Layered dynamic routing system and method for computing power-energy fusion network

Abstract

The invention belongs to the field of urban traffic management, and provides a hierarchical dynamic routing system and method for a computing power-energy fusion network, and the method comprises the steps: dividing a fusion network graph into a plurality of region sub-graphs, and dividing a path table and a connection point path table based on each region sub-graph; monitoring the running states of the road, the edge computing node and the charging station in real time, and predicting the road congestion intensity, the edge computing node queuing estimation information and the charging constraint; and determining a candidate path set in the partition path table and the connection point path table according to an online request of the vehicle, screening each candidate path in the candidate path set based on the road congestion intensity, the edge computing node queuing estimation information and the charging constraint, and determining an optimal path. According to the method, online retrieval is converted into rapid combination and evaluation of small candidate sets from full-graph traversal; traffic, energy and computing power are incorporated into the same decision framework, and systematicness deviation caused by evaluation of the future at present is avoided.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of urban traffic management, and particularly relates to a layered dynamic routing system and method for a computing power-energy fusion network. BACKGROUND

[0002] The statements in this section merely provide background information related to the present application and do not necessarily constitute prior art.

[0003] With the deep integration of intelligent electric vehicles and Internet of Things technology, modern urban transportation systems are gradually evolving into a complex coupled network that integrates traffic flow, energy flow, and computing power flow (further described as a vehicle-energy-computing fusion network). In this context, electric vehicles are no longer just transportation vehicles, but also mobile computing nodes and distributed energy storage units. Through dynamic interaction with charging facilities, edge computing nodes, and other infrastructure, they achieve intelligent energy replenishment, collaborative offloading of computing tasks, and real-time data processing. Accordingly, the vehicle path planning problem has expanded from traditional geographical space trajectory optimization to a multi-dimensional decision-making process that needs to coordinate travel time, energy consumption status, computing resource availability, and energy supply dynamics. It has become a typical multi-constrained, multi-objective, and real-time collaborative optimization problem.

[0004] In terms of path planning methods, traditional shortest path algorithms such as Dijkstra's algorithm and Floyd-Warshall algorithm have good routing performance in static road networks, but their computational complexity increases significantly with the size of the network, making them difficult to apply to large-scale dynamic road networks at the city or even regional level. To balance computational efficiency and routing quality, researchers have proposed various hierarchical routing strategies. These methods effectively reduce the computational burden of real-time queries by abstracting network hierarchical structures, making them more suitable for efficient path generation in dynamic traffic conditions.

[0005] However, existing routing mechanisms still mainly focus on optimizing a single traffic time or distance indicator, and have not fully considered the complex needs of electric vehicles in energy-sensitive and computing-dependent scenarios. Specifically, electric vehicles need to balance battery charge status, charging station availability and electricity price fluctuations, edge server computing load and task processing delay, and other multi-dimensional dynamic constraints during travel. If the real-time status of energy and computing resources is ignored in path planning and decisions are made based solely on static or historical data, it can easily lead to problems such as vehicles running out of power, computing tasks timing out, or service interruptions. Currently, research on collaborative optimization under the triple constraints of travel, energy charging, and computing is still in its early stages, and there is a lack of systematic fusion modeling methods and real-time response mechanisms.

[0006] In the face of large-scale, high-dynamic and strong-coupling characteristics of the vehicle-energy-computing integrated network, the current method cannot comprehensively perceive the network state and dynamically coordinate multiple types of resources, and it is difficult to support joint optimization of path planning and task scheduling. SUMMARY

[0007] To solve the above problems, the application proposes a layered dynamic routing system and method for computing power-energy integrated network, which uses two-level computing paradigm of offline recalculation and online light splicing as the base. The offline side structurally preprocesses the global road network by region to generate intra-regional paths and connection point backbone paths that can be directly reused; the online side considers driving time, energy consumption, task delay and constraint feasibility in a small-scale candidate set to complete joint optimal decision.

[0008] According to some embodiments, the first aspect of the application provides a layered dynamic routing method for computing power-energy integrated network, which adopts the following technical solution: A layered dynamic routing method for computing power-energy integrated network includes: Divide the vehicle-energy-computing integrated network graph into several regional subgraphs, and generate a partition path table and a connection point path table based on each regional subgraph; Real-time monitor the running state of roads, edge computing nodes and charging stations, and predict road congestion intensity, edge computing node queuing estimation information and charging constraints; Determine a candidate path set in the partition path table and the connection point path table according to the online request of the vehicle, and screen each candidate path in the candidate path set based on road congestion intensity, edge computing node queuing estimation information and charging constraints to determine the optimal path.

[0009] Further, the vehicle-energy-computing integrated network graph is divided into several regional subgraphs, and a partition path table and a connection point path table are generated based on each regional subgraph, which includes: According to the spatial relationship, the vehicle-energy-computing integrated network graph is divided into several regional subgraphs, and a connection point virtual graph is generated based on the identification of boundary connection points of each regional subgraph; Calculate the all-pair shortest path matrix and predecessor pointer in each regional subgraph, and aggregate the time, energy and delay statistics triplets to obtain a partition path record, and the partition path records of each regional subgraph constitute a partition path table; Calculate the backbone shortest path on the connection point virtual graph to form a connection point path table.

[0010] Further, the vehicle-energy-computing integrated network graph is divided into several regional subgraphs, and a partition path table and a connection point path table are generated based on each regional subgraph, which includes: A road, a charging station, and an edge computing are taken as nodes in a global traffic, energy, and computing fusion graph, a basic driving time from a road to the charging station, and an equivalent energy consumption are taken as fusion edges, and a car-energy-computing fusion network graph is constructed; A region division mapping is determined according to a spatial relationship, and the car-energy-computing fusion network graph is divided into a plurality of region subgraphs based on the region division mapping; A virtual connection point is determined based on an intersection relationship between a cross-region edge and a boundary of the region subgraphs, a boundary connection point is determined according to a splitting cost of the virtual connection point, and a connection point virtual graph is constructed according to all the boundary connection points.

[0011] Further, the candidate path set is determined in the partition path table and the connection point path table according to the online request of the vehicle, and includes: A starting point region and an ending point region are determined based on the online request of the vehicle; If the starting point region and the ending point region are the same region, a shortest path from the starting point to the ending point in the region is determined according to the partition path table, and the candidate path set is constructed; If the starting point region and the ending point region are not the same region, a starting point intra-region path is determined in the region subgraph of the starting point, a backbone path is determined in the connection point virtual graph, and an ending point intra-region path is determined in the region subgraph of the ending point, and the three paths are spliced to obtain the candidate path set.

[0012] Further, each candidate path in the candidate path set is screened based on a road congestion intensity, edge computing node queuing estimation information, and charging constraints, and an optimal path is determined, and includes: An edge weight cost, a computing task execution cost, and a charging cost are determined according to the road congestion intensity, the edge computing node queuing estimation information, and the charging constraints; Each candidate path in the candidate path set is screened to determine the optimal path, with an optimal comprehensive cost composed of the edge weight cost, the computing task execution cost, and the charging cost as a target.

[0013] Further, the edge weight cost, the computing task execution cost, and the charging cost are determined according to the road congestion intensity, the edge computing node queuing estimation information, and the charging constraints, and include: A basic edge weight is determined according to a basic driving time and energy consumption of a road section, an amplification factor is determined by using a load rate of the road congestion intensity, the edge computing node queuing estimation information, and an energy sufficiency of the charging constraint, a dynamic edge weight is determined based on the amplification factor and the basic edge weight, and serves as the edge weight cost; A waiting time and an actual processing time are determined based on an arrival rate and a service rate of a computing task of the edge node queuing estimation information, and serve as the computing task execution cost; An expected charging time is determined by using an average charging power and a charging queuing time, and serves as the charging cost.

[0014] According to some embodiments, a second aspect of the present application provides a layered dynamic routing system for a computing power-energy fusion network, adopting the following technical solution: A layered dynamic routing system for a computing power-energy fusion network, comprising: An offline pre-computation module, configured to divide a vehicle-energy-computing fusion network graph into a plurality of regional sub-graphs, and generate a partition path table and a connection point path table based on each regional sub-graph; A state monitoring and prediction module, configured to monitor the running states of roads, edge computing nodes and charging stations in real time, and predict road congestion intensity, edge computing node queuing estimation information and charging constraints; A cooperative routing module, configured to determine a candidate path set in the partition path table and the connection point path table according to an online request of a vehicle, and screen each candidate path in the candidate path set based on road congestion intensity, edge computing node queuing estimation information and charging constraints to determine an optimal path.

[0015] According to some embodiments, a third aspect of the present application provides a computer readable storage medium.

[0016] A computer readable storage medium, having a computer program stored thereon, the program being executed by a processor to implement the steps in the layered dynamic routing method for a computing power-energy fusion network according to the first aspect described above.

[0017] According to some embodiments, a fourth aspect of the present application provides a computer device.

[0018] A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps in the layered dynamic routing method for a computing power-energy fusion network according to the first aspect described above when executing the program.

[0019] According to some embodiments, a fifth aspect of the present application provides a computer program product or computer program.

[0020] A computer program product or computer program, comprising computer instructions stored in a computer readable storage medium, the computer instructions being read by a processor of a computer device from the computer readable storage medium, and the processor executing the computer instructions to cause the computer device to perform the steps in the layered dynamic routing method for a computing power-energy fusion network according to the first aspect described above.

[0021] Compared with the prior art, the present application has the following beneficial effects: The application takes "offline recalculation + online light splicing" as the base, converts the online search from full map traversal to fast combination and evaluation of small candidate set through regional decomposition, ZPT / JPT precalculation and DB-Cache resident, and the same area request can be hit in O(1), and the cross-area request is only selected on the limited set of three splicing, the query delay is weakly related to the size of the full map.

[0022] In terms of optimization target and feasibility guarantee, the application integrates traffic, energy and computing power into the same decision framework: on the one hand, the time-varying evaluation of edge weight and node cost is driven by the dynamic situation (congestion, energy adequacy, computing power load) of ETA alignment, avoiding the systematic deviation of evaluating the future with the present. On the other hand, the RCSP label propagation first checks the feasibility (SoC bottom line, total time limit, task completion), and then compares the cost, significantly reducing the risk of power failure and timeout. BRIEF DESCRIPTION OF DRAWINGS

[0023] The accompanying drawings, which form a part of this specification, are included to provide a further understanding of the application, and are incorporated in and constitute a part of this specification. The embodiments of these drawings are set forth to explain the present application and are not intended to limit the present application.

[0024] Figure 1 is the overall architecture diagram of a layered dynamic routing system for a computing power-energy integrated network according to the application. DETAILED DESCRIPTION

[0025] The application will be further described below in conjunction with the drawings and embodiments.

[0026] It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the application. Unless otherwise indicated, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the application pertains.

[0027] It should be noted that the terms used herein are only for the purpose of describing specific embodiments and are not intended to limit exemplary embodiments according to the application. As used herein, the singular form is intended to include the plural form unless the context clearly indicates otherwise, and it should be understood that when the terms "comprise" and / or "include" are used in the specification, there is a presence of the features, steps, operations, devices, components and / or combinations thereof.

[0028] The embodiments in the application and the features in the embodiments can be combined with each other without conflict.

[0029] Embodiment one The embodiment provides a layered dynamic routing method for a computing power-energy fusion network. The method is applied to a server for example. It can be understood that the method can also be applied to a terminal and can also be applied to a system including a terminal and a server and is realized through interaction of the terminal and the server. The server can be a physical server, a server cluster composed of multiple physical servers or a distributed system. It can also be a cloud server providing basic cloud computing services such as cloud service, cloud database, cloud computing, cloud function, cloud storage, network server, cloud communication, middleware service, domain name service, security service CDN, and big data and artificial intelligence platform. The terminal can be a smart phone, a tablet computer, a notebook computer, a desktop computer, a smart speaker, a smart watch and the like, but is not limited thereto. The terminal and the server can be directly or indirectly connected through wired or wireless communication, which is not limited in the application. In the embodiment, the method includes the following steps: The vehicle-energy-computing fusion network graph is divided into a plurality of regional subgraphs, and a partition path table and a connection point path table are generated based on each regional subgraph; The running states of the road, the edge computing node and the charging station are monitored in real time, and the road congestion intensity, the edge computing node queuing estimation information and the charging constraint are predicted; According to the online request of the vehicle, a candidate path set is determined in the partition path table and the connection point path table, and each candidate path in the candidate path set is screened based on the road congestion intensity, the edge computing node queuing estimation information and the charging constraint to determine an optimal path.

[0030] In a large-scale urban road network, the embodiment uses offline recalculation and online light splicing as a base, superimposes two-layer mechanisms of energy and computing power cooperation and feasibility label propagation, so that online decision-making only completes the feasible optimal of three targets of time, energy consumption and task in a small-scale candidate set. Offline partition, connection points and two types of precomputed path tables (partition path table, Zone Path Table, ZPT and connection point path table, Junction Path Table, JPT) reside in the database cache (Database Cache, DB-Cache), ensuring that the same area query can hit, cross-area query can quickly generate a candidate set through three splicing, and then the Dijkstra or Yen is used to select the optimal path on the candidate set. DB-Cache supports parallel access and incremental write-back, facilitating fast update under topology or weight change.

[0031] Step S1: The vehicle-energy-computing fusion network graph is divided into a plurality of regional subgraphs, a connection point virtual graph is generated based on the boundary connection points of each regional subgraph, full-pair shortest paths in each regional subgraph are calculated to generate a partition path table, and backbone shortest paths on the connection point virtual graph are calculated to form a connection point path table, including: This step is mainly to perform offline recalculation. The global traffic, computing power and energy fusion graph is divided into several regions according to the spatial relationship, the boundary connection points are identified and embedded, and the all-pair shortest path precalculation is completed on the subgraph in the region and the virtual graph of the connection point, and the structured path table is resident in the database cache (DB-Cache) for online stage constant / quasi-constant time reuse. To ensure that the cross-region edge splitting does not change the global optimality, the system introduces virtual connection points at the boundary and proportionally allocates edge weights, so that the equivalent travel cost before and after splitting is strictly conserved, thereby improving the modular degree of cross-region combination while maintaining the correctness of the path.

[0032] Step S1.1: According to the spatial relationship, the car-energy-computing fusion network graph is divided into several regional subgraphs, and the boundary connection points are identified to generate a connection point virtual graph, the process is as follows: The road, charging station and edge computing in the global traffic, energy and computing power fusion graph are taken as nodes, the basic driving time and equivalent energy consumption from road to charging station are taken as fusion edges, and a car-energy-computing fusion network graph is constructed, including: Let the global traffic, energy and computing power fusion graph be modeled as a car-energy-computing fusion network graph , the car-energy-computing fusion network graph is a directed weighted graph, wherein the vertex set contains road nodes, charging station nodes and edge computing nodes, and the edge set contains fusion edges composed of basic driving time and equivalent energy consumption ; The basic driving time is the basic driving time from road to charging station, and the equivalent energy consumption is the equivalent energy consumption on edge , including road driving energy consumption, energy consumption for accessing charging station nodes (queuing and parking), and computing energy consumption.

[0033] According to the spatial relationship, the region division mapping is determined, and the car-energy-computing fusion network graph is divided into several regional subgraphs based on the region division mapping, specifically: First, determine the region division mapping , which is used for standard graph division, and the target considers balance and minimum cut edge, typical targets are:

[0034] Among them, , that is, after all nodes are allocated according to the partition mapping , the number of nodes belonging to the th region; is the number of regions (partition number), and the city big graph is divided into regions (zones), is the total number of nodes; is the node belongs to, is the node belongs to; , is the adjustment weight, increase it if you want to balance the region size , increase it if you want to reduce the number of cut edges (reduce the cross-region edges) . ) is the region size balancing term, the purpose is to make each region size similar; ( ) is the number of cross-region edges, to control the connection of cross-region.

[0035] Based on the intersection relationship between the cross-region edges and the boundary of each region subgraph, the virtual connection point is determined, the split cost of the virtual connection point is determined, and the connection point virtual graph is constructed according to all the boundary connection points, the process is as follows: If the cross-region edge between each region subgraph intersects with the boundary and a certain node, a virtual connection point is generated at the node; Take the virtual connection point as the split point to determine the split ratio of the cross-region edge corresponding to the virtual connection point; If the split ratio corresponding to the split cost does not change the total travel cost and the correctness of the global shortest path, the virtual connection point is taken as the boundary connection point; Based on all the boundary connection points and the relationship between the boundary connection points, a connection point virtual graph is constructed.

[0036] First, determine the boundary connection point set . The boundary connection point is the Internet node between each region subgraph, which is equivalent to the gateway node on the region boundary, which is used for inter-regional path docking. That is, after the region division is completed, for any cross-region edge If it intersects with the boundary and a certain node , a virtual connection point is generated at the node , and the original cross-region edge is split into two virtual edges , . In order to make the cross-region edge reusable in ZPT and JPT, the split must meet: the path cost before splitting = the sum of the two segments after splitting, so a formal allocation method is needed to meet the following proportion formula:

[0037]

[0038] Among them, is the basic travel cost of the cross-region edge ; Splitting cost of two virtual edges obtained by splitting the cross-zone edge; two values are guaranteed The proportional allocation ensures that the splitting does not change the correctness of the total trip cost and the global shortest path, and requires embedding the connection points as intra-zone nodes before the intra-zone shortest path calculation to maintain the consistency of the local / global representation and create conditions for the construction of the virtual interconnection network.

[0039] Next, a connection point virtual graph is established between all boundary connection points , whose boundary connection edge set is composed of pairs of boundary connection points directly connected by cross-zone edges, which is a compressed representation of the original cross-zone path and is used to depict the backbone connectivity relationship between regions.

[0040] The construction of the virtual graph follows the following principles: any two boundary connection points belonging to adjacent regions and connected by a direct road in the original road network are connected by a boundary connection edge in the virtual graph, thereby, abstractly representing the connectivity backbone structure between regions, which is greatly reduced in scale compared to the original road network and only contains boundary connection points at the region boundaries and cross-zone edges.

[0041] Step S1.2: Calculate the all-pair shortest path matrix and predecessor pointer in each regional subgraph, and aggregate the time, energy, and delay triplets to obtain a partition path record. The partition path records of each regional subgraph constitute a partition path table, including: Each region Floyd-Warshall (all-source shortest path algorithm) is performed as an independent subgraph. According to the classic recursive formula, check if there is a shorter path through , if is allowed to be a transit, will it make shorter? The classic recursive formula is as follows:

[0042] wherein, is the node number in the regional subgraph, is the shortest path from node to node with the first nodes as intermediate nodes; is the shortest path without requiring to pass through node , i.e., the shortest distance from node to node under the premise of only allowing these nodes as transit; indicates first walking from node to node , and then from node Total distance from node to node , i.e. if forced to pass node , then the shortest total cost is what. That is, the present embodiment uses the iterative way of loyd-warshall to update the shortest path table by systematically covering all possible combinations of transit nodes.

[0043] Based on the above classical recursive formula, the all-pairs shortest distance matrix and predecessor pointer are obtained, which includes internal nodes and boundary connection points in the region subgraph. The predecessor pointer represents the previous node in the region subgraph.

[0044] To adapt the three-objective online optimization of the present embodiment, the shortest path in each region is calculated in the offline phase Pre-aggregation time, energy, and delay statistics triplets, as follows:

[0045] Wherein, , and is the average additional delay estimated by long-term statistics at the edge computing node / charging station node (0 if not). Finally, a path record item of a ZPT is formed, as follows:

[0046] Wherein, is the next-hop pointer (predecessor pointer), which is the identification of the next-hop node from node to node along the shortest path, used to expand the path by next-hop in the online phase is the path sequence compression encoding, is the basic distance metric used in the offline all-source shortest path calculation, is the region identification to which the path record belongs. The path record set of all region subgraphs constitutes the ZPT, and these ZPT path records are serialized and resident in storage as the result of offline pre-computation. The path record item is serialized and written into DB-Cache, so that the path segment can be directly extracted without real-time graph search in online query. In short, ZPT provides the optimal path and its cost summary between any two points in each region, so that graph search does not need to be performed again in the region in online query, which is one of the key innovative points of the present application to improve real-time response speed.

[0047] Step S1.3: Calculate the backbone shortest path on the connection point virtual graph to form a connection point path table, specifically including:​​​ Perform global shortest path calculation on the virtual graph of connection points to generate JPT offline.

[0048] Specifically, in Run classic shortest path algorithms (such as Dijkstra's algorithm) to calculate the shortest path between any pair of connected points; when multiple alternative backbone paths need to be generated, Yen's algorithm can also be used. The shortest path algorithm finds several suboptimal paths. Regardless of the algorithm used, the optimal (and suboptimal) path information between boundary connection points can be pre-calculated offline and recorded in the JPT. The JPT can be viewed as a lookup table. Through the above offline calculations, the JPT stores key data on backbone paths between regions, laying the foundation for rapid path splicing in the online phase.

[0049] Each record in JPT corresponds to a specific pair of start and end points. This includes the pre-calculated path and related cost information for the pair of origin and destination points, as shown in the following formula:

[0050] in, Virtual graph for connecting points Above, starting and ending points The shortest path scalar cost; This is a sequence of link points (used for expanding relay link points during online splicing), starting from the origin. Departure to destination The sequence of relay connection points traversed in sequence. By recording the jump relationships of relay connection points in the path, the cross-regional backbone path can be quickly reconstructed based on this sequence during the online phase. When multiple suboptimal backbone paths are generated using the Yen algorithm, the same pair can also be represented in JPT. Store multiple records, sorted in ascending order by path cost, for online selection of the best option.

[0051] During the online phase, leveraging the pre-generated partition path table ZPT and connector path table JPT, this invention employs a three-segment path concatenation strategy, significantly reducing real-time computation overhead. Specifically, any cross-regional global path is decomposed into three segments: a path segment within the source region, a connector backbone path segment, and a path segment within the destination region. The source region segment and the destination region segment directly obtain the shortest path within their respective regions from the ZPT by looking up the tables from the origin to a connector and from a connector to the destination, respectively; the intermediate connector backbone segment is determined based on the path provided by the JPT. The sequence obtains the corresponding cross-zone shortest path. Since only splicing combinations need to be performed on a limited candidate set, the system finally produces only a small number of candidate global paths, on the basis of which the optimal scheme that meets the constraints is selected. Compared with the traditional method of searching for a complete path on a large original road network map, the three-stage mechanism makes the online solution no longer perform high-cost traversal operations on the entire map, and the query process is equivalent to path segment extraction and splicing in constant time levels, significantly reducing the computational complexity of real-time path planning.

[0052] Example scenario: Assume that a vehicle departs from area A and needs to go to cross-city area C, passing through area B along the way. The ZPTs of areas A, B, and C and the JPTs of cross-regions are precomputed in the offline stage. In the online stage, first, the starting point is determined to be in area A and the ending point is determined to be in area C (which belongs to a cross-region scenario) according to the vehicle position and destination. Then, the shortest path from the starting point to the relay connection point X in area A (the source-in-zone segment) is extracted from the cached ZPT; the backbone shortest path from the relay connection point X in area A to the relay connection point Z via area B (the connection-point backbone segment, which may include the relay connection point Y of area B, etc.) is extracted from the JPT; and the shortest path from the relay connection point Z to the ending point in area C (the destination-in-zone segment) is extracted from the ZPT. Finally, by sequentially splicing the A-in-zone segment→X-Z backbone segment→C-in-zone segment, the complete A-to-C optimal path is obtained. In the entire process, each segment of the path is directly derived from the path table obtained by offline calculation, without any real-time map search on the original road network, thereby fully demonstrating the advantages of the layered routing scheme of the present application in terms of efficiency and practicality.

[0053] Step S2: Real-time monitoring of the operating states of roads, edge computing nodes, and charging stations, prediction of road congestion intensity, edge computing node queuing estimation information, and charging constraints, including: This step is responsible for converting the operating states of roads, edge computing nodes, and charging stations of the urban road network into time-varying variables that can be directly consumed by online scheduling, and injecting the real-time engine and DB-Cache metadata view with a unified time axis and version number.

[0054] Real-time monitoring of the operating states of roads, edge computing nodes, and charging stations, including: Road traffic data, including road segment travel speed, traffic volume, and road occupancy rate, etc. Such data reflect the current road congestion intensity (degree of traffic congestion), for example, if the average speed of a road segment decreases and the occupancy rate increases, the congestion intensity is high; Edge computing node data, performance telemetry from edge computing servers or roadside units, including task arrival rate (number of tasks received per unit of time), task service rate (number of tasks that can be processed per unit of time), etc. These data are used to compute the node load rate (i.e. how busy the node is, usually the ratio of arrival rate to service rate). Charging infrastructure inventory and queuing, edge computing node arrival and service rates; charging infrastructure inventory and queuing, edge computing node arrival and service rates.

[0055] Energy charging infrastructure data: monitoring information from the back-end of a charging station, such as the number of available charging piles (idle charging gun ports / vehicle spaces), the number of currently queued vehicles, and the power supply status of the station, etc. Through these data, the queue waiting time (the time a vehicle waits in the queue for charging at a charging station) can be estimated.

[0056] Predicted road congestion intensity, edge computing node queuing estimation information, and charging constraints, including: Congestion intensity for edge weight amplification , representing the degree of traffic congestion on the road edge at time ; Load rate for edge computing node queuing estimation , the arrival rate and service rate of computing tasks , respectively describing the queuing congestion level, task arrival, and task processing capacity of the computing node at time ; Energy adequacy for charging equality and inequality constraints , reflecting the ratio of the current remaining energy to the maximum capacity of the charging station node, the larger the value, the more energy is adequate, the more suitable for energy scheduling, and it also includes the average charging power and the charging queuing time .

[0057] After all quantities are cleaned, aligned, and filtered, they are published with timestamp and version identifier , so that the system can construct the edge weight amplification factor and node cost estimation on the same perception snapshot when solving in the online phase, thereby ensuring the repeatability and timing consistency of the path selection process.

[0058] For edge computing nodes, if a single-server approximation is used, the waiting time and actual processing time can be estimated through a simplified queuing model (such as M / M / 1), where the waiting time is as follows:

[0059] where , is the equivalent statistic for queue modeling, is the congestion level of the node in unit time, is the equivalent task processing capacity of the node in the current running state.

[0060] actual processing time with task size and effective computing power determined as follows:

[0061] wherein, is the available computing resource size of the node, is the effective throughput efficiency per unit computing power.

[0062] For a charging station node, if the number of parallel charging guns and the charging service rate are available, the expected charging queue time can be approximated by Erlang-C under the assumption of multiple service stations, as follows:

[0063] wherein, is calculated by according to the standard formula, is the number of vehicles predicted to arrive at the charging station node in unit time, is the number of charging services that a charging gun can complete in unit time, i.e., the charging service rate, is the number of parallel charging guns.

[0064] average charging power is given by the comprehensive consideration of the rated power of the pile, the grid-connected voltage state, and the temperature control current limit, as follows:

[0065] wherein, is estimated from station-level telemetry and historical statistics, thereby obtaining the expected charging time as follows:

[0066] wherein, is the target energy increase.

[0067] energy sufficiency , the calculation formula is as follows:

[0068] wherein, indicates the current remaining energy of the charging station node, ​Charging station node Maximum capacity of the charging station node, Reflects the ratio of the current remaining energy of the charging station node to the maximum capacity, the larger the value, the more energy is abundant, the more suitable for energy scheduling; To avoid using outdated current values in the online phase, forward-looking samples along the path are provided to the scheduler. Let the edge sequence of a candidate path be , under the premise that the online solution is at the entry time , the expected arrival time of the edge is obtained by sequentially accumulating the prefix travel time. In the coupling with the online engine, the submodule publishes a situation snapshot at a fixed period , and triggers an event frame in the event of a sudden event (stake offline, node degradation, accident closure). The snapshot timestamp (i.e. the entry time of the online solution) and the cache version number are bound at the beginning of each online solution, forming a consistent view of .

[0069] Step S3: Determine the candidate path set in the partition path table and the junction path table according to the online request of the vehicle, and screen each candidate path in the candidate path set based on road congestion intensity, edge computing node queuing estimation information and charging constraint to determine the optimal path, including: This step is based on the partition path table (ZPT) and the junction path table (JPT) generated in the offline phase, and the path segments and metadata are obtained in the database buffer (DB-Cache) in a read-only manner. After receiving the online request of the vehicle, the road, computing power and energy state are integrated to complete the three-segment splicing of the candidate path, the feasibility label propagation and the minimum comprehensive cost selection, and the optimal path containing "travel path, energy supplement, task offloading" is generated. The input includes the starting and ending node identifier, the current state of charge (SoC) of the vehicle, the time limit for reaching and the task size constraint parameters, as well as the road congestion intensity, the edge computing node queuing estimation information and the charging constraint short-term situation quantity. The output is the feasible optimal or front joint scheme and its planned arrival time, energy supplement power and time slot, task allocation and pre-queuing information.

[0070] Step S.3.1: Determine the candidate path set in the partition path table and the junction path table according to the online request of the vehicle, including: Determine the starting area and the ending area according to the online request of the vehicle; If the starting area and the ending area are the same area, determine the shortest path from the starting point to the ending point in the area according to the partition path table to form the candidate path set; If the starting region and the ending region are not the same region, then determine the path within the starting region subgraph, determine the backbone path within the connection point virtual graph, and determine the path within the ending region subgraph. Concatenate the three paths to obtain the candidate path set.

[0071] During the online phase, first locate the starting point. and the finish line area.

[0072] like Directly hit from ZPT The shortest path within the region constitutes the basic element of the candidate path set.

[0073] like If the starting point and the ending point are located in different regions, then the path query will construct a candidate path set by concatenating three segments. .

[0074] First, within the subgraph containing the starting point, enumerate all boundary connection points within that subgraph to form the set of connection points for the starting point region. These connection points are nodes that have direct cross-region edges with other regions. Retrieval starting point. Connecting points to the boundaries of each starting region The path segments form the first segment: the path segment within the source region.

[0075] Next, in the virtual graph of the connection points The above enumeration from Starting from each boundary connection point in the middle, the set of connection points leading to the target area. The backbone path of any connection point in the network. Indicates the end point All boundary connection points within the area. This constitutes the second segment: the cross-regional backbone path segment.

[0076] Finally, within the region where the endpoint is located, retrieve the connection points from the boundary of each target region. To the finish line The path segments are then combined to form the third segment: the path segment within the target area. A candidate path set is formed by concatenating these three path segments. Compared to traversing all paths across the entire graph, this method effectively reduces the search space and significantly improves online solution efficiency.

[0077] Feasibility verification adopts the label propagation framework of Resource-Constrained Shortest Path (RCSP).

[0078] For candidate path set For each candidate path, the vehicle status labels are sequentially assigned along the path. propagation is performed, wherein is the current node, is the SoC normalization level, is the task completion status, is the accumulated travel time. When passing edge , if is a computing node or a charging station node, the label is obtained according to the following set of formulas , the formulas are as follows:

[0079]

[0080]

[0081] mandatory constraints during propagation and and requires when reaching the destination. To control the online complexity, the system adopts a dominance rule: if two labels reaching the same node satisfy and the accumulated cost , then is deleted.

[0082] Step S3.2: According to the road congestion intensity, edge computing node queuing estimation information and charging constraints, determine the edge weight cost, computing task execution cost and charging cost, including: To control the candidate size, the set of starting area connection points and the set of target area connection points may be limited according to geometric proximity, expected travel time threshold or connection degree of the connection points, so as to avoid forming a too large path combination space in the whole graph. Each path in the candidate path set is spliced by the precomputed path segments in ZPT and JPT, and in the online stage, only needs to be unfolded by and the forward pointer lookup table, and no longer needs to perform real-time graph search, which greatly reduces the online calculation overhead.

[0083] To reflect the influence of dynamic running situation on the pros and cons of the path, the comprehensive cost weight is integrated, and the basic part is composed of the basic travel time and energy consumption of the road segment, and the two are weighted to form the basic edge weight through the weight coefficient.

[0084] The basic edge weight is defined for each edge , as follows:

[0085] wherein, is the base travel time, is the equivalent energy consumption, is the strategy weight.

[0086] Road congestion, computing load and energy sufficiency are embedded in edge weights with amplification factors to adjust the current cost of edges, and the amplification factors are constructed as follows:

[0087] Finally, based on the base edge weight and the amplification factor, the dynamic edge weight is determined as follows:

[0088] wherein, is the node computing load rate, is the energy available ratio, that is, the energy sufficiency, is the congestion intensity, . The dynamic edge weight is the adjustment result after further considering the current system situation (such as congestion, computing load, energy shortage, etc.) on the basis of the base edge weight.

[0089] For the case of needing to perform a task or energy supplement at a node, the system does not explicitly model this operation as stopping at the node, but rather adds an equivalent service cost (such as waiting time and processing time) on the adjacent edge connected to the node to avoid destroying the reusability of the path segment.

[0090] In the modeling of node service cost, the queuing time of the task can be simplified to estimate as the "current load rate" divided by the "remaining processing capacity", and the processing time is calculated according to the task size and available computing power. The waiting time on the computing side and the actual processing time can be approximately recorded as:

[0091]

[0092]

[0093] wherein, is the service rate, is the arrival rate, is the available computing resource size of the node, is the effective throughput efficiency of unit computing power, is the task size, is the actual processing time.

[0094] ​Similarly, the service time of a charging station node is determined by both the current queuing time and the average charging rate. The sum of the queuing time and the charging time on the charging side is used as the expected charging time. It can be approximated as:

[0095] in, For charging queue time, Boost energy for the goal This represents the equivalent average charging power.

[0096] The dynamic weights on the road side are used as edge weight costs; the waiting time and actual processing time on the computation side are used as computation task execution costs; and the expected charging time on the energy replenishment side is used as charging costs. All of the above costs are weighted and accumulated into the total path cost based on whether the node is selected to perform the relevant operation, and are used for path optimization and ranking.

[0097] Step S3.3: With the goal of optimizing the comprehensive cost consisting of edge weight cost, computation task execution cost, and charging cost, each candidate path in the candidate path set is screened to determine the optimal path, including: In the candidate path set The above uses the combined cost of the three objectives as the optimization criterion to evaluate each candidate path. Overall cost:

[0098] in, In candidate paths The set of edge computing nodes that actually perform computing tasks. In candidate paths The set of charging station nodes that actually perform energy replenishment. For strategy weights. Indicate candidate path The sum of the dynamic costs of all edges (i.e. road segments) under the current system state, i.e. the sum of edge weight costs; For the part of calculating the task execution cost, Candidate paths The set of nodes that actually perform the computational tasks. The policy weights reflect the system's emphasis on computational task latency. For computational tasks at nodes The waiting time For computational tasks at nodes The actual processing time. For the cost of charging, Indicate candidate path The set of charging station nodes (charging stations) that actually provide energy replenishment. is a strategy weight, represents the degree of attention to the charging cost, including the charging queuing time at the node + the actual charging time required.

[0099] According to the ranking of the comprehensive cost of each candidate path, the candidate path with the minimum comprehensive cost is selected as the optimal path.

[0100] The embodiment adopts offline recalculation and online light splicing as the base in large-scale urban road network, superimposes two-layer mechanisms of energy and computing power cooperation and feasibility label propagation, so that online decision only completes the feasible optimal of three targets of time, energy consumption and task in small-scale candidate set. Offline partition, connection point and two types of pre-computed path table (zone path table, ZPT and junction path table, JPT) reside in database cache (database cache, DB-Cache), which ensures that the same area query can hit, cross-area query can generate candidate set through three splicing, and then select the optimal path on the candidate set using Dijkstra or Yen. DB-Cache supports parallel access and incremental write-back, which is convenient for rapid update under the change of topology or weight.

[0101] Embodiment two The embodiment provides a layered dynamic routing system for a computing power-energy fusion network, comprising: An offline pre-computation module is configured to divide a vehicle-energy-computing fusion network graph into a plurality of regional subgraphs, and generate a zone path table and a junction path table based on each regional subgraph; A state monitoring and prediction module is configured to monitor the operating states of roads, edge computing nodes and charging stations in real time, and predict road congestion intensity, edge computing node queuing estimation information and charging constraints; A cooperative routing module is configured to determine a candidate path set in the zone path table and the junction path table according to online requests of vehicles, screen each candidate path in the candidate path set based on road congestion intensity, edge computing node queuing estimation information and charging constraints, and determine an optimal path.

[0102] The above modules and corresponding steps achieve the same examples and application scenarios as the above embodiment one, but are not limited to the content disclosed in the above embodiment one. It should be noted that the above modules as part of the system can be executed in a computer system such as a group of computer executable instructions.

[0103] The description of each embodiment in the above embodiments has its own emphasis, and the parts not described in detail in a certain embodiment can refer to the related description of other embodiments.

[0104] The proposed system can be implemented in other ways. For example, the system embodiments described above are only illustrative, and for example, the division of the above modules is only a logical functional division, and in actual implementation, other division manners can be used, for example, a plurality of modules can be combined or integrated into another system, or some features can be ignored or not executed.

[0105] Embodiment three The embodiment provides a computer readable storage medium, which stores a computer program, and the program is executed by a processor to implement steps in the layered dynamic routing method for the computing power-energy fusion network.

[0106] Embodiment four The embodiment provides a computer device, which includes a memory, a processor and a computer program stored in the memory and executable on the processor, and the processor implements steps in the layered dynamic routing method for the computing power-energy fusion network when executing the program.

[0107] Embodiment five The embodiment provides a computer program product or a computer program, which includes computer instructions stored in a computer readable storage medium, and a processor of a computer device reads the computer instructions from the computer readable storage medium, and the processor executes the computer instructions, so that the computer device executes steps in the layered dynamic routing method for the computing power-energy fusion network.

[0108] Those skilled in the art should understand that the embodiments of the present application can provide a method, a system or a computer program product. Therefore, the present application can take the form of a hardware embodiment, a software embodiment or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer usable storage media (including but not limited to disk storage and optical storage, etc.) containing computer usable program code.

[0109] The present application is described with reference to flowcharts and / or block diagrams of the method, device (system) and computer program product according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams and the combination of the flows and / or blocks can be implemented by computer program instructions. These computer program instructions can be provided to a general purpose computer, a special purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the computer or other programmable data processing device produce a device that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 One flow or multiple flows and / or blocks Figure 1apparatuses that carry out or implement designated functions in one or more blocks or multiple blocks.

[0110] These computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture including instructions which implement the flow Figure 1 one or more flows and / or blocks Figure 1 one or more blocks or multiple blocks.

[0111] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the flow Figure 1 one or more flows and / or blocks Figure 1 one or more blocks or multiple blocks.

[0112] Those of ordinary skill in the art can understand that all or part of the flow of the above-mentioned embodiment methods can be completed by a computer program instructing relevant hardware, and the program can be stored in a computer readable storage medium. When the program is executed, it can include the flow of the above-mentioned embodiment methods. The storage medium can be a magnetic disc, an optical disc, a read-only memory (ROM), a random access memory (RAM), or the like.

[0113] Although the specific embodiments of the present application are described above with reference to the drawings, the description is not a limitation on the scope of protection of the present application. Those of ordinary skill in the art should understand that various modifications or variations made on the basis of the technical solutions of the present application without creative labor are still within the scope of protection of the present application.

Claims

1. A hierarchical dynamic routing method for computing power-energy fusion networks, characterized in that, include: The vehicle-energy-computing fusion network graph is divided into several regional subgraphs, and a partition path table and a connection point path table are generated based on each regional subgraph. Real-time monitoring of the operational status of roads, edge computing nodes, and charging stations; prediction of road congestion intensity, edge computing node queuing estimation information, and charging constraints. Based on the online requests from vehicles, a set of candidate paths is determined in the partition path table and the connector path table. Each candidate path in the set is then filtered based on road congestion intensity, edge computing node queuing estimation information, and charging constraints to determine the optimal path.

2. The hierarchical dynamic routing method for computing power-energy fusion networks as described in claim 1, characterized in that, The process of dividing the vehicle-energy-computing fusion network graph into several regional subgraphs and generating a partition path table and a connection point path table based on each regional subgraph includes: The vehicle-energy-computing fusion network graph is divided into several regional subgraphs according to spatial relationships. Based on the boundary connection points of each regional subgraph, a virtual graph of connection points is generated. Calculate the shortest path matrix and predecessor pointers for each region subgraph, and aggregate the triplet of time, energy and delay statistics to obtain a partition path record. The partition path records of each region subgraph form a partition path table. Calculate the shortest path of the backbone on the virtual graph of the connection points to form a connection point path table.

3. The hierarchical dynamic routing method for computing power-energy fusion networks as described in claim 1, characterized in that, The process of dividing the vehicle-energy-computing fusion network graph into several regional subgraphs according to spatial relationships, and generating a virtual graph of connection points based on the boundary connection points of each regional subgraph, includes: Using roads, charging stations, and edge computing in the global traffic, energy, and computing power fusion graph as nodes, and the basic travel time from road to charging station and equivalent energy consumption as fusion edges, a vehicle-energy-computing fusion network graph is constructed. The regional division mapping is determined according to spatial relationships, and the vehicle-energy-computing fusion network graph is divided into several regional subgraphs based on the regional division mapping. Virtual connection points are determined based on the intersection relationship between cross-region edges and boundaries in each subgraph. Boundary connection points are determined based on the splitting cost of the virtual connection points. A virtual graph of connection points is constructed based on all boundary connection points.

4. The hierarchical dynamic routing method for computing power-energy fusion networks as described in claim 1, characterized in that, The step of determining the candidate path set in the partition path table and the connector path table based on the vehicle's online request includes: The origin and destination areas are determined based on the vehicle's online request. If the starting region and the ending region are the same region, the shortest path from the starting point to the ending point within the region is determined according to the partition path table, forming a candidate path set; If the starting region and the ending region are not the same region, then determine the path within the starting region subgraph, determine the backbone path within the connection point virtual graph, and determine the path within the ending region subgraph. Concatenate the three paths to obtain the candidate path set.

5. A hierarchical dynamic routing method for computing power-energy fusion networks as described in claim 1, characterized in that, The process of filtering each candidate path in the candidate path set based on road congestion intensity, edge computing node queuing estimation information, and charging constraints to determine the optimal path includes: Based on road congestion intensity, edge computing node queuing estimation information, and charging constraints, determine the edge weight cost, computing task execution cost, and charging cost; With the goal of optimizing the comprehensive cost consisting of edge weight cost, computation task execution cost, and charging cost, each candidate path in the candidate path set is screened to determine the optimal path.

6. A hierarchical dynamic routing method for computing power-energy fusion networks as described in claim 5, characterized in that, The process of determining the edge weight cost, computation task execution cost, and charging cost based on road congestion intensity, edge computing node queuing estimation information, and charging constraints includes: The basic edge weights are determined based on the basic travel time and energy consumption of the road segment. The amplification factor is determined by using the road congestion intensity, the load rate of the queuing estimation information of the edge computing nodes and the energy sufficiency of the charging constraints. The dynamic edge weights are determined based on the amplification factor and the basic edge weights, and are used as the edge weight cost. Based on the arrival rate and service rate of the computation task estimated by edge node queuing information, the waiting time and actual processing time are determined as the execution cost of the computation task. The expected charging time is determined by using the average charging power and the charging queue time, and this is taken as the charging cost.

7. A hierarchical dynamic routing system for computing power-energy fusion networks, characterized in that, include: The offline pre-computation module is used to divide the vehicle-energy-computing fusion network graph into several regional subgraphs and generate partition path tables and connection point path tables based on each regional subgraph. The status monitoring and prediction module is used to monitor the operating status of roads, edge computing nodes and charging stations in real time, and predict road congestion intensity, edge computing node queuing estimation information and charging constraints. The collaborative routing module is used to determine a set of candidate paths in the partition path table and the connector path table based on the online requests of vehicles. It then filters each candidate path in the set based on road congestion intensity, edge computing node queuing estimation information, and charging constraints to determine the optimal path.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by the processor, the program implements the steps in the hierarchical dynamic routing method for computing power-energy fusion networks as described in any one of claims 1-6.

9. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps in the hierarchical dynamic routing method for computing power-energy fusion networks as described in any one of claims 1-6.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the steps in a hierarchical dynamic routing method for a computing power-energy fusion network as described in any one of claims 1-6.

Images