Multi-service multi-modal power communication network planning method based on double-agent collaborative optimization

Abstract

The application discloses a multi-service multi-modal power communication network planning method based on double-agent collaborative optimization, and relates to the technical field of power communication network planning.The application unifies multiple service demands, multiple transmission systems and engineering constraints into the same planning model, takes the minimum comprehensive construction cost of the whole network as the target, constructs a feasible search space for each service by the first agent, and pre-generates a candidate path set meeting the service adaptation, hierarchical permission and quality of service requirements; the second agent takes the candidate path set as the input, performs global collaborative optimization, and outputs the final bearing path of each service, the bandwidth configuration amount of each transmission system on each link and the comprehensive construction cost of the whole network.The application realizes joint decision of the service path and the transmission system, solves the technical problem that the existing method cannot realize global collaborative planning under the condition of coexistence of multiple services and multiple systems, and the planning result has complete engineering implementability.

Description

Technical Field

[0001] This invention relates to the field of power communication network planning technology, specifically a multi-service, multi-modal power communication network planning method with dual-agent collaborative optimization. Background Technology

[0002] With the continuous advancement of new power system construction, power communication networks need to simultaneously support multiple services such as relay protection, distribution automation, electricity consumption information collection, and video surveillance, and be able to collaboratively utilize various heterogeneous transmission systems such as fiber optic communication, high-speed power line communication, and wireless private network communication. Different services exhibit significant differences in bandwidth, latency, jitter, and reliability, while different transmission systems vary in construction costs, bandwidth capacity, transmission performance, and applicable scenarios. This constitutes a typical multi-service, multi-modal, heterogeneous power communication network planning problem.

[0003] Existing power communication network planning methods typically separate service path selection and transmission system configuration into two independent decision-making stages when addressing the aforementioned problems, or focus solely on path optimization under a single communication system. Specifically, existing methods first determine the service transmission path and then match available transmission systems to that path, failing to achieve joint and collaborative decision-making between the service path and heterogeneous transmission systems. This fragmented planning approach leads to two key problems: firstly, service path selection fails to consider the performance differences and cost characteristics of different transmission systems, easily resulting in suboptimal solutions with short path distances but requiring high-cost systems; secondly, transmission system configuration is constrained by the determined paths, failing to leverage the complementary advantages of multiple heterogeneous systems on a global scale. Therefore, existing methods struggle to output a complete collaborative planning scheme that simultaneously includes service-bearing paths and the configuration relationships of transmission systems across links.

[0004] To address this, the present invention proposes a dual-agent collaborative optimization method for planning multi-service, multi-modal power communication networks. Summary of the Invention

[0005] The purpose of this invention is to provide a dual-agent collaborative optimization method for planning multi-service, multi-modal power communication networks, in order to solve the problems mentioned in the background art.

[0006] According to a first aspect of the present invention, in order to achieve the above-mentioned objective, the present invention provides the following technical solution: a multi-service, multi-modal power communication network planning method with dual-agent cooperative optimization, comprising the following steps: Obtain service requirement information for various services in the power communication network and performance attribute information for various transmission systems; Based on the service requirements information and the transmission system performance attribute information, a mathematical model of a multi-service multimodal power communication network is established. The mathematical model has an objective function that minimizes the overall construction cost of the power communication network and sets multiple types of constraints. A first intelligent agent is constructed. Based on the service requirement information and the transmission system performance attribute information, the first intelligent agent constructs a feasible search space for each service, eliminates infeasible paths, and generates a set of candidate paths for a single service within the feasible search space. Each candidate path includes a link sequence and a transmission system sequence. The candidate path sets of each generated single business are aggregated to form the full business candidate path set; A second intelligent agent is constructed. The second intelligent agent takes the set of candidate paths for all services as input and performs global iterative optimization with the objective function as the goal under the premise of satisfying multiple constraints. After reaching the preset termination condition, the final planning result is output. The final planning result includes the final bearer path of each service, the bandwidth configuration of each transmission system on each link, and the overall construction cost of the entire network.

[0007] Furthermore, the service requirement information includes the source node, destination node, service type, bandwidth requirement, and quality of service requirement for each service, wherein the quality of service requirement includes latency requirement and jitter requirement. The system attribute information includes the performance characteristics, applicable scope, construction attributes, and compatibility relationships between different service types and different transmission systems for each transmission system.

[0008] Furthermore, the mathematical model of the multi-service, multi-modal power communication network is specifically represented as follows: In the formula, Representative from the system Switch to The one-time switchover cost includes equipment, wiring, and commissioning fees. It is a 0-1 variable, representing business logic. At the node Whether the system occurs Switching, Representing a path The above adopts a system The amount of bandwidth that needs to be configured at that time. This represents the cost reduction that can be achieved through resource reuse.

[0009] Furthermore, the objective function is specifically expressed as: In the formula, Represents a link, Indicates the transmission system, Represents the business it carries. Represents a set of network nodes. Represents the link The above is the system The unit cost of building bandwidth. Indicates link The above adopts a system The amount of bandwidth that needs to be configured at that time. It is a 0-1 variable, representing whether or not it is in Enable resource reuse, that is: In the formula, This means that resource reuse can reduce costs. Representative from the system Switch to The one-time switchover cost includes equipment, wiring, and commissioning fees. It is a 0-1 variable, representing business logic. At the node Whether the system occurs The switching, that is: .

[0010] Furthermore, the various constraints specifically include service connectivity constraints, consistency constraints between bearer and construction, service type and transmission system permission constraints, link layering and transmission system permission constraints, quality of service constraints, bandwidth capacity constraints, and backbone network node ring formation constraints.

[0011] Furthermore, the service connectivity constraint is expressed by the flow conservation formula, as follows: in, It is a 0-1 variable, representing the business logic. Does the main path use : Representative business The source node, Representative business The host node, Representing all from The starting link, Represents all pointers The link; The consistency constraint between load and construction is specifically expressed as follows: in, It is a 0-1 variable, indicating whether it is on the link. Construction system : ; The specific restrictions on service types and transmission system permissions are as follows: in, It is a 0-1 variable representing the allowable matrix, i.e., according to "business category". system The given allow / disallow parameters are expressed as: ; The link layering and transmission system licensing constraints are specifically represented as follows: in, It is a 0-1 variable representing the physical link. Does the network layer allow for the construction of a system? ,Right now: ; The quality of service constraints include latency requirement constraints and jitter constraints, specifically as follows: in, It is a 0-1 variable representing whether the business has a low latency requirement: It is a 0-1 variable, representing the system. Can it be considered a low-latency system? Similarly, the jitter constraint is expressed as: in, It is a 0-1 variable representing whether the business requires low jitter: It is also a 0-1 variable, representing the system. Could this be considered a low-jitter system? ; The bandwidth capacity constraint is specifically expressed as follows: In the formula Representative business bandwidth requirements, Representing a path The above adopts a system The amount of bandwidth that needs to be configured at that time. It is a parameter with a value between 0 and 1, representing the upper limit of link utilization; The ring constraint of the backbone network nodes is specifically represented as follows: in, Represents all connected nodes The set of candidate links, It is a 0-1 variable, representing the link. Whether it is adopted or constructed, i.e.: Representative node Existing links, It is also a 0-1 variable, representing whether this node forms a cycle, that is: when hour, Take 1; if 0 or 1, Set to 0; Represents the set of key nodes in the backbone network. This represents the number of key nodes in the backbone network.

[0012] Furthermore, the first intelligent agent is a single-business feasible path generation intelligent agent. The first intelligent agent adopts a path generation architecture based on the K shortest path algorithm, specifically including a feasible edge screening layer, a business-specific weighted graph construction layer, a K shortest path search layer, and a candidate path output layer. The first intelligent agent takes as input the business requirement information, system attribute information, and various constraints, and outputs a set of candidate paths for each business. The specific processing flow is as follows: For each business k, a feasible search space is constructed, infeasible paths are eliminated, and a set of candidate paths is generated within the feasible search space. Specifically, it is expressed as: in, Indicates business The Alternative routes; A feasible search space is constructed for all business operations, and the candidate path sets for individual businesses are aggregated to form a complete set of candidate paths for all businesses: In the formula, This represents the total number of business transactions.

[0013] Furthermore, when constructing the feasible search space, the first intelligent agent sequentially determines whether each candidate service link and transmission system combination satisfies the following: The transmission system allows the carrying of the service type, the network layer where the candidate link is located allows the deployment of the transmission system, and the transmission system meets the quality of service requirements of the service. Candidate links and transmission scheme combinations are retained as searchable edges only when all conditions are met simultaneously; Within the feasible search space, the first intelligent agent generates several candidate paths from the source node to the destination node for each service based on the flow conservation formula. Each candidate path contains a complete node link sequence and a corresponding transmission system sequence.

[0014] Furthermore, the second intelligent agent takes the full set of candidate paths for all services as input, and adopts a global collaborative optimization architecture based on genetic algorithms, including a path index encoding layer, a population initialization layer, a chromosome decoding layer, a network-wide resource statistics layer, a fitness evaluation layer, a genetic iterative update layer, and an optimal solution output layer. The second intelligent agent takes the entire set of candidate paths for all services as input, and performs global iterative optimization with the objective function as the goal, under the premise of satisfying multiple constraints. After reaching the preset termination condition, it outputs the final planning result. The specific steps are as follows: The full-service candidate path set As input, where Total number of business transactions; Candidate planning schemes are generated using a path index encoding method, and the candidate planning schemes are represented as follows: in For the purpose of indicating business From its candidate path set Select the first The index of the path; Decode the candidate planning scheme This allows us to obtain the final bearer path and corresponding transmission system for each service. In response to the decoding results, the service load is statistically analyzed according to two dimensions: link and transmission system. The bandwidth required for each system on each link is calculated, and the comprehensive construction cost of the candidate planning scheme is calculated according to the objective function. The bandwidth is expressed as: Check whether the candidate planning scheme satisfies the bandwidth constraint and the ring formation constraint of the backbone network nodes in the constraint set. Based on the comprehensive construction cost and constraint check results, a fitness function is constructed, where solutions that satisfy the constraints and have lower costs have high fitness, while solutions that do not satisfy the constraints have their fitness reduced through a penalty term, as follows: For any candidate programming scheme First, based on each gene locus Decoding to obtain business The selected candidate path is used to determine the service carrying variables. Link construction variables Institutional Development Variables Resource reuse variables System transformation variables and links upper system bandwidth configuration ; Calculate the overall construction cost of this candidate planning scheme based on the objective function: And construct constraint penalty terms based on various constraints: in, This is a penalty item for business connectivity constraints. To accommodate the constraints and penalties related to consistency between the carrying capacity and construction, For business-related institutional restrictions and penalties, This is a penalty term for bandwidth capacity constraints. The constraint penalty term for the ring formation of backbone network nodes. For the corresponding penalty weight; Each penalty item is calculated as follows: in, in, Based on the comprehensive construction cost and constraint penalty terms, the second agent constructs a fitness function: in, Indicates candidate planning schemes fitness For each candidate planning scheme in the population, selection, crossover, and mutation are performed iteratively based on fitness until a preset termination condition is met. The candidate planning scheme with the highest fitness is then output as the final planning result.

[0015] According to a second aspect of the present invention, the present invention provides a dual-agent cooperative optimization multi-service multimodal power communication network planning system for implementing the dual-agent cooperative optimization multi-service multimodal power communication network planning method described in the first aspect, comprising: The information acquisition module is used to acquire service requirement information for various services in the power communication network and transmission system performance attribute information for various transmission systems. The model building module is used to establish a mathematical model of a multi-service multimodal power communication network based on the service requirement information and the transmission system performance attribute information. The mathematical model has an objective function that minimizes the overall construction cost of the power communication network and sets multiple types of constraints. The first intelligent agent construction module is used to construct the first intelligent agent. The first intelligent agent constructs a feasible search space for each service based on the service requirement information and the transmission system performance attribute information, filters out infeasible paths, and generates a candidate path set for a single service within the feasible search space. Each candidate path includes a link sequence and a transmission system sequence. The dataset aggregation module is used to aggregate the candidate path sets of each generated single business to form a full business candidate path set; The second intelligent agent construction module is used to construct a second intelligent agent. The second intelligent agent takes the set of candidate paths for all services as input, and performs global iterative optimization with the objective function as the goal under the premise of satisfying multiple constraints. After reaching the preset termination condition, it outputs the final planning result. The final planning result includes the final bearer path of each service, the bandwidth configuration of each transmission system on each link, and the overall construction cost of the entire network.

[0016] This invention has at least the following beneficial effects: 1. This invention incorporates path selection and system selection into the same optimization process. The output planning results not only include the carrying path of each service, but also the transmission system adopted by each link segment on the path and the bandwidth scale required for each system. This enables the planning results to directly support subsequent equipment selection, investment estimation, engineering construction plan preparation and construction organization, avoiding the problem of the output results of traditional methods being out of touch with the actual engineering construction needs.

[0017] 2. In planning scenarios involving multiple services, systems, and constraints, a direct global search across the entire network's original link space results in an extremely large decision space and is prone to generating numerous illegal solutions, such as disconnected service paths, critical services being assigned to disallowed systems, and paths failing to meet quality of service requirements. The first intelligent agent pre-constructs a feasible search space for each service, eliminating link-system combinations that clearly do not meet engineering requirements, and retaining only candidate paths with basic engineering feasibility. This allows subsequent global optimization to move beyond the original massive link space and instead focus on combining and selecting feasible candidate paths for each service. This reduces the computational overhead of invalid searches and illegal solution repairs, improves solution efficiency, enhances algorithm convergence stability, and ensures that candidate solutions entering the global optimization stage possess basic engineering feasibility.

[0018] 3. This invention utilizes a second intelligent agent to comprehensively consider the link sharing relationship between multiple services, the configuration relationship of different transmission systems, capacity occupancy, resource reuse, and system conversion costs from a network-wide perspective. It coordinates resource usage relationships among multiple services, enabling path selection, system configuration, and capacity planning to form an overall optimal solution. This ensures reliable service transmission while balancing construction economy, network reliability, and resource utilization, avoiding resource waste caused by over-configuration and preventing a decline in the communication quality of critical services due to excessive pursuit of low costs.

[0019] 4. This invention establishes adaptation constraints between service types and transmission systems to prevent critical services from being assigned to unsuitable bearer modes; it establishes matching constraints between service quality requirements and system capabilities to ensure performance guarantees for low-latency, low-jitter services; it establishes applicable rules between network layering and transmission systems, enabling the backbone layer and access layer to adopt communication methods consistent with their respective functional positioning; it establishes constraints on the backbone network loop formation rate to avoid generating tree-like networks with insufficient redundancy; and it establishes constraints on the reuse of existing engineering resources, ensuring that the planning results reflect the actual construction costs of different candidate paths. Through the combined effect of the above constraints, the planning scheme output by this invention can optimize the overall construction cost of the entire network while meeting the power system safety verification standards and engineering specifications.

[0020] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a schematic diagram of the overall process of the method described in this invention; Figure 2 This is a schematic diagram of the operation flow of the first intelligent agent in this invention; Figure 3 This is a schematic diagram of the operation process of the second intelligent agent in this invention. Detailed Implementation

[0023] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0024] Example 1: Explanation of related terms: Dual-Agent Collaborative Optimization: Dual-agent collaborative optimization refers to decomposing the planning process of a multi-service, multi-modal power communication network into two cooperating intelligent optimization units. These units respectively complete candidate path generation and global capacity quantification optimization, thereby reducing the solution complexity of the planning problem and improving optimization quality. The first intelligent agent is the service feasible path generation agent. Its core function is to preemptively eliminate illegal links that do not meet service access, hierarchy matching, and service quality requirements, ensuring that subsequent optimization processes only occur within the feasible path space. The second intelligent agent is the global optimization agent. Its core function is to coordinate link sharing, system selection, and capacity configuration among different services across the entire network, minimizing construction costs while meeting constraints.

[0025] Multi-service power communication network: A multi-service power communication network refers to a dedicated communication network capable of simultaneously carrying multiple types of power services to meet the needs of operation, control, monitoring, and management of new power systems. The services it carries include not only control-type services with high real-time and reliability requirements, such as relay protection, stability control, and distribution automation, but also services with varying requirements for bandwidth, latency, jitter, and reliability, such as electricity consumption information collection, equipment status monitoring, video surveillance, inspection image transmission, and production management.

[0026] Multimodal power communication network: A multimodal power communication network refers to a network configuration in which multiple communication systems coexist and can be used collaboratively within the same power communication planning area. These communication systems may include, but are not limited to, fiber optic communication (GPON), high-speed power line communication (HPLC), private wireless network communication (5G), and other communication methods suitable for carrying power services.

[0027] Heterogeneous communication systems refer to communication bearer methods in power communication networks that have different physical transmission mechanisms, performance attributes, and cost characteristics. This heterogeneity is mainly reflected in three aspects: First, heterogeneous physical media. Different communication systems rely on different transmission media; for example, fiber optic communication relies on optical cable resources, high-speed power line communication relies on power line carrier channels, and wireless communication relies on wireless spectrum resources. Second, heterogeneous performance capabilities. Different communication systems differ in bandwidth capacity, latency levels, jitter stability, reliability, security, and anti-interference capabilities. Third, heterogeneous construction costs. Different communication systems differ in unit bandwidth construction costs, line laying costs, equipment costs, construction difficulty, and resource reuse capabilities. Therefore, during planning, all links cannot be simply treated as homogeneous links; instead, construction costs and availability attributes should be set separately for different links and different systems.

[0028] Please see Figure 1 This invention provides a technical solution: a dual-agent collaborative optimization method for planning multi-service, multi-modal power communication networks, comprising the following steps: S1. Obtain service requirement information for various services in the power communication network and performance attribute information for various transmission systems; The business requirements information includes the source node, destination node, business type, bandwidth requirements, and quality of service requirements for each business. Among them, the quality of service requirements include latency requirements and jitter requirements. System attribute information includes the performance characteristics, applicable scope, construction attributes, and compatibility relationships between different service types and different transmission systems for each transmission system; S2. Based on business demand information and transmission system performance attribute information, establish a mathematical model of multi-service multimodal power communication network. The mathematical model has an objective function that minimizes the overall construction cost of the power communication network and sets multiple types of constraints. The mathematical model of a multi-service, multi-modal power communication network is specifically represented as follows: In the formula, Representative from the system Switch to The one-time switchover cost includes equipment, wiring, and commissioning fees. It is a 0-1 variable, representing business logic. At the node Whether the system occurs Switching, Representing a path The above adopts a system The amount of bandwidth that needs to be configured at that time. This represents the cost reduction that can be achieved through resource reuse; S21. The objective function is specifically expressed as follows: In the formula, Represents a link, Indicates the transmission system, Represents the business it carries. Represents a set of network nodes. Represents the link The above is the system The unit cost of building bandwidth. Indicates link The above adopts a system The amount of bandwidth that needs to be configured at that time. It is a 0-1 variable, representing whether or not it is in Enable resource reuse, that is: In the formula, This means that resource reuse can reduce costs. Representative from the system Switch to The one-time switchover cost includes equipment, wiring, and commissioning fees. It is a 0-1 variable, representing business logic. At the node Whether the system occurs The switching, namely: ; This shows that the cost consists of three parts: First, for each candidate link e and each transmission system t, the bandwidth configuration Be,t of that link and system is multiplied by its unit cost ce,t to obtain the corresponding bandwidth construction cost, and this is summed over all links and transmission systems; Second, if some links or transmission systems can reuse existing resources, the corresponding resource reuse benefits are deducted from the total cost; Third, if there is a conversion between different transmission systems in the service path, the corresponding system conversion cost is included in the total cost. Therefore, the objective function can directly reflect the overall construction cost formed by the allocation of network communication resources, resource reuse, and system conversion. S22. To ensure the feasibility of the planning results, this embodiment sets constraints such as service connectivity constraints, consistency constraints between bearer and construction constraints, service type and transmission system permission constraints, link layering and transmission system permission constraints, quality of service constraints, bandwidth capacity constraints, and backbone network node ring formation constraints, among which: S22.1 Service Connectivity: Flow Conservation Constraint First, there is the business connectivity constraint. For each business, the model requires that it originates from the source node and eventually reaches the destination node, maintaining a balance between inflow and outflow at intermediate nodes. This constraint is used to ensure that each business forms a continuous and complete end-to-end transmission path, avoiding path interruption, invalid forks, or failure to reach the destination node. This mathematically guarantees that each business can form a loop-free, continuous, and closed logical transmission path. Business connectivity constraints are expressed using the flow conservation formula, as follows: in, It is a 0-1 variable, representing the business logic. Does the main path use : Representative business The source node, Representative business The host node, Representing all from The starting link, Represents all pointers The link; S22.2 Consistency between infrastructure and construction: Business operations can only utilize existing institutional resources. Services can only use a link and its corresponding transmission system if that link has been built or the corresponding transmission system has been activated. This ensures strict consistency between the planning scheme and the "physical construction" and "logical transmission." The consistency constraint between transmission and construction is specifically expressed as follows: in, It is a 0-1 variable, indicating whether it is on the link. Construction system : ; S22.3 Service Type and Transmission System Licensing Constraints Different types of services have specific requirements for the security and anti-interference capabilities of the communication medium. For example, critical services such as relay protection usually require the use of dedicated fiber optic networks and strictly prohibit the use of public wireless networks or power line carriers. Therefore, the model introduces a service-system licensing matrix. Define parameters If and only if the communication system Allowed service types When the parameter is 1, this constraint mathematically limits the decision variables. The feasible domain is determined by forcibly eliminating system choices that violate technical specifications, ensuring that the model output scheme meets the power system safety verification standards. The specific restrictions on service types and transmission systems are as follows: in, It is a 0-1 variable representing the allowable matrix, i.e., according to "business category". system The given allow / disallow parameters are expressed as: ; S22.4 Link Layering and Transmission System Licensing Constraints Power communication networks have typical hierarchical and heterogeneous characteristics. According to the "Technical Guidelines for Planning and Design of Power Communication Networks", the backbone transmission layer carries the core services of the entire network and has extremely high requirements for reliability and bandwidth stability. It is usually mandatory to use fiber optic private network coverage. The access layer, on the other hand, allows for flexible selection of low-cost systems such as wireless or carrier according to the scenario. To this end, the model introduces a link-layered system matching matrix η to quantify the hierarchical rules. This constraint ensures that the selected system conforms to the construction rules of the corresponding network layer. Link layering and transmission system licensing constraints are specifically represented as follows: in, It is a 0-1 variable representing the physical link. Does the network layer allow for the construction of a system? ,Right now: ; S22.5 Service Quality Constraints The model should meet the service quality constraints. For services with low latency, low jitter or high reliability requirements, they are only allowed to choose the transmission system with the corresponding capabilities. The purpose of this type of constraint is to ensure that critical services are not allocated to communication channels that do not meet the performance requirements simply because of cost reasons. Quality of service constraints include latency requirement constraints and jitter constraints, which are specifically represented as follows: in, It is a 0-1 variable representing whether the business has a low latency requirement: It is a 0-1 variable, representing the system. Can it be considered a low-latency system? Similarly, the jitter constraint is expressed as: in, It is a 0-1 variable representing whether the business requires low jitter: It is also a 0-1 variable, representing the system. Could this be considered a low-jitter system? ; Table 1. Latency and jitter under different systems S22.6 Bandwidth Capacity Constraints Multiple services may converge on the same link and using the same transmission system. Therefore, it is necessary to calculate the total service load at this location and ensure that it does not exceed the configured capacity. The bandwidth capacity constraint is specifically expressed as follows: In the formula Representative business bandwidth requirements, Representing a path The above adopts a system The amount of bandwidth that needs to be configured at that time. It is a parameter with a value between 0 and 1, representing the upper limit of link utilization; S22.7 Backbone Network Node Ring Constraints In power communication network planning, if the goal is only to minimize construction costs, the model tends to choose a tree topology with fewer edges. Although the construction cost of the tree topology is lower, it lacks redundant links and is difficult to meet the reliability and fault tolerance requirements of the power backbone communication network. Therefore, this embodiment further sets a backbone network ring formation rate constraint to ensure that key nodes in the backbone layer have the necessary ring network access capability. The loop constraint of backbone network nodes is specifically represented as follows: in, Represents all connected nodes The set of candidate links, It is a 0-1 variable, representing the link. Whether it is adopted or constructed, i.e.: Representative node Existing links, It is also a 0-1 variable, representing whether this node forms a cycle, that is: when hour, Take 1; if 0 or 1, Set to 0; Represents the set of key nodes in the backbone network. This represents the number of critical nodes in the backbone network. It should be noted that the backbone network is a ring network structure, so the ring formation rate must be 1, while there is no such requirement for the access layer; S3. Construct the first intelligent agent. Based on the business requirement information and the transmission system performance attribute information, the first intelligent agent constructs a feasible search space for each business, eliminates infeasible paths, and generates a set of candidate paths for a single business within the feasible search space. Each candidate path includes a link sequence and a transmission system sequence. The first intelligent agent is used to solve the path generation problem at the single service level. This agent can be called the single-service feasible path generation agent. The first intelligent agent adopts a path generation architecture based on the K-shortest path algorithm, including a feasible edge screening layer, a service-specific weighted graph construction layer, a K-shortest path search layer, and a candidate path output layer. Among them, the feasible edge screening layer is used to screen candidate links and their optional transmission systems based on the service type and transmission system permission relationship, service quality requirements, link engineering availability, and transmission system performance attributes, retaining the "link-system" combinations that meet the constraints as searchable edges. The service-specific weighted graph construction layer is used to use power communication network nodes as graph nodes, the screened "link-system" combinations as graph edges, and determine the graph edge weights based on link construction costs, system bandwidth costs, resource reuse benefits, and system conversion costs to form a service-specific weighted graph for a single service. The K-shortest path search layer is used in the service-specific weighted graph to search for the source node of the service as the starting point and the destination node as the ending point, using K-shortest path algorithm. The shortest path algorithm searches for K feasible paths with low cost that meet the constraints. The candidate path output layer represents each feasible path as a sequence of nodes, a sequence of links, and a sequence of corresponding transmission systems, and outputs a set of candidate paths for the service. This set serves as the input for the second agent to perform global collaborative optimization. The input for the first agent includes the candidate topology of the power communication network, the set of transmission systems, the source and destination nodes of the service, the service type, the service requirements, and various adaptation rules. For each service, the first agent does not blindly search through all links in the entire network, but first constructs a feasible search space specifically for the service based on its own requirements. Specifically, for business The first intelligent agent sequentially judges whether each candidate link and its optional transmission system meet the following conditions: First, whether the transmission system allows the carrying of this type of service; Second, whether the network layer where the link is located allows the deployment of the transmission system; Third, whether the transmission system meets the service quality requirements of the service; Fourth, whether the combination of link and system has basic engineering availability.

[0029] Only "link-system" combinations that simultaneously meet the above conditions will be retained as searchable edges for that service. Thus, the first intelligent agent can form a dedicated search space for each service, which has eliminated obviously illegal choices. Subsequently, the first intelligent agent performs path search within this search space to provide services. Generate several lines from the source node To the destination node Alternative paths; Each alternative path includes not only the physical links it traverses, but also the transmission protocol used in each link segment. This allows us to determine the service... The set of alternative paths: in, Indicates business The Alternative routes; Specifically, the reason for using the first intelligent agent for preprocessing is that this type of planning problem involves service routing, system selection, capacity configuration and cost optimization. If a global search is performed directly on the entire network link and all transmission systems, the decision space will be very large and it will easily generate a large number of illegal solutions that do not meet the requirements of connectivity, service compatibility or service quality. By generating a set of alternative paths for each service in advance, the first intelligent agent can eliminate obviously infeasible paths in advance. This prevents subsequent global optimization from dealing with the original massive link space, and instead allows for the selection of combinations among feasible candidate paths for each service. This reduces invalid searches and lowers the dimensionality of the solution, making the subsequent optimization process more stable and efficient, and providing a high-quality candidate solution foundation for the global optimization of the second intelligent agent. S4. Summarize the candidate path sets of each generated single business to form a full business candidate path set; A feasible search space is constructed for all business operations, and the candidate path sets for individual businesses are aggregated to form a complete set of candidate paths for all businesses: In the formula, Total number of business transactions; S5. Construct a second intelligent agent. The second intelligent agent takes the set of candidate paths for all services as input, and performs global iterative optimization with the objective function as the goal under the premise of satisfying multiple constraints. After reaching the preset termination condition, it outputs the final planning result. The final planning result includes the final bearer path of each service, the bandwidth configuration of each transmission system on each link, and the overall construction cost of the entire network. A second intelligent agent is constructed to solve the combinatorial optimization problem across the entire network. This intelligent agent can be called the global collaborative optimization intelligent agent. The input of the second intelligent agent is the set of all service alternative paths generated by the first intelligent agent. Its core task is to select a final carrying path from the set of alternative paths for each service so that the overall construction cost of the network solution formed by all services is optimal under the premise of meeting the constraints. The second intelligent agent adopts a global collaborative optimization architecture based on genetic algorithms, including a path index encoding layer, a population initialization layer, a chromosome decoding layer, a network-wide resource statistics layer, a fitness evaluation layer, a genetic iterative update layer, and an optimal solution output layer. The path index encoding layer represents the entire set of candidate paths for the entire service as the search space of the genetic algorithm. Let the entire set of candidate paths be denoted as... ,in For the number of business transactions, Indicates business Given a set of candidate paths, a candidate planning scheme is encoded as a chromosome. ,in , indicating business Select its candidate path set The first in The system consists of several layers: a path initialization layer, a population initialization layer, and a chromosome decoding layer. The population initialization layer randomly generates multiple chromosomes that satisfy the path index range, forming an initial candidate planning scheme set. The chromosome decoding layer extracts the actual bearer path, link sequence, and transmission system sequence corresponding to each service based on the path index in each chromosome, and generates the corresponding service bearer relationship. The network-wide resource statistics layer summarizes the bandwidth requirements of all services according to the two dimensions of links and transmission systems, obtains the service superimposed load on each link and each transmission system, and determines the bandwidth configuration, construction status, resource reuse status, and system conversion status of the corresponding links and transmission systems. The fitness evaluation layer calculates the overall network construction cost of each candidate planning scheme according to the objective function, and checks whether service connectivity, bearer-construction consistency, service-system licensing relationship, bandwidth capacity, and backbone network ring formation constraints are satisfied, and identifies violations of the overall construction cost and constraints. Penalties are used as the basis for fitness evaluation. The genetic iterative update layer performs selection, crossover, and mutation operations on the current population. Selection retains candidate schemes with higher fitness, crossover swaps path indices of some services between two candidate schemes to generate new service path combinations, and mutation randomly changes the path index of a service, switching it to another path in its candidate path set, thus expanding the global search scope. When the number of iterations reaches a preset limit, or the optimal objective function value no longer improves after several consecutive generations, the optimal solution output layer outputs the chromosome with the best fitness and that satisfies the constraints, and decodes it into the final planning result. The final planning result includes the final bearer path for each service, the transmission system configuration results on each link, the final bandwidth configuration results of each transmission system across the entire network, and the overall network construction cost. This is based on the descriptions in the original draft regarding "path index encoding, decoding, link-system load statistics, objective function and constraint penalty evaluation, selection, crossover, mutation iteration, and output of the final planning result."

[0030] In one implementation, the second agent can represent a candidate planning scheme using a path index encoding method, assuming a total of If there are several business operations, then a candidate solution can be represented as: in, Indicates business Select its set of alternative paths The first in One path. The advantage of this encoding method is that each gene position corresponds to one service path selection, and the chromosome length is only related to the number of services, without the need to encode every link and every system in the entire network separately; For any candidate scheme, the second intelligent agent first performs decoding to obtain the actual bearer path and corresponding transmission system of all services; Subsequently, the service load was statistically analyzed according to two dimensions: link and transmission system. The bandwidth required for each system on each link was calculated, as follows: For any candidate planning scheme, after the second intelligent agent decodes and obtains the path selection results for each service, it performs a process for any link. and transmission system Judge each business individually Whether to use this link and this system; if Then the bandwidth requirement of this service Included in the link upper system The business load, if If so, it will not be included; Therefore, the link upper system The service load is: According to bandwidth constraints: The required bandwidth for this link and this system can be obtained. It should meet the following requirements: Since the objective function aims to minimize the overall network construction cost, under the premise of satisfying bandwidth constraints, the link... upper system The bandwidth allocation is taken as the minimum value that satisfies the above constraints, that is: Therefore, for all , Calculate each one individually to obtain the bandwidth required for each transmission system on each link; Then, the overall construction cost of the scheme is calculated according to the objective function, and the constraints such as capacity, service quality, and engineering rules are checked. New candidate schemes are continuously generated through operations such as selection, crossover, and mutation, and the schemes are evaluated according to the fitness function. Furthermore, the fitness function can be composed of the comprehensive construction cost and the constraint penalty term. Specifically, for any candidate planning scheme... The second intelligent agent first determines the location of each gene. Decoding to obtain business The selected candidate path is used to determine the service carrying variables. Link construction variables Institutional Development Variables Resource reuse variables System transformation variables and links upper system bandwidth configuration Subsequently, the overall construction cost of the candidate planning scheme is calculated based on the objective function: And construct constraint penalty terms based on various constraints: in, This is a penalty item for business connectivity constraints. To accommodate the constraints and penalties related to consistency between the carrying capacity and construction, For business-related institutional restrictions and penalties, This is a penalty term for bandwidth capacity constraints. The constraint penalty term for the ring formation of backbone network nodes. For the corresponding penalty weight; Each penalty item is calculated as follows: in, in, Based on the comprehensive construction cost and constraint penalty terms, the second agent constructs a fitness function: ; in, Indicates candidate planning schemes The fitness of the candidate program. If the candidate program satisfies all constraints, then At this time, the overall construction cost The lower the fitness The higher the penalty, the better. If a candidate planning scheme has issues such as service disconnection, service using undeveloped infrastructure resources, service mismatch with transmission infrastructure, insufficient link bandwidth capacity, or backbone network nodes not meeting ring-forming requirements, then a corresponding penalty item will be applied. Increasing the fitness of the proposed solution decreases its fitness. The second agent performs the above decoding, cost calculation, constraint penalty calculation, and fitness calculation for each candidate planning solution in the population, and performs selection, crossover, and mutation iterations based on the fitness value until a preset termination condition is reached. Finally, it outputs the candidate planning solution with the highest fitness as the final planning result.

[0031] A solution has high fitness when it has low cost and meets the constraints; when a solution has insufficient capacity, does not meet the engineering rules, or is otherwise infeasible, its fitness is reduced by a penalty term. After multiple iterations, the second intelligent agent outputs the current optimal global planning scheme. This scheme includes the following results: First, the final bearer path for each service; second, the configuration results of different transmission systems on each link; third, the final bandwidth configuration results of each transmission system across the entire network; and fourth, the overall construction cost of the entire network.

[0032] Specifically, the second intelligent agent addresses the problem of "how to achieve overall optimization when all services are combined." It does not select the cheapest path for each service in isolation, but rather comprehensively considers the relationship between shared links for multiple services, transmission system configuration, capacity requirements, and construction costs across the entire network, ultimately arriving at a power communication network planning scheme that balances economy, reliability, and resource utilization.

[0033] Example 2: This embodiment provides a dual-agent cooperative optimization multi-service multimodal power communication network planning system, used to implement the dual-agent cooperative optimization multi-service multimodal power communication network planning method described in Embodiment 1, including: The information acquisition module is used to acquire service requirement information for various services in the power communication network and transmission system performance attribute information for various transmission systems. The model building module is used to establish a mathematical model of a multi-service multimodal power communication network based on the service requirement information and the transmission system performance attribute information. The mathematical model has an objective function that minimizes the overall construction cost of the power communication network and sets multiple types of constraints. The first intelligent agent construction module is used to construct the first intelligent agent. The first intelligent agent constructs a feasible search space for each service based on the service requirement information and the transmission system performance attribute information, filters out infeasible paths, and generates a candidate path set for a single service within the feasible search space. Each candidate path includes a link sequence and a transmission system sequence. The dataset aggregation module is used to aggregate the candidate path sets of each generated single business to form a full business candidate path set; The second intelligent agent construction module is used to construct a second intelligent agent. The second intelligent agent takes the set of candidate paths for all services as input, and performs global iterative optimization with the objective function as the goal under the premise of satisfying multiple constraints. After reaching the preset termination condition, it outputs the final planning result. The final planning result includes the final bearer path of each service, the bandwidth configuration of each transmission system on each link, and the overall construction cost of the entire network.

[0034] In the description of this specification, the references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0035] The foregoing has shown and described the basic principles, main features, and advantages of this application. Those skilled in the art should understand that this application is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of this application. Various changes and modifications can be made to this application without departing from the spirit and scope thereof, and all such changes and modifications fall within the scope of the claims of this application.

Claims

1. A planning method for multi-service, multi-modal power communication networks using dual-agent collaborative optimization, characterized in that: Includes the following steps: Obtain service requirement information for various services in the power communication network and performance attribute information for various transmission systems; Based on the service requirements information and the transmission system performance attribute information, a mathematical model of a multi-service multimodal power communication network is established. The mathematical model has an objective function that minimizes the overall construction cost of the power communication network and sets multiple types of constraints. A first intelligent agent is constructed. Based on the service requirement information and the transmission system performance attribute information, the first intelligent agent constructs a feasible search space for each service, eliminates infeasible paths, and generates a set of candidate paths for a single service within the feasible search space. Each candidate path includes a link sequence and a transmission system sequence. The candidate path sets of each generated single business are aggregated to form the full business candidate path set; A second intelligent agent is constructed. The second intelligent agent takes the set of candidate paths for all services as input and performs global iterative optimization with the objective function as the goal under the premise of satisfying multiple constraints. After reaching the preset termination condition, the final planning result is output. The final planning result includes the final bearer path of each service, the bandwidth configuration of each transmission system on each link, and the overall construction cost of the entire network.

2. The dual-agent cooperative optimization method for planning multi-service multimodal power communication networks according to claim 1, characterized in that: The business requirement information includes the source node, destination node, business type, bandwidth requirement, and quality of service requirement for each business, wherein the quality of service requirement includes latency requirement and jitter requirement. The system attribute information includes the performance characteristics, applicable scope, construction attributes, and compatibility relationships between different service types and different transmission systems for each transmission system.

3. The dual-agent cooperative optimization method for planning multi-service multimodal power communication networks according to claim 1, characterized in that: The mathematical model of a multi-service, multi-modal power communication network is specifically represented as follows: In the formula, Representative from the system Switch to The one-time switchover cost includes equipment, wiring, and commissioning fees. It is a 0-1 variable, representing business logic. At the node Whether the system occurs Switching, Representing a path The above adopts a system The amount of bandwidth that needs to be configured at that time. This represents the cost reduction that can be achieved through resource reuse.

4. The dual-agent cooperative optimization method for planning multi-service multimodal power communication networks according to claim 1, characterized in that: The objective function is specifically expressed as follows: In the formula, Represents a link, Indicates the transmission system, Represents the business it carries. Represents a set of network nodes. Represents the link The above is the system The unit cost of building bandwidth. Indicates link The above adopts a system The amount of bandwidth that needs to be configured at that time. It is a 0-1 variable, representing whether or not it is in Enable resource reuse, that is: In the formula, This means that resource reuse can reduce costs. Representative from the system Switch to The one-time switchover cost includes equipment, wiring, and commissioning fees. It is a 0-1 variable, representing business logic. At the node Whether the system occurs The switching, namely: 。 5. The dual-agent cooperative optimization method for planning multi-service multimodal power communication networks according to claim 1, characterized in that: The various constraints specifically include service connectivity constraints, consistency constraints between bearer and construction, service type and transmission system permission constraints, link layering and transmission system permission constraints, quality of service constraints, bandwidth capacity constraints, and backbone network node ring formation constraints.

6. The dual-agent cooperative optimization method for planning multi-service multimodal power communication networks according to claim 5, characterized in that: The business connectivity constraints are expressed by the flow conservation formula, as follows: in, It is a 0-1 variable, representing the business logic. Does the main path use : Representative business The source node, Representative business The host node, Representing all from The starting link, Represents all pointers The link; The consistency constraint between load and construction is specifically expressed as follows: in, It is a 0-1 variable, indicating whether it is on the link. Construction system : ； The specific restrictions on service types and transmission system permissions are as follows: in, It is a 0-1 variable representing the allowable matrix, i.e., according to "business category". system The given allow / disallow parameters are expressed as: ； The link layering and transmission system licensing constraints are specifically represented as follows: in, It is a 0-1 variable representing the physical link. Does the network layer allow for the construction of a system? ,Right now: ； The quality of service constraints include latency requirement constraints and jitter constraints, specifically as follows: in, It is a 0-1 variable representing whether the business has a low latency requirement: It is a 0-1 variable, representing the system. Can it be considered a low-latency system? Similarly, the jitter constraint is expressed as: in, It is a 0-1 variable representing whether the business requires low jitter: It is also a 0-1 variable, representing the system. Could this be considered a low-jitter system? ； The bandwidth capacity constraint is specifically expressed as follows: In the formula Representative business bandwidth requirements, Representing a path The above adopts a system The amount of bandwidth that needs to be configured at that time. It is a parameter with a value between 0 and 1, representing the upper limit of link utilization; The ring constraint of the backbone network nodes is specifically represented as follows: in, Represents all connected nodes The set of candidate links, It is a 0-1 variable, representing the link. Whether it is adopted or constructed, i.e.: Representative node Existing links, It is also a 0-1 variable, representing whether this node forms a cycle, that is: when hour, Take 1; if 0 or 1, Set to 0; Represents the set of key nodes in the backbone network. This represents the number of key nodes in the backbone network.

7. The dual-agent cooperative optimization method for planning multi-service multimodal power communication networks according to claim 5, characterized in that: The first intelligent agent is a single-business feasible path generation intelligent agent. The first intelligent agent adopts a path generation architecture based on the K shortest path algorithm, which specifically includes a feasible edge screening layer, a business-specific weighted graph construction layer, a K shortest path search layer, and a candidate path output layer. The first intelligent agent takes as input the business requirement information, system attribute information, and various constraints, and outputs a set of candidate paths for each business. The specific processing flow is as follows: For each business k, a feasible search space is constructed, infeasible paths are eliminated, and a set of candidate paths is generated within the feasible search space. Specifically, it is expressed as: in, Indicates business The Alternative routes; A feasible search space is constructed for all business operations, and the candidate path sets for individual businesses are aggregated to form a complete set of candidate paths for all businesses: In the formula, This represents the total number of business transactions.

8. The dual-agent cooperative optimization method for planning multi-service multimodal power communication networks according to claim 7, characterized in that: When the first intelligent agent constructs the feasible search space, it sequentially determines whether each candidate service link and transmission system combination satisfies the following: The transmission system allows the carrying of the service type, the network layer where the candidate link is located allows the deployment of the transmission system, and the transmission system meets the quality of service requirements of the service. Candidate links and transmission scheme combinations are retained as searchable edges only when all conditions are met simultaneously; Within the feasible search space, the first intelligent agent generates several candidate paths from the source node to the destination node for each service based on the flow conservation formula. Each candidate path contains a complete node link sequence and a corresponding transmission system sequence.

9. The dual-agent cooperative optimization method for planning multi-service multimodal power communication networks according to claim 1, characterized in that: The second intelligent agent takes the full set of candidate paths for all services as input. The second intelligent agent adopts a global collaborative optimization architecture based on genetic algorithm, including a path index encoding layer, a population initialization layer, a chromosome decoding layer, a full network resource statistics layer, a fitness evaluation layer, a genetic iterative update layer, and an optimal solution output layer. The second intelligent agent takes the entire set of candidate paths for all services as input, and performs global iterative optimization with the objective function as the goal, under the premise of satisfying multiple constraints. After reaching the preset termination condition, it outputs the final planning result. The specific steps are as follows: The full-service candidate path set As input, where Total number of business transactions; Candidate planning schemes are generated using a path index encoding method, and the candidate planning schemes are represented as follows: in For the purpose of indicating business From its candidate path set Select the first The index of the path; Decode the candidate planning scheme This allows us to obtain the final bearer path and corresponding transmission system for each service. In response to the decoding results, the service load is statistically analyzed according to two dimensions: link and transmission system. The bandwidth required for each system on each link is calculated, and the comprehensive construction cost of the candidate planning scheme is calculated according to the objective function. The bandwidth is expressed as: ; Check whether the candidate planning scheme satisfies the bandwidth constraint and the ring formation constraint of the backbone network nodes in the constraint set. Based on the comprehensive construction cost and constraint check results, a fitness function is constructed, where solutions that satisfy the constraints and have lower costs have high fitness, while solutions that do not satisfy the constraints have their fitness reduced through a penalty term, as follows: For any candidate programming scheme First, based on each gene locus Decoding to obtain business The selected candidate path is used to determine the service carrying variables. Link construction variables Institutional Development Variables Resource reuse variables System transformation variables and links upper system bandwidth configuration ; Calculate the overall construction cost of this candidate planning scheme based on the objective function: And construct constraint penalty terms based on various constraints: in, This is a penalty item for business connectivity constraints. To accommodate the constraints and penalties related to consistency between the carrying capacity and construction, For business-related institutional restrictions and penalties, This is a penalty term for bandwidth capacity constraints. The constraint penalty term for the ring formation of backbone network nodes. For the corresponding penalty weight; Each penalty item is calculated as follows: in, in, Based on the comprehensive construction cost and constraint penalty terms, the second agent constructs a fitness function: in, Indicates candidate planning schemes The fitness of; For each candidate planning scheme in the population, selection, crossover, and mutation are performed iteratively based on fitness until a preset termination condition is met. The candidate planning scheme with the highest fitness is then output as the final planning result.

10. A dual-agent cooperative optimization multi-service multimodal power communication network planning system, used to implement the dual-agent cooperative optimization multi-service multimodal power communication network planning method according to any one of claims 1 to 9, characterized in that, include: The information acquisition module is used to acquire service requirement information for various services in the power communication network and transmission system performance attribute information for various transmission systems. The model building module is used to establish a mathematical model of a multi-service multimodal power communication network based on the service requirement information and the transmission system performance attribute information. The mathematical model has an objective function that minimizes the overall construction cost of the power communication network and sets multiple types of constraints. The first intelligent agent construction module is used to construct the first intelligent agent. The first intelligent agent constructs a feasible search space for each service based on the service requirement information and the transmission system performance attribute information, filters out infeasible paths, and generates a candidate path set for a single service within the feasible search space. Each candidate path includes a link sequence and a transmission system sequence. The dataset aggregation module is used to aggregate the candidate path sets of each generated single business to form a full business candidate path set; The second intelligent agent construction module is used to construct a second intelligent agent. The second intelligent agent takes the set of candidate paths for all services as input, and performs global iterative optimization with the objective function as the goal under the premise of satisfying multiple constraints. After reaching the preset termination condition, it outputs the final planning result. The final planning result includes the final bearer path of each service, the bandwidth configuration of each transmission system on each link, and the overall construction cost of the entire network.

Images