 303, configured to construct a graph theory-based energy scheduling decision model according to the wireless network topology structure characteristics of the target area, perform delay sensitivity analysis and energy efficiency balance processing on the optimized allocation strategy, optimize the energy storage scheduling timing, and generate a dynamic energy storage scheduling scheme with a fault tolerance mechanism to achieve energy storage for base station management.

[0106] It can be seen that, according to the historical traffic data, real-time user movement pattern data, and meteorological data of each virtual base station in the target area, perform multi-modal data fusion processing to obtain the dynamic load prediction results of each virtual base station in the target period; based on the dynamic load prediction results, use a distributed constraint optimization algorithm to perform dynamic allocation of energy storage resources and generate an optimized allocation strategy including the cross-base station energy scheduling path; according to the wireless network topology structure characteristics of the target area, construct a graph theory-based energy scheduling decision model, optimize the energy storage scheduling timing, and generate a dynamic energy storage scheduling scheme with a fault tolerance mechanism to achieve energy storage for base station management, thereby being able to improve the energy efficiency of the base station, reduce energy waste, and achieve efficient configuration and real-time scheduling of energy storage resources.

[0107] The embodiment of the present invention further provides a storage medium, in which a computer program is stored, and the computer program is configured to execute the steps in any one of the above method embodiments when running.

[0108] Specifically, in this embodiment, the above storage medium may be configured to store a computer program for executing the following steps: S201, according to the historical traffic data, real-time user movement pattern data, and meteorological data of each virtual base station in the target area, perform multi-modal data fusion processing to obtain the dynamic load prediction results of each virtual base station in the target period; S202. Based on the dynamic load prediction results, in combination with the current state of charge of the energy storage units associated with each virtual base station, the charge and discharge efficiency curves, and the energy transmission path loss parameters between adjacent base stations, a distributed constrained optimization algorithm is used to dynamically allocate energy storage resources, generating an optimized allocation strategy that includes cross-base station energy scheduling paths. S203. According to the characteristics of the wireless network topology in the target area, a graph theory-based energy scheduling decision model is constructed to perform delay sensitivity analysis and energy efficiency balance processing on the optimized allocation strategy, optimizing the energy storage scheduling timing, and generating a dynamic energy storage scheduling scheme with a fault tolerance mechanism to achieve energy storage for base station management.

[0109] It can be seen that according to the historical traffic data, real-time user movement pattern data, and meteorological data of each virtual base station in the target area, multi-modal data fusion processing is performed to obtain the dynamic load prediction results of each virtual base station during the target period; based on the dynamic load prediction results, a distributed constrained optimization algorithm is used to dynamically allocate energy storage resources, generating an optimized allocation strategy that includes cross-base station energy scheduling paths; according to the characteristics of the wireless network topology in the target area, a graph theory-based energy scheduling decision model is constructed to optimize the energy storage scheduling timing, generating a dynamic energy storage scheduling scheme with a fault tolerance mechanism to achieve energy storage for base station management, thereby being able to improve the energy efficiency of the base station, reduce energy waste, and achieve efficient configuration and real-time scheduling of energy storage resources.

[0110] An embodiment of the present invention further provides an electronic device, including a memory and a processor, where a computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0111] Specifically, the above electronic device may further include a transmission device and an input / output device, where the transmission device is connected to the above processor, and the input / output device is connected to the above processor.

[0112] Specifically, in this embodiment, the above processor may be configured to execute the following steps through a computer program: S201. According to the historical traffic data, real-time user movement pattern data, and meteorological data of each virtual base station in the target area, multi-modal data fusion processing is performed to obtain the dynamic load prediction results of each virtual base station during the target period; S202. Based on the dynamic load prediction results, in combination with the current state of charge of the energy storage units associated with each virtual base station, the charge and discharge efficiency curves, and the energy transmission path loss parameters between adjacent base stations, a distributed constrained optimization algorithm is used to dynamically allocate energy storage resources, generating an optimized allocation strategy that includes cross-base station energy scheduling paths; S203. Based on the characteristics of the wireless network topology in the target area, construct an energy scheduling decision model based on graph theory, conduct delay sensitivity analysis and energy efficiency balance processing on the optimized allocation strategy, optimize the energy storage scheduling timing, and generate a dynamic energy storage scheduling scheme with a fault tolerance mechanism to achieve energy storage for base station management.

[0113] It can be seen that by performing multi-modal data fusion processing on the historical traffic data, real-time user movement pattern data, and meteorological data of each virtual base station in the target area, the dynamic load prediction results of each virtual base station in the target time period are obtained; based on the dynamic load prediction results, a distributed constraint optimization algorithm is used for dynamic allocation of energy storage resources to generate an optimized allocation strategy including cross-base station energy scheduling paths; according to the characteristics of the wireless network topology in the target area, an energy scheduling decision model based on graph theory is constructed to optimize the energy storage scheduling timing and generate a dynamic energy storage scheduling scheme with a fault tolerance mechanism to achieve energy storage for base station management, thereby being able to improve the energy efficiency of the base station, reduce energy waste, and achieve efficient configuration and real-time scheduling of energy storage resources.

[0114] The above has detailed the structure, characteristics, and function effects of the present invention according to the illustrated embodiments. The above is only the preferred embodiment of the present invention, but the present invention is not limited to the scope defined by the drawings. Any changes made according to the concept of the present invention, or modified into equivalent embodiments with equivalent changes, still within the spirit covered by the specification and the drawings, shall be within the protection scope of the present invention.

Claims

1. An energy storage method for base station management, characterized in that: The method comprises: Based on the historical traffic data, real-time user mobility pattern data and meteorological data of each virtual base station in the target area, multi-modal data fusion processing is performed to obtain the dynamic load prediction results of each virtual base station in the target period; Based on the dynamic load prediction results, combined with the current charge state of the energy storage unit associated with each virtual base station, the charging and discharging efficiency curve, and the energy transmission path loss parameters between adjacent base stations, a distributed constraint optimization algorithm is used to dynamically allocate energy storage resources and generate an optimized allocation strategy that includes cross-base station energy scheduling paths; According to the wireless network topology characteristics of the target area, a graph-theory-based energy scheduling decision model is constructed. The delay sensitivity analysis and energy efficiency balancing processing are performed on the optimization allocation strategy, the energy storage scheduling sequence is optimized, and a dynamic energy storage scheduling scheme with a fault-tolerant mechanism is generated to realize energy storage for base station management.

2. The method according to claim 1, characterized in that The multi-modal data fusion processing is performed based on the historical traffic volume data, real-time user mobility pattern data and meteorological data of each virtual base station in the target area to obtain the dynamic load prediction result of each virtual base station in the target period, including: According to the historical traffic data, real-time user mobility pattern data and meteorological data of each virtual base station in the target area, a multi-source data alignment algorithm based on dynamic time warping is used to eliminate the problem of inconsistent data timestamps and interpolate missing values ​​to obtain a time-synchronized multimodal data set. For time-synchronized multimodal datasets, a feature extraction network based on a multi-head attention mechanism is used to extract the temporal features of historical traffic, the spatial features of user mobility patterns, and the environmental features of meteorological data. The features of different modalities are weightedly fused through a cross-modal attention mechanism to generate a fused multimodal feature representation. The fused multimodal feature representation is input into the pre-trained prediction model based on the spatiotemporal graph convolutional network. Combined with the base station topology structure of the target area, the spatiotemporal dependency between base stations is captured, and the dynamic load change trend of each virtual base station in the target period is predicted. The dynamic load prediction result of each virtual base station in the target period is obtained.

3. The method according to claim 2, characterized in that Based on the dynamic load prediction results, combined with the current state of charge of the energy storage unit associated with each virtual base station, the charging and discharging efficiency curve and the energy transmission path loss parameters between adjacent base stations, a distributed constraint optimization algorithm is used to dynamically allocate energy storage resources, and an optimized allocation strategy including an energy scheduling path across base stations is generated, including: According to the current state of charge of the energy storage unit associated with each virtual base station, the charge and discharge efficiency curve, and the energy transmission path loss parameters between adjacent base stations, a multi-dimensional constraint model of the energy storage state is constructed, and the feasible solution space of energy storage resource allocation is defined through energy conservation constraints, charge and discharge rate constraints, and transmission loss constraints. The dynamic load prediction results are combined with the multi-dimensional constraint model of energy storage status, and a distributed optimization framework based on Lagrangian relaxation method is adopted to decompose the global optimization problem of energy storage resource allocation into multiple sub-problems. Each sub-problem corresponds to a local optimization goal of a virtual base station. The local optimization goals include load balancing, energy transmission efficiency and energy storage life. For the decomposed sub-problems, a path optimization method based on genetic algorithm is used, combined with the energy transmission path loss parameters between adjacent base stations, to dynamically search for the optimal energy scheduling path and generate a preliminary cross-base station energy scheduling solution; For the preliminary cross-base station energy scheduling scheme, a global optimization algorithm based on the consensus mechanism is adopted. Through information exchange and iterative update among distributed nodes, the solutions of each sub-problem are coordinated to generate an optimal allocation strategy that meets the global constraints.

4. The method according to claim 3, characterized in that According to the wireless network topology characteristics of the target area, an energy scheduling decision model based on graph theory is constructed, delay sensitivity analysis and energy efficiency balancing processing are performed on the optimization allocation strategy, the energy storage scheduling sequence is optimized, and a dynamic energy storage scheduling scheme with a fault-tolerant mechanism is generated to realize energy storage for base station management, including: According to the wireless network topology characteristics and optimization allocation strategy of the target area, an energy scheduling decision model based on graph theory is constructed, in which nodes represent base stations or energy storage units, edges represent energy scheduling paths, and weights are assigned to edges through path loss parameters and transmission delay data to generate a weighted network topology graph. For the energy scheduling paths in the network topology, a delay sensitivity analysis method based on the shortest path algorithm is used to calculate the transmission delay of each path, and the delay-sensitive critical paths are screened out through the delay tolerance threshold. For the critical paths that are sensitive to latency, we use an energy efficiency optimization method based on load balancing. We combine the dynamic load prediction results of the base station and the charging and discharging efficiency curve of the energy storage unit to adjust the allocation ratio of the energy scheduling path, ensure that high-load base stations get energy support first, and generate an energy-efficient scheduling plan. For energy efficiency balanced scheduling schemes, a fault-tolerant mechanism based on redundant paths is introduced. By constructing backup energy scheduling paths and dynamic switching strategies, it is ensured that energy transmission can be maintained when some paths fail, and finally a dynamic energy storage scheduling scheme with a fault-tolerant mechanism is generated.

5. The method according to claim 4, characterized in that According to the wireless network topology characteristics and optimization allocation strategy of the target area, an energy scheduling decision model based on graph theory is constructed, wherein nodes represent base stations or energy storage units, edges represent energy scheduling paths, and weights are assigned to edges through path loss parameters and transmission delay data to generate a weighted network topology graph, including: According to the wireless network topology characteristics of the target area, the base stations and energy storage units are abstracted as nodes, and the energy scheduling paths in the optimization allocation strategy are abstracted as edges. The nodes and edges are initialized through the physical connection relationship between base stations and the energy transmission requirements to generate a preliminary network topology diagram. For each energy scheduling path, a path loss calculation model based on environment perception is used to calculate the path loss parameters in combination with the distance between base stations, transmission medium characteristics and environmental interference factors. The path loss parameters are normalized by optimizing the energy transmission efficiency target in the allocation strategy to generate the path loss weight. According to the real-time scheduling requirements in the optimization allocation strategy, combined with the historical transmission delay data and the current network status, a delay prediction method based on time series analysis is used to dynamically update the transmission delay data of each energy scheduling path and use it as the delay weight; The path loss weight and transmission delay weight are integrated for multi-objective optimization. A comprehensive weight assignment method based on the weighted sum method is adopted. Combined with the priority and fault tolerance requirements of the energy scheduling path, comprehensive weights are assigned to the edges in the preliminary network topology diagram to generate a weighted network topology diagram.

6. An energy storage system for base station management, characterized in that: The system comprises: A fusion module is used to perform multi-modal data fusion processing based on the historical traffic data, real-time user mobility pattern data and meteorological data of each virtual base station in the target area to obtain the dynamic load prediction results of each virtual base station in the target period; The allocation module is used to dynamically allocate energy storage resources based on the dynamic load prediction results, combined with the current charge state of the energy storage units associated with each virtual base station, the charge and discharge efficiency curve, and the energy transmission path loss parameters between adjacent base stations using a distributed constraint optimization algorithm to generate an optimized allocation strategy that includes an energy scheduling path across base stations; The building module is used to construct an energy scheduling decision model based on graph theory according to the wireless network topology characteristics of the target area, perform delay sensitivity analysis and energy efficiency balancing on the optimization allocation strategy, optimize the energy storage scheduling timing, and generate a dynamic energy storage scheduling plan with a fault-tolerant mechanism to realize energy storage for base station management.

7. The system according to claim 6, characterized in that The fusion module is specifically used for: According to the historical traffic data, real-time user mobility pattern data and meteorological data of each virtual base station in the target area, a multi-source data alignment algorithm based on dynamic time warping is used to eliminate the problem of inconsistent data timestamps and interpolate missing values ​​to obtain a time-synchronized multimodal data set. For time-synchronized multimodal datasets, a feature extraction network based on a multi-head attention mechanism is used to extract the temporal features of historical traffic, the spatial features of user mobility patterns, and the environmental features of meteorological data. The features of different modalities are weightedly fused through a cross-modal attention mechanism to generate a fused multimodal feature representation. The fused multimodal feature representation is input into the pre-trained prediction model based on the spatiotemporal graph convolutional network. Combined with the base station topology structure of the target area, the spatiotemporal dependency between base stations is captured, and the dynamic load change trend of each virtual base station in the target period is predicted. The dynamic load prediction result of each virtual base station in the target period is obtained.

8. The system according to claim 7, characterized in that The allocation module is specifically used for: According to the current state of charge of the energy storage unit associated with each virtual base station, the charge and discharge efficiency curve, and the energy transmission path loss parameters between adjacent base stations, a multi-dimensional constraint model of the energy storage state is constructed, and the feasible solution space of energy storage resource allocation is defined through energy conservation constraints, charge and discharge rate constraints, and transmission loss constraints. The dynamic load prediction results are combined with the multi-dimensional constraint model of energy storage status, and a distributed optimization framework based on Lagrangian relaxation method is adopted to decompose the global optimization problem of energy storage resource allocation into multiple sub-problems. Each sub-problem corresponds to a local optimization goal of a virtual base station. The local optimization goals include load balancing, energy transmission efficiency and energy storage life. For the decomposed sub-problems, a path optimization method based on genetic algorithm is used, combined with the energy transmission path loss parameters between adjacent base stations, to dynamically search for the optimal energy scheduling path and generate a preliminary cross-base station energy scheduling solution; For the preliminary cross-base station energy scheduling scheme, a global optimization algorithm based on the consensus mechanism is adopted. Through information exchange and iterative update among distributed nodes, the solutions of each sub-problem are coordinated to generate an optimal allocation strategy that meets the global constraints.

9. A storage medium, characterized in that: The storage medium stores a computer program, wherein the computer program is configured to execute the method according to any one of claims 1 to 5 when executed.

10. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to run the computer program to perform the method according to any one of claims 1 to 5.

Images