Power distribution network differential protection service security collaboration method and system for 5G network slice

Abstract

The invention relates to a 5G network slice-oriented power distribution network differential protection service security collaboration method and system, and the method comprises the steps: constructing a cloud-side-end communication architecture which introduces 5G user plane function sinking, and achieving the security isolation through independently deploying a special network unit of a power grid. On the basis, a three-level collaborative optimization mechanism is adopted: multi-MEC node load balancing is realized based on terminal access balance control of a genetic algorithm; based on edge resource dynamic allocation of deep reinforcement learning, ultra-low time delay guarantee is provided for differential protection service through a time delay demand matching degree award mechanism; based on cross-domain routing planning of an improved ant colony algorithm, dual optimization of low time delay and load balance is realized in combination with pheromone concentration and a bandwidth balance factor. The technical problem that static network slice resource allocation is not flexible and low-delay requirements are mutually restricted is solved, end-to-end deterministic guarantee of differential protection service is achieved, and the reliability of power distribution network protection is remarkably improved.

Description

Technical Field

[0001] This application relates to the field of power grid security technology, and in particular to a method and system for security collaboration of differential protection services in distribution networks for 5G network slicing. Background Technology

[0002] With the rapid development of new power systems based on new energy sources, the penetration rate of distributed generation in distribution networks continues to increase, transforming distribution networks from passive to active networks. This has resulted in complex and variable power flow distributions and increasingly complex fault characteristics. Against this backdrop, differential protection based on current phase and amplitude comparison has become an important development direction for distribution network protection due to its absolute selectivity.

[0003] Currently, the industry is exploring the use of 5G network slicing technology to support differential protection services. The main technical approach involves constructing an end-to-end logically isolated private network on the physical network using network function virtualization (NFV) and software-defined networking (SDN) technologies. Operators allocate dedicated network slices for power differential protection services. These slices, based on pre-allocated bandwidth resources and fixed quality of service (QoS) policies, provide a relatively prioritized transmission channel for differential protection data. In terms of security, a unified encryption and authentication mechanism is employed, with data transmission protected through a security gateway. This approach, to a certain extent, meets the basic communication requirements of differential protection services.

[0004] Regarding the aforementioned technologies, network slicing with static resource configuration cannot adapt to dynamic fluctuations in traffic volume. When differential protection services experience a sudden surge, fixed bandwidth resources cannot guarantee service quality, while during periods of service idleness, resources are wasted, affecting the use of differential protection services.

[0005] Based on this, this application provides a method and system for security collaboration of differential protection services in distribution networks oriented towards 5G network slicing. Summary of the Invention

[0006] To address the issue that statically configured network slices cannot adapt to dynamic fluctuations in traffic volume, and that fixed bandwidth resources cannot guarantee service quality when differential protection services experience sudden increases, while resources are wasted during idle periods, thus affecting the use of differential protection services, this application provides a method and system for secure coordination of differential protection services in distribution networks based on 5G network slices.

[0007] Firstly, this application provides a security coordination method for distribution network differential protection services oriented towards 5G network slicing, employing the following technical solution: including: The user plane functions of the 5G core network are pushed down to the power mobile edge computing nodes, forming a communication architecture consisting of power terminals, power mobile edge computing nodes, and a power cloud platform; when a differential protection service is initiated, the SDN controller receives a differential protection service slice request containing quality of service requirements; Based on the SDN controller, with the optimization objective of minimizing the variance of bandwidth resource utilization of all power mobile edge computing nodes in the region, a genetic algorithm is used to calculate the optimal access strategy for the differential protection terminal that initiates the request. The fitness function of the genetic algorithm is based on node bandwidth constraints and signal-to-noise ratio constraints. After the differential protection terminal accesses the target power mobile edge computing node according to the access strategy, based on the target power mobile edge computing node and with the optimization objective of maximizing the matching degree of network slice communication requirements, the DDPG algorithm is used to dynamically allocate bandwidth and computing resources to the internal differential protection service slices; After the target power mobile edge computing node completes the resource allocation for the differential protection service slice, based on the SDN controller with the optimization objectives of minimizing path latency and balancing the entire network load, an improved ant colony algorithm with an introduced bandwidth balancing factor is adopted to plan the optimal forwarding route from the target power mobile edge computing node to the power cloud platform for the differential protection service slice. The pheromone parameter of the improved ant colony algorithm is designed according to the link latency, and its path selection probability is guided by both the pheromone concentration and the bandwidth balancing factor.

[0008] Preferably, the user plane functions of the 5G core network are pushed down to the power mobile edge computing node, forming a communication architecture consisting of power terminals, power mobile edge computing nodes, and a power cloud platform; when a differential protection service is initiated, the SDN controller receives a differential protection service slice request containing quality of service requirements, including: The user plane function network elements of the 5G core network are deployed in the power mobile edge computing nodes controlled by the power grid company, enabling the power mobile edge computing nodes to have local data forwarding and processing capabilities; In the aforementioned communication architecture, a control unit and user plane unit of the wireless access network are deployed independently for power services, achieving physical or logical isolation from public user services; The power terminal establishes a connection with the power mobile edge computing node via a 5G wireless air interface; the power mobile edge computing node connects to the power cloud platform via a dedicated power communication network or operator network. When a differential protection service is initiated, the SDN controller deployed in the communication architecture receives a differential protection service slice request that includes end-to-end latency requirements, bandwidth requirements, and reliability metrics.

[0009] Preferably, the SDN-based controller optimizes the access strategy for the requesting differential protection terminal by using a genetic algorithm to calculate the optimal access strategy for the power mobile edge computing node with the objective of minimizing the variance of bandwidth resource utilization of all power mobile edge computing nodes within the region, including: Based on the SDN controller, the real-time bandwidth resource status of each power mobile edge computing node in the area, the channel quality information between each differential protection terminal and each power mobile edge computing node, and the total amount of available 5G resource blocks in the area are obtained; The access relationship between the differential protection terminal and the power mobile edge computing node is encoded using chromosomes to form an initial population; The fitness function is used to evaluate each individual in the population. The fitness function is configured as follows: based on the evaluation of the variance of bandwidth resource utilization, a first penalty term is applied to the terminal access scheme that does not meet the signal-to-noise ratio threshold, and a second penalty term is applied to the access scheme that causes the total bandwidth of the power mobile edge computing node to exceed the limit. The first penalty term: When the signal-to-noise ratio (SNR) between the differential protection terminal and the target power mobile edge computing node is lower than a preset SNR threshold, a preset first penalty value is added to the fitness function. The magnitude of the first penalty value is positively correlated with the degree to which the SNR is lower than the threshold. The second penalty term: When the access of the differential protection terminal causes the total bandwidth demand of the power mobile edge computing node to exceed the maximum bandwidth capacity of the node, a preset second penalty value is added to the fitness function. The magnitude of the second penalty value is positively correlated with the degree to which the bandwidth exceeds the limit. The population is iteratively evolved through selection, crossover, and mutation operations to gradually optimize the access strategy. When the preset number of iterations or the fitness function converges, the optimal differential protection terminal access strategy is output. The access strategy specifies the best power mobile edge computing node that each differential protection terminal should connect to.

[0010] Preferably, the step of iteratively evolving the population through selection, crossover, and mutation operations to gradually optimize the access strategy, and outputting the optimal differential protection terminal access strategy when a preset number of iterations or fitness function convergence is reached, includes: Based on the evaluation results of the fitness function, a roulette wheel selection method is used to select superior individuals from the current population to enter the next generation; A single-point crossover operation is performed on the selected individuals, exchanging partial chromosome segments to generate a new access scheme; basic position mutation operations are then performed on the crossover individuals to randomly change individual gene positions in the chromosomes with a preset probability. Repeatedly perform selection, crossover, and mutation operations until the preset maximum number of iterations is reached or the evaluation result of the fitness function remains stable for multiple consecutive generations, and output the optimal access strategy for the differential protection terminal.

[0011] Preferably, after the differential protection terminal accesses the target power mobile edge computing node according to the access strategy, based on the target power mobile edge computing node with the optimization objective of maximizing the matching degree of network slice communication needs, the DDPG algorithm is used to dynamically allocate bandwidth and computing resources for the internal differential protection service slice, including: The internal resource status of the target power mobile edge computing node is obtained, including the current usage of total bandwidth resources, total computing power resources, and the quality of service requirements parameters of each network slice it supports; A state space for the DDPG algorithm is constructed, which includes the real-time bandwidth resource utilization and computing power resource utilization of the target power mobile edge computing node, as well as the rate and latency requirements of each network slice it carries; Define the action space of the DDPG algorithm, whereby the action space represents the amount of bandwidth and computing resources allocated to each network slice; A reward function for the DDPG algorithm is constructed, which is calculated based on the latency requirement matching degree of network slices. For differential protection service slices, the latency requirement matching degree is constructed to incentivize the algorithm to optimize its actual processing latency to a level significantly lower than its required latency. For non-differential protection service slices, the latency requirement matching degree is constructed to incentivize the algorithm to maintain its actual processing latency at a level close to its required latency. The Actor network generates resource allocation actions based on the current state, and the Critic network evaluates the long-term benefits of these resource allocation actions. Based on the feedback of the reward function, the parameters of the Actor network and Critic network are continuously optimized through experience replay and target network update mechanisms to obtain the optimal bandwidth and computing power resource allocation strategy. The bandwidth and computing power resource allocation strategy allocates dedicated resources to the differential protection service slice to meet the ultra-low latency requirements.

[0012] Preferably, after the resource allocation for the differential protection service slice is completed at the target power mobile edge computing node, based on the SDN controller with the optimization objectives of minimizing path delay and balancing the entire network load, an improved ant colony algorithm incorporating a bandwidth balancing factor is used to plan the optimal forwarding route from the target power mobile edge computing node to the power cloud platform for the differential protection service slice, including: The SDN controller acquires the overall network topology, real-time status information of each link, and bandwidth requirement parameters of differential protection service slices after resource allocation. An improved ant colony algorithm is initialized by deploying multiple ant agents along a feasible path between the target power mobile edge computing node and the power cloud platform. Each ant agent selects the next node hop by hop based on the path selection probability, which simultaneously considers the pheromone concentration of the link and the bandwidth balancing factor. The pheromone concentration is negatively correlated with the link delay, while the bandwidth balancing factor is positively correlated with the remaining bandwidth of the link. After each iteration, the pheromone concentration on each link is updated based on the latency performance of the paths explored by the ant agent. Paths with lower latency receive a greater pheromone enhancement. At the same time, the bandwidth equalization factor is used as an adjustment parameter for the pheromone evaporation coefficient. When the preset number of iterations is reached or the path quality tends to stabilize, the optimal forwarding route with the lowest latency and balanced load is obtained. The optimal forwarding route meets the end-to-end transmission requirements of the differential protection service slice.

[0013] Preferably, before each ant agent selects the next node hop-by-hop according to the path selection probability, wherein the path selection probability simultaneously considers the pheromone concentration of the link and the bandwidth balancing factor, wherein the pheromone concentration is negatively correlated with the link delay and the bandwidth balancing factor is positively correlated with the remaining bandwidth of the link, the process further includes: The pheromone concentration of a link is negatively correlated with its latency performance; the lower the latency, the higher the accumulated pheromone concentration. The bandwidth equalization factor of a link is positively correlated with the remaining bandwidth resources of the link; the more abundant the remaining bandwidth, the higher the bandwidth equalization factor value. Under the premise of meeting the end-to-end latency requirements of differential protection service slices, links with high pheromone concentration and large bandwidth equalization factor values ​​should be selected first; When the ant agent selects the next node, a roulette wheel selection mechanism is used, with the path selection probability as the selection basis. This ensures both the trend of convergence to the optimal path and the necessary exploration capability to discover potential optimal paths.

[0014] Secondly, this application discloses a security coordination device for differential protection services in distribution networks oriented towards 5G network slicing, which adopts the following technical solution, including: The communication architecture module is used to push the user plane functions of the 5G core network down to the power mobile edge computing node, forming a communication architecture consisting of power terminals, power mobile edge computing nodes, and a power cloud platform; when a differential protection service is initiated, the SDN controller receives a differential protection service slice request containing quality of service requirements; The edge node module is used to optimize the access strategy of power mobile edge computing nodes based on the SDN controller, with the goal of minimizing the variance of bandwidth resource utilization of all power mobile edge computing nodes in the region. It uses a genetic algorithm to calculate the optimal access strategy for the differential protection terminal that initiates the request. The fitness function of the genetic algorithm is based on node bandwidth constraints and signal-to-noise ratio constraints. The resource allocation module is used to dynamically allocate bandwidth and computing resources to the internal differential protection service slices based on the target power mobile edge computing node after the differential protection terminal accesses the target power mobile edge computing node according to the access strategy, with the optimization objective of maximizing the matching degree of network slice communication requirements. The routing planning module is used to plan the optimal forwarding route from the target power mobile edge computing node to the power cloud platform for the differential protection service slice after the resource allocation for the differential protection service slice is completed at the target power mobile edge computing node. Based on the SDN controller, the module aims to minimize path delay and balance the load of the entire network. It adopts an improved ant colony algorithm with the introduction of a bandwidth balancing factor to optimize the path delay and optimize the load balancing of the entire network. The pheromone parameter of the improved ant colony algorithm is designed according to the link delay, and its path selection probability is guided by both the pheromone concentration and the bandwidth balancing factor.

[0015] Thirdly, this application also provides a control device, the device comprising: It includes a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and executed, such as the above-described method for security collaboration of distribution network differential protection services for 5G network slicing.

[0016] Fourthly, this application also provides a computer-readable storage medium storing a computer program that can be loaded by a processor and executed as described above regarding the security coordination method for distribution network differential protection services oriented towards 5G network slicing.

[0017] In summary, this application constructs a cloud-edge-device communication architecture. By sinking the user plane functions of the 5G core network to the power mobile edge computing nodes controlled by the power grid, and independently deploying network units for power services, it achieves secure isolation from public network services, laying a dedicated physical foundation for differential protection. When a differential protection service initiates a request, end-to-end collaborative protection is achieved through three closely linked intelligent algorithm stages: First, the SDN controller uses a genetic algorithm to minimize the load imbalance of all MEC nodes in the region as the optimization objective. Taking into account multiple constraints such as signal-to-noise ratio, node capacity, and total resource blocks, it intelligently selects the optimal access node for each differential protection terminal, avoiding network congestion from the source. Next, after the terminal connects to the target MEC node, the node uses a deep reinforcement learning algorithm to dynamically allocate bandwidth and computing resources to the internal differential protection service slices, with the core objective of maximizing latency demand matching. This sets a highly challenging latency optimization target for the differential protection service, incentivizing the algorithm to drive its actual processing latency to a level far below the demand value. This provides unconventional resource guarantees and intelligently decides whether to adopt strategies such as data compression to trade computing power for bandwidth. Finally, based on the MEC resource allocation results, the SDN controller intervenes again, using an improved ant colony algorithm that introduces a bandwidth balancing factor to plan the optimal forwarding route from the edge to the cloud for the differential protection data. This algorithm ensures that the path selection probability is guided by both link latency and remaining bandwidth, guaranteeing that the final selected path is not only the one with the lowest latency but also the lightest load, reducing the creation of new network bottlenecks. This solves the core problems of static network slice resources being rigid and lacking end-to-end collaboration. Through an intelligent collaboration mechanism, it provides ultra-low latency and high-reliability deterministic communication guarantees for distribution network differential protection services, while maximizing network resource utilization efficiency and global load balancing, significantly improving the reliability and speed of differential protection. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating a security collaboration method for differential protection services in distribution networks oriented towards 5G network slicing.

[0019] Figure 2 This is a structural block diagram of a distribution network differential protection service security coordination device for 5G network slicing. Detailed Implementation

[0020] The following combination Figures 1-2 This application will be described in further detail.

[0021] Currently, in the practice of using 5G network slicing to carry distribution network differential protection services, three major technical bottlenecks are commonly encountered: static resource allocation is difficult to adapt to dynamic fluctuations in services; security mechanisms and low latency requirements are mutually constrained; and terminal access, edge processing, and network transmission are disconnected and lack coordination. This makes it difficult to reliably guarantee the end-to-end deterministic quality of service for differential protection services, limiting the deep application of 5G technology in critical power control services. Based on this, this application proposes a collaborative control scheme with deep integration of intelligent decision-making. Its core lies in constructing a novel, fully adaptive communication support system from the terminal to the cloud through a progressive three-level intelligent processing flow: terminal access balancing based on genetic algorithms, dynamic allocation of edge resources based on deep reinforcement learning, and cross-domain routing optimization based on improved ant colony algorithms.

[0022] Reference Figure 1 The embodiments of this application include at least steps S10 to S40.

[0023] S10 pushes the user plane functions of the 5G core network down to the power mobile edge computing node, forming a communication architecture consisting of power terminals, power mobile edge computing nodes and power cloud platform; when differential protection service is initiated, the SDN controller receives the differential protection service slice request containing quality of service requirements.

[0024] S20, based on the SDN controller, takes minimizing the variance of bandwidth resource utilization of all power mobile edge computing nodes in the region as the optimization objective, and uses a genetic algorithm to calculate the optimal power mobile edge computing node access strategy for the differential protection terminal that initiates the request. The fitness function of the genetic algorithm is based on node bandwidth constraints and signal-to-noise ratio constraints.

[0025] S30: After the differential protection terminal accesses the target power mobile edge computing node according to the access policy, based on the target power mobile edge computing node and with the optimization objective of maximizing the matching degree of network slice communication requirements, the DDPG algorithm is used to dynamically allocate bandwidth and computing resources for the internal differential protection service slice.

[0026] S40: After the target power mobile edge computing node completes the resource allocation for the differential protection service slice, based on the SDN controller with the optimization objectives of minimizing path delay and balancing the entire network load, an improved ant colony algorithm with an introduced bandwidth balancing factor is adopted to plan the optimal forwarding route from the target power mobile edge computing node to the power cloud platform for the differential protection service slice. The pheromone parameter of the improved ant colony algorithm is designed according to the link delay, and its path selection probability is guided by both the pheromone concentration and the bandwidth balancing factor.

[0027] Among them, control units and user plane units are deployed independently for power grid services to achieve secure isolation from the public network; the state space of the DDPG algorithm includes the real-time resource utilization rate of the target power mobile edge computing node and the bandwidth and latency requirements of the differential protection service slice, its action space is the amount of bandwidth and computing resources allocated for the differential protection service slice, and its reward function is constructed based on the latency requirement matching degree defined for the differential protection service slice.

[0028] Specifically, a cloud-edge-device security isolation architecture is constructed by introducing 5G user plane functions to provide a dedicated communication environment for differential protection. Based on this, end-to-end collaborative protection is achieved through a three-level intelligent algorithm: First, the SDN controller uses a genetic algorithm to optimize terminal access, selecting the best edge node for each differential protection terminal with load balancing as the goal; then, the target edge node uses the DDPG algorithm for dynamic resource allocation, which, based on a state space containing node resource status and service requirements, is driven by a reward function centered on latency requirement matching, allocating dedicated bandwidth and computing resources to the differential protection slice to ensure ultra-low latency; finally, the SDN controller uses an improved ant colony algorithm to plan cross-domain routes, selecting a fast and smooth end-to-end transmission path by simultaneously considering latency-related pheromone and load-related bandwidth balancing factors.

[0029] In some embodiments, step S10 specifically includes the following steps: deploying the user plane function network elements of the 5G core network in the power mobile edge computing node controlled by the power grid company, so that the power mobile edge computing node has local data forwarding and processing capabilities; in the communication architecture, deploying the control unit and user plane unit of the wireless access network independently for power services to achieve physical or logical isolation from public user services; establishing a connection between the power terminal and the power mobile edge computing node through the 5G wireless air interface; connecting the power mobile edge computing node to the power cloud platform through the power communication private network or operator network; when the differential protection service is initiated, the SDN controller deployed in the communication architecture receives the differential protection service slice request containing end-to-end latency requirements, bandwidth requirements and reliability indicators.

[0030] Specifically, by deploying 5G core network user plane functional elements to power mobile edge computing nodes under dedicated power grid control, a three-tiered collaborative communication architecture of power terminals, edge nodes, and cloud platforms is constructed. The core features of this architecture include: independently deploying radio access network control plane and user plane units for power services, achieving secure isolation from public services; terminals accessing edge nodes via 5G air interfaces, and edge nodes connecting to the cloud platform via dedicated power networks or operator networks. When differential protection services are activated, the SDN controller deployed in the architecture receives slice requests containing end-to-end latency, bandwidth, and reliability metrics. This architecture provides a collaborative computing foundation with secure isolation capabilities and layered processing characteristics for subsequent intelligent algorithms.

[0031] In some embodiments, step S20 specifically includes the following steps: obtaining the real-time bandwidth resource status of each power mobile edge computing node in the area, the channel quality information between each differential protection terminal and each power mobile edge computing node, and the total amount of available 5G resource blocks in the area based on the SDN controller; performing chromosome encoding on the access relationship between the differential protection terminal and the power mobile edge computing node to form an initial population; evaluating each individual in the population based on a fitness function, wherein the fitness function is configured as follows: based on evaluating the variance of bandwidth resource utilization, a first penalty term is applied to the terminal access scheme that does not meet the signal-to-noise ratio threshold, and a second penalty term is applied to the access scheme that causes the total bandwidth of the power mobile edge computing node to exceed the limit; the first penalty term: when the differential protection terminal and the power mobile edge computing node... When the signal-to-noise ratio (SNR) between target power mobile edge computing nodes is lower than a preset SNR threshold, a preset first penalty value is added to the fitness function. The magnitude of the first penalty value is positively correlated with the degree to which the SNR is lower than the threshold. A second penalty term is added to the fitness function when the access of a differential protection terminal causes the total bandwidth demand of the power mobile edge computing node to exceed the node's maximum bandwidth capacity. The magnitude of the second penalty value is positively correlated with the degree to which the bandwidth exceeds the limit. The population is iteratively evolved through selection, crossover, and mutation operations to gradually optimize the access strategy. When a preset number of iterations is reached or the fitness function converges, the optimal differential protection terminal access strategy is output. The access strategy specifies the best power mobile edge computing node that each differential protection terminal should access.

[0032] The specific expression for the fitness function r is: ; Where H is the set of mobile edge computing nodes in the region, and N is the set of differential protection terminals. This is a binary variable representing the connection status between terminal n and node h. Let n be the transmission rate of terminal n. The total bandwidth that node h can provide. This represents the total bandwidth requirement that node h needs to carry under the current access scheme, calculated using the following formula: μ is the average resource utilization rate of all nodes. The signal-to-noise ratio between terminal n and node h. This is the signal-to-noise ratio threshold.

[0033] Specifically, the SDN controller leads the implementation of terminal access optimization. By acquiring real-time data such as the bandwidth status of each edge node in the region, terminal channel quality, and the total amount of 5G resource blocks, the access relationship between terminals and nodes is encoded using chromosomes to form an initial population. Then, a fitness function incorporating multiple constraints is used for evaluation: based on calculating the variance of node bandwidth utilization, a triple penalty mechanism positively correlated with the degree of violation is set—increasing penalties are applied to access schemes with substandard signal-to-noise ratio, node bandwidth overload, and excessive regional resource limits. Based on this evaluation system, iterative evolution is performed through selection, crossover, and mutation operations using a genetic algorithm, ultimately outputting a load balancing strategy specifying the optimal access node for each differentially protected terminal, establishing an accurate terminal-node mapping relationship for subsequent edge resource allocation.

[0034] Furthermore, the population is iteratively evolved through selection, crossover, and mutation operations to gradually optimize the access strategy. When the preset number of iterations or the fitness function converges, the optimal differential protection terminal access strategy is output. Specifically, this includes: selecting high-quality individuals from the current population to enter the next generation using roulette wheel selection based on the fitness function evaluation results; performing a single-point crossover operation on the selected individuals to exchange some chromosome segments to generate a new access scheme; performing a basic position mutation operation on the crossover individuals to randomly change individual gene positions in the chromosome with a preset probability; and repeating the selection, crossover, and mutation operations until the preset maximum number of iterations is reached or the fitness function evaluation results remain stable for multiple generations, outputting the optimal differential protection terminal access strategy.

[0035] Specifically, based on the fitness assessment results, a roulette wheel selection method is used to select high-quality individuals to enter the next generation of evolution; a new access scheme is generated by exchanging chromosome segments through single-point crossover operation, and then gene positions are randomly adjusted with a preset probability through basic position mutation operation to introduce diversity; the selection, crossover and mutation operations are continuously repeated until the maximum number of iterations or the fitness function is stable for multiple generations convergence condition is met, and finally the optimal access strategy is output. This strategy realizes the self-evolution and optimization of the terminal access scheme by simulating the biological evolution process.

[0036] In some embodiments, step S30 specifically includes the following steps: obtaining the internal resource status of the target power mobile edge computing node, including the current usage of total bandwidth resources, total computing power resources, and the service quality requirement parameters of each network slice it carries; constructing the state space of the DDPG algorithm, the state space including the real-time bandwidth resource utilization rate, computing power resource utilization rate of the target power mobile edge computing node, and the rate requirement and latency requirement of each network slice it carries; defining the action space of the DDPG algorithm, the action space being represented by the amount of bandwidth resources and computing power resources allocated to each network slice; constructing the reward function of the DDPG algorithm, the reward function being calculated based on the latency requirement matching degree of the network slice, wherein for differential protection... For service slices, the latency requirement matching degree is constructed to incentivize the algorithm to optimize the actual processing latency to a level significantly lower than the required latency. For non-differential protection service slices, the latency requirement matching degree is constructed to incentivize the algorithm to maintain the actual processing latency at a level close to the required latency. The Actor network generates resource allocation actions based on the current state, and the Critic network evaluates the long-term benefits of the resource allocation actions. Based on the feedback of the reward function, the parameters of the Actor network and the Critic network are continuously optimized through experience replay and target network update mechanisms to obtain the optimal bandwidth and computing power resource allocation strategy. The bandwidth and computing power resource allocation strategy allocates dedicated resources to the differential protection service slices to meet the ultra-low latency requirements.

[0037] The specific expression of the reward function is as follows: ; in, and These are sets of delay-sensitive slices and delay-insensitive slices, respectively. The preset weighting coefficients, For network slice m, the latency requirement matching degree is [value missing]. For differential protection service slices, its [value missing] =0.5 indicates that for a specific type of business slice, the ratio is... The optimal target value, and These are penalties for failing to meet slicing requirements and for failing to meet the total equipment resource limits.

[0038] Specifically, after the terminal completes node access, edge-side resource optimization is initiated. A state space containing node resource utilization and service quality requirements for each slice is constructed, defining an action space to output specific resource allocation schemes. A reward function centered on latency requirement matching is designed: a reinforcement mechanism that incentivizes ultra-low latency is adopted for differential protection services, while a conservative strategy that maintains baseline latency is adopted for other services. Finally, an Actor-Critic network architecture is used to achieve interactive optimization of resource allocation and value assessment. Through experience replay and target network update mechanisms, continuous iterative training is performed, ultimately generating a dynamic allocation strategy that provides dedicated ultra-low latency resource guarantees for differential protection services.

[0039] In some embodiments, step S40 specifically includes the following steps: obtaining the entire network topology, real-time status information of each link, and bandwidth requirement parameters of the differential protection service slice after resource allocation based on the SDN controller; initializing the improved ant colony algorithm and deploying multiple ant agents on the feasible path between the target power mobile edge computing node and the power cloud platform; each ant agent selects the next node hop by hop according to the path selection probability, which considers both the pheromone concentration of the link and the bandwidth balancing factor; after each iteration, updating the pheromone concentration on each link according to the latency performance of the path explored by the ant agents, the path with lower latency receives a greater pheromone enhancement, and the bandwidth balancing factor is used as an adjustment parameter for the pheromone evaporation coefficient; when the preset number of iterations is reached or the path quality tends to stabilize, the optimal forwarding route with the lowest latency and load balancing is obtained, and the optimal forwarding route meets the end-to-end transmission requirements of the differential protection service slice.

[0040] Among them, the pheromone concentration of the link is negatively correlated with the link's latency performance; the lower the latency, the higher the accumulated pheromone concentration. The bandwidth balancing factor of the link is positively correlated with the remaining bandwidth resources of the link; the more abundant the remaining bandwidth, the larger the bandwidth balancing factor value. Under the premise of meeting the end-to-end latency requirements of differential protection service slices, links with high pheromone concentration and large bandwidth balancing factor values ​​are given priority. When the ant agent selects the next node, a roulette wheel selection mechanism is adopted, using the path selection probability as the selection basis, which ensures both the trend of convergence to the optimal path and maintains the necessary exploration capability to discover potential optimized paths.

[0041] Among them, path selection probability The calculation formula is: ; For time t link The concentration of pheromones on the surface For link The bandwidth equalization factor, where α and β are adjustment parameters. This is a list of links that ant k has already traversed.

[0042] The pheromone update rule is: ; The standardized bandwidth equalization factor, In this iteration, in the link The value of the newly added pheromone is inversely proportional to the total delay of the path traversed by the ant.

[0043] Specifically, after edge resource allocation is completed, cross-domain route optimization is initiated. The SDN controller obtains the network topology, link status, and specific bandwidth requirements of differential protection services. An improved ant colony algorithm is used to deploy multiple ant agents between edge nodes and the cloud platform for path exploration. Each agent selects a path based on a composite probability of fused pheromone concentration and bandwidth balancing factor, balancing convergence and exploration through a roulette wheel mechanism. During the iteration process, pheromone levels are updated based on path latency performance (significant enhancement for low-latency paths), and the bandwidth balancing factor is used as an adjustment parameter to control pheromone evaporation, ultimately converging to the optimal route that simultaneously satisfies ultra-low latency and load balancing.

[0044] The implementation principle of the security collaboration method for differential protection services in distribution networks oriented to 5G network slicing in this application embodiment is as follows: A cloud-edge-device communication architecture is constructed. By sinking the user plane functions of the 5G core network to the power mobile edge computing nodes controlled by the power grid, and independently deploying network units for power services, secure isolation from public network services is achieved, laying a dedicated physical foundation for differential protection. When the differential protection service initiates a request, end-to-end collaborative protection is achieved through three closely linked intelligent algorithm stages: First, the SDN controller adopts a genetic algorithm with the optimization objective of minimizing the load imbalance of all MEC nodes in the region. Taking into account multiple constraints such as signal-to-noise ratio, node capacity, and total resource blocks, it intelligently selects the optimal access node for each differential protection terminal, avoiding network congestion from the source. Next, after the terminal connects to the target MEC node, the node uses a deep reinforcement learning algorithm to dynamically allocate bandwidth and computing resources to the internal differential protection service slices, with the core objective of maximizing latency demand matching. This sets a highly challenging latency optimization target for the differential protection service, incentivizing the algorithm to drive its actual processing latency to a level far below the demand value. This provides unconventional resource guarantees and intelligently decides whether to adopt strategies such as data compression to trade computing power for bandwidth. Finally, based on the MEC resource allocation results, the SDN controller intervenes again, using an improved ant colony algorithm that introduces a bandwidth balancing factor to plan the optimal forwarding route from the edge to the cloud for the differential protection data. This algorithm ensures that the path selection probability is guided by both link latency and remaining bandwidth, guaranteeing that the final selected path is not only the one with the lowest latency but also the lightest load, reducing the creation of new network bottlenecks. This solves the core problems of static network slice resources being rigid and lacking end-to-end collaboration. Through an intelligent collaboration mechanism, it provides ultra-low latency and high-reliability deterministic communication guarantees for distribution network differential protection services, while maximizing network resource utilization efficiency and global load balancing, significantly improving the reliability and speed of differential protection.

[0045] Figure 1 This is a flowchart illustrating a security coordination method for distribution network differential protection services oriented towards 5G network slicing in one embodiment. It should be understood that, although... Figure 1 The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows; unless explicitly stated herein, there is no strict order requirement for the execution of these steps, and they can be executed in other orders; and Figure 1At least some of the steps in the process may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least some of the sub-steps or stages of other steps.

[0046] Based on the same technical concept, referring to Figure 2 This application also provides a security coordination device for distribution network differential protection services oriented to 5G network slicing, which adopts the following technical solution: The device includes: The communication architecture module is used to push the user plane functions of the 5G core network down to the power mobile edge computing node, forming a communication architecture consisting of power terminals, power mobile edge computing nodes, and a power cloud platform; when a differential protection service is initiated, the SDN controller receives a differential protection service slice request containing quality of service requirements; The edge node module is used to optimize the access strategy of power mobile edge computing nodes based on the SDN controller, with the goal of minimizing the variance of bandwidth resource utilization of all power mobile edge computing nodes in the region. It employs a genetic algorithm to calculate the optimal access strategy for the differential protection terminal that initiates the request. The fitness function of the genetic algorithm is based on node bandwidth constraints and signal-to-noise ratio constraints. The resource allocation module is used to dynamically allocate bandwidth and computing resources to the internal differential protection service slices after the differential protection terminal accesses the target power mobile edge computing node according to the access policy. The optimization objective is to maximize the matching degree of network slice communication requirements. The routing planning module is used to plan the optimal forwarding route from the target power mobile edge computing node to the power cloud platform for the differential protection service slice after the resource allocation of the differential protection service slice is completed at the target power mobile edge computing node. Based on the SDN controller, the module aims to minimize path delay and balance the load of the entire network. It adopts an improved ant colony algorithm with the introduction of a bandwidth balancing factor to optimize the path delay and optimize the load balancing of the entire network. The pheromone parameter of the improved ant colony algorithm is designed according to the link delay, and its path selection probability is guided by both the pheromone concentration and the bandwidth balancing factor.

[0047] In some embodiments, the communication architecture module is specifically used to deploy the user plane function network elements of the 5G core network in the power mobile edge computing node controlled by the power grid company, so that the power mobile edge computing node has local data forwarding and processing capabilities; In the communication architecture, the control unit and user plane unit of the wireless access network are deployed independently for power services to achieve physical or logical isolation from public user services; The power terminal establishes a connection with the power mobile edge computing node via a 5G wireless air interface; the power mobile edge computing node connects to the power cloud platform via a dedicated power communication network or operator network. When a differential protection service is initiated, the SDN controller deployed in the communication architecture receives a differential protection service slice request that includes end-to-end latency requirements, bandwidth requirements, and reliability metrics.

[0048] In some embodiments, the edge node module is specifically used to obtain, based on the SDN controller, the real-time bandwidth resource status of each power mobile edge computing node in the area, the channel quality information between each differential protection terminal and each power mobile edge computing node, and the total amount of 5G resource blocks available in the area; The access relationship between the differential protection terminal and the power mobile edge computing node is encoded using chromosomes to form an initial population; The fitness function is used to evaluate each individual in the population. The fitness function is configured as follows: based on the evaluation of the variance of bandwidth resource utilization, a first penalty term is applied to the terminal access scheme that does not meet the signal-to-noise ratio threshold, and a second penalty term is applied to the access scheme that causes the total bandwidth of the power mobile edge computing node to exceed the limit. First penalty: When the signal-to-noise ratio (SNR) between the differential protection terminal and the target power mobile edge computing node is lower than a preset SNR threshold, a preset first penalty value is added to the fitness function. The magnitude of the first penalty value is positively correlated with the degree to which the SNR is lower than the threshold. Second penalty: When the access of the differential protection terminal causes the total bandwidth demand of the power mobile edge computing node to exceed the maximum bandwidth capacity of the node, a preset second penalty value is added to the fitness function. The magnitude of the second penalty value is positively correlated with the degree to which the bandwidth exceeds the limit. The population is iteratively evolved through selection, crossover, and mutation operations to gradually optimize the access strategy. When the preset number of iterations or the fitness function converges, the optimal differential protection terminal access strategy is output. The access strategy specifies the best power mobile edge computing node that each differential protection terminal should connect to.

[0049] In some embodiments, the edge node module is also used to select high-quality individuals from the current population to enter the next generation based on the evaluation results of the fitness function using a roulette wheel selection method; A single-point crossover operation is performed on the selected individuals, exchanging partial chromosome segments to generate a new access scheme; basic position mutation operations are then performed on the crossover individuals to randomly change individual gene positions in the chromosomes with a preset probability. Repeatedly perform selection, crossover, and mutation operations until the preset maximum number of iterations is reached or the evaluation result of the fitness function remains stable for multiple generations, and output the optimal access strategy for the differential protection terminal.

[0050] In some embodiments, the resource allocation module is specifically used to obtain the internal resource status of the target power mobile edge computing node, including the current usage of total bandwidth resources, total computing power resources, and the quality of service requirement parameters of each network slice already supported; The state space of the DDPG algorithm is constructed, which includes the real-time bandwidth resource utilization and computing power resource utilization of the target power mobile edge computing node, as well as the rate and latency requirements of each network slice it carries; Define the action space of the DDPG algorithm, where the action space is represented by the amount of bandwidth and computing resources allocated to each network slice; The reward function of the DDPG algorithm is constructed. The reward function is calculated based on the latency requirement matching degree of the network slice. For differential protection service slices, the latency requirement matching degree is constructed to incentivize the algorithm to optimize the actual processing latency to a level significantly lower than the required latency. For non-differential protection service slices, the latency requirement matching degree is constructed to incentivize the algorithm to maintain the actual processing latency at a level close to the required latency. The Actor network generates resource allocation actions based on the current state, and the Critic network evaluates the long-term benefits of these resource allocation actions. Based on the feedback of the reward function, the parameters of the Actor network and Critic network are continuously optimized through experience replay and target network update mechanism to obtain the optimal bandwidth and computing resource allocation strategy. The bandwidth and computing resource allocation strategy allocates dedicated resources to the differential protection service slice to meet the ultra-low latency requirements.

[0051] In some embodiments, the routing planning module is specifically used to obtain the overall network topology, real-time status information of each link, and bandwidth requirement parameters of the differential protection service slice after resource allocation based on the SDN controller; An improved ant colony algorithm is initialized by deploying multiple ant agents along a feasible path between the target power mobile edge computing node and the power cloud platform. Each ant agent selects the next node hop by hop based on the path selection probability. The path selection probability takes into account both the pheromone concentration of the link and the bandwidth balancing factor. The pheromone concentration is negatively correlated with the link latency, while the bandwidth balancing factor is positively correlated with the remaining bandwidth of the link. After each iteration, the pheromone concentration on each link is updated based on the latency performance of the paths explored by the ant agent. Paths with lower latency receive a greater pheromone enhancement. At the same time, the bandwidth equalization factor is used as an adjustment parameter for the pheromone evaporation coefficient. When the preset number of iterations is reached or the path quality tends to stabilize, the optimal forwarding route with the lowest latency and balanced load is obtained. The optimal forwarding route meets the end-to-end transmission requirements of the differential protection service slice.

[0052] In some embodiments, the routing planning module is also used to determine the pheromone concentration of a link, the value of which is negatively correlated with the link's latency performance; links with lower latency accumulate higher pheromone concentrations. The bandwidth equalization factor of a link is positively correlated with the remaining bandwidth resources of the link; the more abundant the remaining bandwidth, the higher the bandwidth equalization factor value. Under the premise of meeting the end-to-end latency requirements of differential protection service slices, links with high pheromone concentration and large bandwidth equalization factor values ​​should be selected first; When the ant agent selects the next node, a roulette wheel selection mechanism is used, which uses the path selection probability as the selection basis. This ensures both the trend of convergence to the optimal path and the necessary exploration capability to discover potential optimal paths.

[0053] This application also discloses a control device.

[0054] Specifically, the control device includes a memory and a processor. The memory stores a computer program that can be loaded by the processor and executed to provide a secure collaborative method for distribution network differential protection services oriented towards 5G network slicing.

[0055] This application also discloses a computer-readable storage medium.

[0056] Specifically, the computer-readable storage medium stores a computer program that can be loaded by a processor and executed, such as the aforementioned security collaboration method for distribution network differential protection services oriented towards 5G network slicing. The computer-readable storage medium includes, for example, various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0057] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. A method for secure collaborative operation of differential protection services in distribution networks for 5G network slicing, characterized in that... ,include: The user plane functions of the 5G core network are pushed down to the power mobile edge computing nodes, forming a communication architecture consisting of power terminals, power mobile edge computing nodes, and a power cloud platform; when a differential protection service is initiated, the SDN controller receives a differential protection service slice request containing quality of service requirements; Based on the SDN controller, with the optimization objective of minimizing the variance of bandwidth resource utilization of all power mobile edge computing nodes in the region, a genetic algorithm is used to calculate the optimal access strategy for the differential protection terminal that initiates the request. The fitness function of the genetic algorithm is based on node bandwidth constraints and signal-to-noise ratio constraints. After the differential protection terminal accesses the target power mobile edge computing node according to the access strategy, based on the target power mobile edge computing node and with the optimization objective of maximizing the matching degree of network slice communication requirements, the DDPG algorithm is used to dynamically allocate bandwidth and computing resources to the internal differential protection service slices; After the target power mobile edge computing node completes the resource allocation for the differential protection service slice, based on the SDN controller with the optimization objectives of minimizing path latency and balancing the entire network load, an improved ant colony algorithm with an introduced bandwidth balancing factor is adopted to plan the optimal forwarding route from the target power mobile edge computing node to the power cloud platform for the differential protection service slice. The pheromone parameter of the improved ant colony algorithm is designed according to the link latency, and its path selection probability is guided by both the pheromone concentration and the bandwidth balancing factor.

2. The method for secure collaborative operation of differential protection services in distribution networks oriented towards 5G network slicing as described in claim 1, characterized in that... The process involves pushing the user plane functions of the 5G core network down to the power mobile edge computing node, forming a communication architecture consisting of power terminals, power mobile edge computing nodes, and a power cloud platform. When a differential protection service is initiated, the SDN controller receives a differential protection service slice request containing quality of service requirements, including: The user plane function network elements of the 5G core network are deployed in the power mobile edge computing nodes controlled by the power grid company, enabling the power mobile edge computing nodes to have local data forwarding and processing capabilities; In the aforementioned communication architecture, a control unit and user plane unit of the wireless access network are deployed independently for power services, achieving physical or logical isolation from public user services; The power terminal establishes a connection with the power mobile edge computing node via a 5G wireless air interface; the power mobile edge computing node connects to the power cloud platform via a dedicated power communication network or operator network. When a differential protection service is initiated, the SDN controller deployed in the communication architecture receives a differential protection service slice request that includes end-to-end latency requirements, bandwidth requirements, and reliability metrics.

3. A method for secure collaborative operation of differential protection services in distribution networks oriented towards 5G network slicing, as described in claim 1, is characterized in that... The SDN-based controller aims to minimize the variance of bandwidth resource utilization of all power mobile edge computing nodes within the region. It employs a genetic algorithm to calculate the optimal access strategy for the requesting differential protection terminal, including: Based on the SDN controller, the real-time bandwidth resource status of each power mobile edge computing node in the area, the channel quality information between each differential protection terminal and each power mobile edge computing node, and the total amount of available 5G resource blocks in the area are obtained; The access relationship between the differential protection terminal and the power mobile edge computing node is encoded using chromosomes to form an initial population; The fitness function is used to evaluate each individual in the population. The fitness function is configured as follows: based on the evaluation of the variance of bandwidth resource utilization, a first penalty term is applied to the terminal access scheme that does not meet the signal-to-noise ratio threshold, and a second penalty term is applied to the access scheme that causes the total bandwidth of the power mobile edge computing node to exceed the limit. The first penalty term: When the signal-to-noise ratio (SNR) between the differential protection terminal and the target power mobile edge computing node is lower than a preset SNR threshold, a preset first penalty value is added to the fitness function. The magnitude of the first penalty value is positively correlated with the degree to which the SNR is lower than the threshold. The second penalty term: When the access of the differential protection terminal causes the total bandwidth demand of the power mobile edge computing node to exceed the maximum bandwidth capacity of the node, a preset second penalty value is added to the fitness function. The magnitude of the second penalty value is positively correlated with the degree to which the bandwidth exceeds the limit. The population is iteratively evolved through selection, crossover, and mutation operations to gradually optimize the access strategy. When the preset number of iterations or the fitness function converges, the optimal differential protection terminal access strategy is output. The access strategy specifies the best power mobile edge computing node that each differential protection terminal should connect to.

4. A method for secure collaborative operation of differential protection services in distribution networks oriented towards 5G network slicing, as described in claim 3, is characterized in that... The process involves iteratively evolving the population through selection, crossover, and mutation operations to gradually optimize the access strategy. When a preset number of iterations or convergence of the fitness function is reached, the optimal differential protection terminal access strategy is output, including: Based on the evaluation results of the fitness function, a roulette wheel selection method is used to select superior individuals from the current population to enter the next generation; A single-point crossover operation is performed on the selected individuals, exchanging partial chromosome segments to generate a new access scheme; basic position mutation operations are then performed on the crossover individuals to randomly change individual gene positions in the chromosomes with a preset probability. Repeatedly perform selection, crossover, and mutation operations until the preset maximum number of iterations is reached or the evaluation result of the fitness function remains stable for multiple consecutive generations, and output the optimal access strategy for the differential protection terminal.

5. A method for secure collaborative operation of differential protection services in distribution networks oriented towards 5G network slicing, as described in claim 4, is characterized in that... After the differential protection terminal accesses the target power mobile edge computing node according to the access strategy, based on the target power mobile edge computing node and with the optimization objective of maximizing the matching degree of network slice communication requirements, the DDPG algorithm is used to dynamically allocate bandwidth and computing resources for the internal differential protection service slices, including: The internal resource status of the target power mobile edge computing node is obtained, including the current usage of total bandwidth resources, total computing power resources, and the quality of service requirements parameters of each network slice it supports; A state space for the DDPG algorithm is constructed, which includes the real-time bandwidth resource utilization and computing power resource utilization of the target power mobile edge computing node, as well as the rate and latency requirements of each network slice it carries; Define the action space of the DDPG algorithm, whereby the action space represents the amount of bandwidth and computing resources allocated to each network slice; A reward function for the DDPG algorithm is constructed. This function is calculated based on the latency requirement matching degree of network slices. The latency requirement matching degree is defined as a function of the ratio of actual processing latency to required latency. By adjusting the parameters of the function, for differential protection service slices, the algorithm is incentivized to optimize its actual processing latency to be lower than its required latency; for non-differential protection service slices, the algorithm is incentivized to maintain its actual processing latency at a level close to its required latency. The Actor network generates resource allocation actions based on the current state, and the Critic network evaluates the long-term benefits of these resource allocation actions. Based on the feedback of the reward function, the parameters of the Actor network and Critic network are continuously optimized through experience replay and target network update mechanisms to obtain the optimal bandwidth and computing power resource allocation strategy. The bandwidth and computing power resource allocation strategy allocates dedicated resources to the differential protection service slice to meet the ultra-low latency requirements.

6. A method for secure collaborative operation of differential protection services in distribution networks oriented towards 5G network slicing, as described in claim 5, is characterized in that... After the resource allocation for the differential protection service slice is completed at the target power mobile edge computing node, based on the SDN controller and with the optimization objectives of minimizing path latency and balancing the entire network load, an improved ant colony algorithm incorporating a bandwidth balancing factor is used to plan the optimal forwarding route from the target power mobile edge computing node to the power cloud platform for the differential protection service slice, including: The SDN controller acquires the overall network topology, real-time status information of each link, and bandwidth requirement parameters of differential protection service slices after resource allocation. An improved ant colony algorithm is initialized by deploying multiple ant agents along a feasible path between the target power mobile edge computing node and the power cloud platform. Each ant agent selects the next node hop by hop based on the path selection probability, which simultaneously considers the pheromone concentration of the link and the bandwidth balancing factor. The pheromone concentration is negatively correlated with the link delay, while the bandwidth balancing factor is positively correlated with the remaining bandwidth of the link. After each iteration, the pheromone concentration on each link is updated based on the latency performance of the paths explored by the ant agent. Paths with lower latency receive a greater pheromone enhancement. At the same time, the bandwidth equalization factor is used as an adjustment parameter for the pheromone evaporation coefficient. When the preset number of iterations is reached or the path quality tends to stabilize, the optimal forwarding route with the lowest latency and balanced load is obtained. The optimal forwarding route meets the end-to-end transmission requirements of the differential protection service slice.

7. A method for secure collaborative operation of differential protection services in distribution networks for 5G network slicing, as described in claim 6, is characterized in that... Before each ant agent selects the next node hop-by-hop according to the path selection probability, where the path selection probability simultaneously considers the pheromone concentration of the link and the bandwidth balancing factor, wherein the pheromone concentration is negatively correlated with the link delay and the bandwidth balancing factor is positively correlated with the remaining bandwidth of the link, the process further includes: The pheromone concentration of a link is negatively correlated with its latency performance; the lower the latency, the higher the accumulated pheromone concentration. The bandwidth equalization factor of a link is positively correlated with the remaining bandwidth resources of the link; the more abundant the remaining bandwidth, the higher the bandwidth equalization factor value. Under the premise of meeting the end-to-end latency requirements of differential protection service slices, links with high pheromone concentration and large bandwidth equalization factor values ​​should be selected first; When the ant agent selects the next node, a roulette wheel selection mechanism is used, with the path selection probability as the selection basis. This ensures both the trend of convergence to the optimal path and the necessary exploration capability to discover potential optimal paths.

8. A security coordination device for differential protection services in distribution networks oriented towards 5G network slicing, characterized in that... The device includes: The communication architecture module is used to push the user plane functions of the 5G core network down to the power mobile edge computing node, forming a communication architecture consisting of power terminals, power mobile edge computing nodes, and a power cloud platform; when a differential protection service is initiated, the SDN controller receives a differential protection service slice request containing quality of service requirements; The edge node module is used to optimize the access strategy of power mobile edge computing nodes based on the SDN controller, with the goal of minimizing the variance of bandwidth resource utilization of all power mobile edge computing nodes in the region. It uses a genetic algorithm to calculate the optimal access strategy for the differential protection terminal that initiates the request. The fitness function of the genetic algorithm is based on node bandwidth constraints and signal-to-noise ratio constraints. The resource allocation module is used to dynamically allocate bandwidth and computing resources to the internal differential protection service slices based on the target power mobile edge computing node after the differential protection terminal accesses the target power mobile edge computing node according to the access strategy, with the optimization objective of maximizing the matching degree of network slice communication requirements. The routing planning module is used to plan the optimal forwarding route from the target power mobile edge computing node to the power cloud platform for the differential protection service slice after the resource allocation for the differential protection service slice is completed at the target power mobile edge computing node. Based on the SDN controller, the module aims to minimize path delay and balance the load of the entire network. It adopts an improved ant colony algorithm with the introduction of a bandwidth balancing factor to optimize the path delay and optimize the load balancing of the entire network. The pheromone parameter of the improved ant colony algorithm is designed according to the link delay, and its path selection probability is guided by both the pheromone concentration and the bandwidth balancing factor.

9. A control device, characterized in that... The device includes: A memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and executed as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that... It stores a computer program that can be loaded by a processor and executed as described in any one of claims 1 to 7.

Images