Computing force load transfer method and system based on Dijkstra optimization algorithm
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-25
- Publication Date
- 2026-03-13
Smart Images
Figure CN121659547A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of power load transfer technology, specifically, it relates to a computing power load transfer method and system based on Dijkstra's optimization algorithm. Background Technology
[0002] Currently, the "computing load" of data centers, 5G base stations, and edge computing nodes is growing rapidly. Their power consumption characteristics are highly elastic, portable, and interruptible, providing new dispatchable resources for the power grid. At the same time, the grid connection of massive distributed photovoltaic, wind power, energy storage, and flexible resources makes it difficult for traditional power transfer models centered on "power balance" to simultaneously achieve economic efficiency, low carbon emissions, and reliability.
[0003] Existing technologies, such as dynamic shortest path planning methods based on improved Dijkstra's algorithm, Dijkstra's optimal traffic path planning methods and systems based on impedance matching, and comprehensive scheduling methods based on Dijkstra's algorithm, still have the following shortcomings: they do not incorporate the portability characteristics of computing load into the transfer decision, resulting in redundant transfer paths and high network losses; static weights cannot reflect the real-time output and electricity price signals of distributed power sources, leading to deviations between optimization results and economic scheduling objectives; when using data such as length, loss, reliability, and carbon emissions as multiple objectives, there is a lack of a unified quantification framework, resulting in low solution efficiency; and traditional Dijkstra's algorithm only considers a single path length, making it difficult to directly apply to novel source-load-computing coupled scenarios. Summary of the Invention
[0004] To address the shortcomings of existing technologies, this invention provides a computing power load transfer method and system based on Dijkstra's optimization algorithm, which solves the problems of correlation between real-time output of distributed power sources and electricity price signals, the establishment of a multi-objective unified quantization framework, and the adaptation to new source-load-computing coupling scenarios.
[0005] The present invention adopts the following technical solution.
[0006] This invention proposes a method for transferring computing power load based on Dijkstra's optimization algorithm, comprising: Real-time acquisition of distribution network line data and distributed power source operation data; establishment of a weighted topology map of the distribution network containing distributed power sources to identify fault areas and loads to be transferred to computing power. By utilizing the total length, total loss, and reliability indicators of the computing load transfer path, an evaluation index for the computing load transfer path is established. Starting from the distributed power node, with the goal of minimizing the evaluation index, the Dijkstra optimization algorithm is used to determine the optimal computing load transfer path.
[0007] Preferably, the distribution network line data includes: line length, equivalent impedance, line fault rate, and number of switching operations; the distributed power generation operation data includes: active power output of the distributed power generation and real-time energy price.
[0008] Preferably, based on the distribution network line data, the initial weights of each branch in the distribution network topology are set; based on the distributed generation operation data, the initial weights of the associated branches of the distributed generation access nodes are corrected to obtain dynamic weights, thus obtaining a weighted topology of the distribution network containing distributed generation.
[0009] Preferably, the distribution network topology is G=(V, E), where V={1,2,…,N} is the set of nodes, N is the number of nodes, and E is the set of branches. , )∈E, node , ∈V; If the nodes connected to the distributed power source are included in the special node set. V; branch road ( , The initial weights are as follows:
[0010] In the formula, For branch roads ( , ) length; , These are the maximum and minimum lengths of all branches, respectively; For branch roads ( , The equivalent impedance of ) , These represent the maximum and minimum equivalent impedances in all branches, respectively. For branch roads ( , Number of times the switch is operated; , These represent the maximum and minimum number of switch operations on all branches, respectively. , , The conversion factor satisfies + + =1.
[0011] Preferably, for special nodes ∈ The dynamic weights are obtained by correcting the initial weights of the associated branches of special nodes based on the distributed power source operation data, as shown in the following formula:
[0012] In the formula, For distributed power nodes Related branches ( , The dynamic weights in time period t, For special nodes The active power of the connected distributed power source in time period t. For special nodes The rated power of the connected distributed power source, For special nodes The real-time price of the connected distributed power source in time period t Sensitive functions, The influence factor of distributed generation, 0≤ ≤1; The sensitivity function uses a piecewise linear model, as shown in the following equation: =1, when < ; =0.5, when ≤ < ; =0, when ≥ ; In the formula, This represents the lower limit of real-time energy prices. This represents the upper limit of real-time energy prices.
[0013] Preferably, the total length and reliability index of the computing power load transfer path are determined based on the fault area, and the total loss is determined based on the computing power load to be transferred; a multi-objective evaluation vector is constructed using the total length, total loss, and reliability index of the computing power load transfer path. As shown in the following formula:
[0014] In the formula, Path for transferring computing load Total length (unit: km) Path for transferring computing load Total losses (unit: kW). Path for transferring computing load Reliability indicators; , For branch roads ( , ) length, For branch roads ( , The line failure rate, The computing load transfer path includes multiple branches connected end to end.
[0015] Preferably, the weight vector is determined using the analytic hierarchy process (AHP). The multi-objective evaluation vector is transformed into an evaluation index for the computing power load transfer path, as shown in the following formula:
[0016] In the formula, Path for transferring computing load Evaluation indicators; Before weighting , , Dimensionless processing is performed.
[0017] Preferably, Dijkstra's optimization algorithm is used to iteratively solve for the optimal power transfer path for computing load. ,include: The initial distance between the distributed power nodes is set to 0, and the distances between the remaining nodes are set to... ; Starting with distributed power nodes, and aiming to minimize the evaluation index, select the corresponding nodes to establish a priority queue of a minimum binary heap structure. In the priority queue of the minimum binary heap structure, the node-to-node computation of the distributed power source is performed. distance Distributed power nodes to adjacent nodes distance If node and adjacent nodes At least one node is a distributed power node, and > + Then update the adjacent nodes. distance = + And record adjacent nodes The predecessor node is a node If node and adjacent nodes Neither of them are distributed power nodes, and > + Then update the adjacent nodes. distance = + And record adjacent nodes The predecessor node is a node ; The calculation stops when the target load node is dequeued or the queue is empty, and the optimal transfer path Path* is generated based on the recorded predecessor node.
[0018] Preferably, the optimal transfer path is constrained and verified. If the verification fails, the evaluation index is updated and the optimal transfer path is regenerated. If the verification passes, the optimal transfer path is output. The constraint checks performed on the optimal power transfer path include: node voltage constraints, branch current constraints, and distributed power constraints.
[0019] This invention also proposes a computing power load transfer system based on Dijkstra's optimization algorithm, comprising: The data acquisition module is used to collect real-time data on power distribution network lines and distributed power source operation data. The power transfer planning module is used to establish a weighted topology map of the distribution network containing distributed power sources in order to identify fault areas and computing power loads to be transferred. The power transfer optimization module is used to establish evaluation indicators for the power transfer path by utilizing the total length, total loss, and reliability indicators of the power transfer path; starting from the distributed power node, it determines the optimal power transfer path for the power load with the goal of minimizing the evaluation indicators.
[0020] The present invention is also a terminal, including a processor and a storage medium; the storage medium is used to store instructions; the processor is used to perform operations according to the instructions to execute the steps of the method.
[0021] The present invention is also a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method.
[0022] The beneficial effects of this invention are as follows, compared with the prior art, at least including: This invention uses a dynamic weight correction mechanism to correct the weights of associated branches based on the real-time active power output and real-time energy prices of distributed power sources, enabling the weights to dynamically reflect the operating status and economic factors of distributed power sources, thereby making the optimization results more aligned with economic dispatch objectives; This invention constructs a multi-objective evaluation vector, determines the weight vector through expert evaluation or analytic hierarchy process, transforms the multi-objective optimization problem into a single-objective function, forms a unified quantitative framework, and improves solution efficiency; This invention improves the Dijkstra algorithm by comprehensively considering a multi-objective function including dynamic weights in path solving, rather than focusing solely on a single path length, making it adaptable to new source-load-computing coupling scenarios.
[0023] This invention constructs a dynamic weighted topology model incorporating distributed power sources, real-time adjusts branch weights, fully utilizes distributed power sources to reduce transfer costs, and adapts to dynamic changes in source and load. It integrates multi-objective optimization of path length, loss, and reliability, and achieves weight normalization through the analytic hierarchy process, balancing the economic and reliability requirements of computing load. A strict constraint verification mechanism is introduced to ensure that voltage, current, and distributed power source operation remain within safe ranges during the transfer process, improving the transfer success rate. The improved Dijkstra algorithm, while maintaining solution efficiency, enhances the global optimality of path optimization, making it suitable for large-scale distribution network computing load transfer scenarios. Attached Figure Description
[0024] Figure 1 This is a flowchart illustrating a computing power load transfer method based on Dijkstra's optimization algorithm proposed in this invention.
[0025] Figure 2 This is a computational example system topology and optimal transfer path diagram for the computing power load transfer method based on Dijkstra's optimization algorithm in an embodiment of the present invention. Detailed Implementation
[0026] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of this invention. The embodiments described in this application are merely some embodiments of this invention, and not all embodiments. Based on the spirit of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of this invention.
[0027] This invention proposes a computing load transfer method based on Dijkstra's optimization algorithm. By constructing a dynamic weighted topology model and integrating multi-objective optimization and constraint verification, it achieves fast, economical, and highly reliable computing load transfer.
[0028] like Figure 1 As shown, the method includes: Step S1: Real-time collection of power distribution network line data and distributed power source operation data.
[0029] Specifically, data on distribution network lines includes, but is not limited to: line length, equivalent impedance, line fault rate, and number of switching operations; data on distributed generation operation includes, but is not limited to: active power output of distributed generation and real-time energy price.
[0030] Step S2: Based on the distribution network line data, set the initial weights of each branch in the distribution network topology diagram; based on the distributed generation operation data, correct the initial weights of the associated branches of the distributed generation access nodes to obtain dynamic weights, and obtain the weighted topology diagram of the distribution network containing distributed generation.
[0031] Specifically, step S2 includes: Step S2.1: Based on the distribution network line data, set the initial weights of each branch in the distribution network topology diagram; The distribution network topology is G=(V, E), where V={1,2,…,N} is the set of nodes, N is the number of nodes, and E is the set of branches. , )∈E, node , ∈V; If the nodes connected to the distributed power source are included in the special node set. V; branch road ( , The initial weights are as follows:
[0032] In the formula, For branch roads ( , Length of ) (unit: km); , These are the maximum and minimum lengths of all branches (unit: km); For branch roads ( , Equivalent impedance (unit: pu); , These are the maximum and minimum equivalent impedances in all branches (unit: pu); For branch roads ( , Number of times the switch is operated; , These represent the maximum and minimum number of switch operations on all branches, respectively. , , The conversion factor satisfies + + =1.
[0033] Step S2.2: Based on the distributed power source operation data, correct the initial weights of the associated branches of the distributed power source access node to obtain dynamic weights. Specifically, for special nodes ∈ The dynamic weights are obtained by correcting the initial weights of the associated branches of special nodes based on the distributed power source operation data, as shown in the following formula:
[0034] In the formula, For distributed power nodes Related branches ( , The dynamic weights in time period t, For special nodes The active power of the connected distributed power source in time period t. For special nodes The rated power of the connected distributed power source, For special nodes The real-time price of the connected distributed power source in time period t Sensitive functions, The influence factor of distributed generation, 0≤ ≤1; The sensitivity function uses a piecewise linear model, as shown in the following equation: =1, when < ; =0.5, when ≤ < ; =0, when ≥ ; In the formula, This represents the lower limit of real-time energy prices. This represents the upper limit of real-time energy prices; the optimal values for the lower and upper limits of real-time energy prices need to be determined by considering the actual range of energy market price fluctuations and the system's economic objectives. In the example, Set the price to the benchmark electricity price or off-peak electricity price, with a preferred value of 0.5 yuan / kWh. Set the peak-hour electricity price, with a preferred value of 1.0 yuan / kWh; Therefore, branches associated with distributed power access nodes correspond to dynamic weights, while branches not associated with distributed power access nodes correspond to initial weights. Existing technologies do not incorporate the portability of computing load into power transfer decisions. This invention constructs a dynamically weighted topology model containing distributed power sources, adjusts branch weights in real time, fully considers the portability of computing load, avoids redundancy in power transfer paths, and reduces network losses.
[0035] The dynamic weight correction mechanism proposed in this invention aims to adjust branch weights based on the active power output of distributed generation sources and real-time energy prices at different times. When the active power output of a distributed generation source is large, it indicates that it can transfer less electricity, and the importance of the associated branches of the distributed generation source access node in the transfer process is relatively reduced, so the weight of the branch should be appropriately decreased. Real-time energy prices reflect the economic value of electricity. When energy prices are high, from an economic perspective, the weight of the associated branches of the distributed generation source access node should also be reduced to lower the electricity transmission cost on the branch. Existing static weights cannot reflect the real-time output and price signals of distributed generation sources. This invention, through a dynamic weight correction mechanism, enables the selection of transfer paths to better conform to the economic efficiency and rationality of actual grid operation. It allows the weights to dynamically reflect the operating status and economic factors of distributed generation sources, making the optimization results more aligned with economic dispatch objectives and the transfer decisions more in line with actual conditions. This avoids the problem of decoupling from actual operation caused by static weights, and improves the economic efficiency and reliability of the transfer scheme.
[0036] Step S3: Based on the weighted topology graph, determine the fault area and the computing power load to be transferred; In the weighted topology of a distribution network containing distributed power sources, fault nodes and their impact ranges are located to determine the fault area, the power supply demand of target load nodes is clarified, and computing power-power collaborative technology is used to determine the computing power loads to be transferred.
[0037] Step S4: Establish evaluation indicators for computing load transfer paths using the total length, total loss, and reliability of the computing load transfer path; starting from the distributed power node, determine the optimal computing load transfer path with the goal of minimizing the evaluation indicators.
[0038] Specifically, step S4 includes: Step S4.1: Determine the total length and reliability index of the computing power load transfer path based on the fault area, and determine the total loss based on the computing power load to be transferred; construct a multi-objective evaluation vector based on the total length, total loss, and reliability index of the computing power load transfer path. As shown in the following formula:
[0039] In the formula, Path for transferring computing load Total length (unit: km) Path for transferring computing load Total losses (unit: kW). Path for transferring computing load Reliability indicators; In the embodiments, , For branch roads ( , ) length (unit: km), For branch roads ( , The line failure rate, The computing load transfer path includes multiple branches connected end to end.
[0040] Step S4.2: Determine the weight vector using the analytic hierarchy process (AHP). The multi-objective evaluation vector is transformed into an evaluation index for the computing power load transfer path, as shown in the following formula:
[0041] In the formula, Path for transferring computing load Evaluation indicators; Before weighting , , Perform dimensionless processing; It is worth noting that using the analytic hierarchy process (AHP) to transform multi-objective evaluation vectors into evaluation indicators for computing power load transfer paths is a non-restrictive and relatively optimal choice.
[0042] Existing technologies, when dealing with multiple objectives such as length, loss, reliability, and carbon emissions, lack a unified quantification framework and suffer from low solution efficiency. This invention constructs a multi-objective evaluation vector and transforms the multi-objective optimization problem into a single-objective function, forming a unified quantification framework and improving solution efficiency.
[0043] Step S4.3: Using Dijkstra's optimization algorithm, the optimal power transfer path for computing power load is iteratively solved. ,include: Step S4.3.1: Set the initial distance of the distributed power node to 0, and the distance of the remaining nodes to... ; Step S4.3.2: Starting from the distributed power node, select the corresponding node to establish a priority queue of minimum binary heap structure with the goal of minimizing the evaluation index. Existing Dijkstra algorithms establish priority queues based on minimizing the branch distance between nodes. While such priority queues can satisfy power dispatching objectives, they cannot meet the computational dispatching objective centered on computational efficiency. However, there are barriers between the modeling standards of energy systems and computing network systems, making it impossible to directly use computing network system models to replace distribution networks. Therefore, this invention improves the priority queue establishment method in existing Dijkstra algorithms by minimizing the evaluation index of the computational load transfer path as the collaborative objective of the energy system and computing network system. The priority queue established accordingly ensures the achievement of power dispatching objectives from the perspective of reducing power transmission losses and improving power supply reliability, and also ensures the achievement of computational dispatching objectives from the perspective of minimizing the computational load transfer path.
[0044] Step S4.3.3: In the priority queue of the minimum binary heap structure, calculate the node-to-node ratio of the distributed power source. distance Distributed power nodes to adjacent nodes distance If node and adjacent nodes At least one node is a distributed power node and > + Then update the adjacent nodes. distance = + And record adjacent nodes The predecessor node is a node If node and adjacent nodes None of them are distributed power nodes and > + Then update the adjacent nodes. distance = + And record adjacent nodes The predecessor node is a node ; Traditional methods only consider a single path length, making it difficult to directly apply Dijkstra's algorithm to novel source-load-computation coupled scenarios. This invention improves Dijkstra's algorithm by comprehensively considering a multi-objective function with dynamic weights in the path solving process, enabling it to adapt to novel source-load-computation coupled scenarios.
[0045] Step S4.3.4: Stop the calculation when the target load node is dequeued or the queue is empty, and generate the optimal transfer path based on the recorded predecessor node. .
[0046] Step S4.4: Perform constraint verification on the optimal transfer path. If the verification fails, return to step S4.2 to update the evaluation indicators and then execute step S4.3 to regenerate the optimal transfer path. If the verification passes, output the optimal transfer path. Among them, the optimal transfer path The constraint checks performed include, but are not limited to: node voltage constraints, branch current constraints, and distributed power constraints.
[0047] While the analytic hierarchy process (AHP) for determining weights and constructing multi-objective evaluation vectors is, to some extent, a feasible approach using existing technologies, this invention combines both for computing power load transfer scenarios. Furthermore, it integrates these with a dynamic weight correction mechanism, an improved Dijkstra algorithm, and other overall technical solutions to form a complete and innovative solution for computing power load transfer in novel source-load-computing coupled scenarios. Multi-objective weight normalization is a key step in solving multi-objective optimization problems, improving solution efficiency, and adapting to new scenarios. Moreover, it fully utilizes distributed power sources to reduce transfer costs and adapt to dynamic changes in source and load; it balances the economic and reliability requirements of computing power loads; and while ensuring solution efficiency, it improves the global optimality of path optimization, making it suitable for computing power load transfer scenarios in large-scale distribution networks.
[0048] Taking the computing load transfer of a data center in an industrial park as an example, such as Figure 2 As shown, the power distribution network in the park contains 10 nodes, V={1,2,…,10}, where node 5 is connected to a distributed photovoltaic power source with a rated capacity of S=500kW, and node 8 is the target computing power load node with a load demand of 800kW; assuming that the line from node 3 to node 4 fails, the power supply needs to be transferred from power node 1 to node 8.
[0049] Real-time collected parameters: Line parameters: Taking branch (1,2) as an example, its line length L(1,2)=0.5km, equivalent impedance Z(1,2)=0.02pu, and number of switching operations S(1,2)=2 times; Distributed power source: Active power output of node 5 P5(t)=300kW, real-time energy price C5(t)=0.6 yuan / kWh; Reliability parameters: Failure rate of line (5,6) λ(5,6)=0.05 times / year, number of power supply users N(5,6)=50 households; Constraint thresholds: U_min=0.95pu, U_max=1.05pu; I_max=600A; S(t)=500kVA.
[0050] The power distribution network of the computing center is abstracted as graph G=(V,E); where the node set V contains all distribution nodes within the park, such as substation nodes, load nodes, and distributed power source access nodes; the branch set E represents the connecting lines between the nodes. Nodes connected to distributed power source DG1 are classified into a special node set. The special node set V_s={5}; initial weights are calculated: let... =0.3, =0.5, =0.2, initial weight of branch (5,6) (5,6)=0.3×0.8+0.5×0.03+0.2×1=0.24+0.015+0.2=0.455.
[0051] Assuming distributed power supply The active power output during the time period t is The distributed energy source has a capacity of 300kW and a rated capacity of 500kW. The real-time energy price is C(t) = 0.8 yuan / kWh. Given that the distributed energy impact factor α = 0.6, in the price sensitivity function, when C(t) < 0.6 yuan / kWh, β(C(t)) = 1; when 0.6 yuan / kWh ≤ C(t) < 1.0 yuan / kWh, β(C(t)) = 0.5; and when C(t) ≥ 1.0 yuan / kWh, β(C(t)) = 0. Therefore, the price sensitivity function β(C(t)) = 0.5. Then, the dynamic weight... (5,6)=0.455×[1-0.6×(300 / 500)×0.5]=0.455×[1-0.6×0.6×0.5]=0.455×(1-0.18)=0.455×0.82≈0.373.
[0052] Construct a multi-objective evaluation vector and determine the weight vector using the Analytic Hierarchy Process (AHP). = (0.2, 0.5, 0.3), transforming the multi-objective optimization problem into a single-objective function: The evaluation vector for a candidate path Path=1-2-5-6-8: =2.5km, =15kW, =0.08; Single objective function value =0.2×2.5+0.5×15+0.3×0.08=0.5+7.5+0.024=8.024.
[0053] Solving using Dijkstra's optimization algorithm The distance d(s) of the distributed power node s is 0, and the distance d(v) of the other nodes is ∞; a minimum binary heap priority queue is used, and the current node is selected each time. The smallest node; path backtracking: generating the optimal transfer path Path*=1-2-5-6-8, its =8.024; The voltage at node 5 is U(5) = 1.02pu (satisfying the constraint); the current in branch (5,6) is I = 450A ≤ 600A; the apparent power of the distributed generation... = ≈424kVA≤500kVA, both constraints are met; Output optimal load transfer path Path*: 1-2-5-6-8; Evaluation indicators: total length of transfer path L*=2.5km, total loss P*=15kW, reliability index SAIFI*=0.08, dynamic weight correction coefficient. (5,6)=0.373.
[0054] Table 1. Comparison of evaluation metrics between the traditional Dijkstra algorithm and the Dijkstra optimization algorithm of this invention.
[0055] Compared to the traditional Dijkstra algorithm, the method of this invention reduces the total power transfer path loss by 18.3%, improves the SAIFI index by 12.5%, and increases the distributed resource utilization rate by 25%, effectively ensuring reliable power supply for computing loads. The dynamic weight correction mechanism makes the power transfer path optimization results more aligned with economic scheduling objectives, improving the utilization rate of distributed power sources by 25% compared to traditional algorithms. The multi-objective normalization framework improves the overall performance of path optimization, reducing the total power transfer path loss by 18.3%, balancing economy and reliability. The improved Dijkstra algorithm is adapted to new source-load-computing coupling scenarios, and combined with the constraint verification mechanism, it ensures the safety and stability of computing load power transfer, improving the SAIFI reliability index by 12.5%. Constraint verification and secondary optimization ensure that voltage, current, and distributed power sources operate within safe ranges during the power transfer process, improving the power transfer success rate.
[0056] This invention also proposes a computing power load transfer system based on Dijkstra's optimization algorithm, comprising: The data acquisition module is used to collect real-time data on power distribution network lines and distributed power source operation data. The power transfer planning module is used to establish a weighted topology map of the distribution network containing distributed power sources in order to identify fault areas and computing power loads to be transferred. The power transfer optimization module is used to establish evaluation indicators for the power transfer path by utilizing the total length, total loss, and reliability indicators of the power transfer path; starting from the distributed power node, it determines the optimal power transfer path for the power load with the goal of minimizing the evaluation indicators.
[0057] This disclosure can be a system, method, and / or computer program product. A computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for causing a processor to implement various aspects of this disclosure.
[0058] Computer-readable storage media can be tangible devices capable of holding and storing instructions for use by an instruction execution device. Computer-readable storage media can be, for example—but not limited to—electrical storage devices, magnetic storage devices, optical storage devices, electromagnetic storage devices, semiconductor storage devices, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of computer-readable storage media include: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static random access memory (SRAM), portable compact disc read-only memory (CD-ROM), digital multifunction disc (DVD), memory sticks, floppy disks, mechanical encoding devices, such as punch cards or recessed protrusions storing instructions thereon, and any suitable combination of the foregoing. The computer-readable storage media used herein are not to be construed as transient signals themselves, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through waveguides or other transmission media (e.g., light pulses through fiber optic cables), or electrical signals transmitted through wires.
[0059] The computer-readable program instructions described herein can be downloaded from computer-readable storage media to various computing / processing devices, or downloaded via a network, such as the Internet, local area network, wide area network, and / or wireless network, to an external computer or external storage device. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards them to the computer-readable storage media in the respective computing / processing device.
[0060] Computer program instructions used to perform the operations of this disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, status setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages such as Smalltalk, C++, etc., and conventional procedural programming languages such as the "C" language or similar programming languages. The computer-readable program instructions may execute entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuitry, such as programmable logic circuitry, field-programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), is personalized by utilizing the status information of the computer-readable program instructions to implement various aspects of this disclosure.
[0061] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.
Claims
1. A method for transferring computing power load based on Dijkstra's optimization algorithm, characterized in that, include: Real-time acquisition of power distribution network line data and distributed power source operation data; Establish a weighted topology map of the distribution network containing distributed power sources to identify fault areas and loads to be transferred to computing power. By utilizing the total length, total loss, and reliability indicators of the computing load transfer path, an evaluation index for the computing load transfer path is established. Starting from the distributed power node, with the goal of minimizing the evaluation index, the Dijkstra optimization algorithm is used to determine the optimal computing load transfer path.
2. The computing power load transfer method based on Dijkstra's optimization algorithm according to claim 1, characterized in that, Distribution network line data includes: line length, equivalent impedance, line fault rate, and number of switching operations; distributed generation operation data includes: active power output of distributed generation and real-time energy price.
3. The computing power load transfer method based on Dijkstra's optimization algorithm according to claim 1, characterized in that, Based on the distribution network line data, set the initial weights of each branch in the distribution network topology diagram; Based on the operation data of distributed generation, the initial weights of the associated branches of the distributed generation access nodes are corrected to obtain dynamic weights, resulting in a weighted topology of the distribution network containing distributed generation.
4. The computing power load transfer method based on Dijkstra's optimization algorithm according to claim 3, characterized in that, The distribution network topology is G=(V, E), where V={1,2,…,N} is the set of nodes, N is the number of nodes, and E is the set of branches. , )∈E, node , ∈V; If the nodes connected to the distributed power source are included in the special node set. V; branch road ( , The initial weights are as follows: In the formula, For branch roads ( , ) length; , These are the maximum and minimum lengths of all branches, respectively; For branch roads ( , The equivalent impedance of ) , These represent the maximum and minimum equivalent impedances in all branches, respectively. For branch roads ( , Number of times the switch is operated; , These represent the maximum and minimum number of switch operations on all branches, respectively. , , The conversion factor satisfies + + =1.
5. The computing power load transfer method based on Dijkstra's optimization algorithm according to claim 4, characterized in that, For special nodes ∈ The dynamic weights are obtained by correcting the initial weights of the associated branches of special nodes based on the distributed power source operation data, as shown in the following formula: In the formula, For distributed power nodes Related branches ( , The dynamic weights in time period t, For special nodes The active power of the connected distributed power source in time period t. For special nodes The rated power of the connected distributed power source, For special nodes The real-time price of the connected distributed power source in time period t Sensitive functions, The influence factor of distributed generation, 0≤ ≤1; The sensitivity function uses a piecewise linear model, as shown in the following equation: =1, when < ; =0.5, when ≤ < ; =0, when ≥ ; In the formula, This represents the lower limit of real-time energy prices. This represents the upper limit of real-time energy prices.
6. The computing power load transfer method based on Dijkstra's optimization algorithm according to claim 1, characterized in that, The total length and reliability index of the computing power load transfer path are determined based on the fault area, and the total loss is determined based on the computing power load to be transferred. A multi-objective evaluation vector is constructed using the total length, total loss, and reliability index of the computing power load transfer path. As shown in the following formula: In the formula, Path for transferring computing load Total length, Path for transferring computing load Total loss, Path for transferring computing load Reliability indicators; , For branch roads ( , ) length, For branch roads ( , The line failure rate, The computing load transfer path includes multiple branches connected end to end.
7. The computing power load transfer method based on Dijkstra's optimization algorithm according to claim 6, characterized in that, Determining the weight vector using the analytic hierarchy process (AHP) The multi-objective evaluation vector is transformed into an evaluation index for the computing power load transfer path, as shown in the following formula: In the formula, Path for transferring computing load Evaluation indicators; Before weighting , , Dimensionless processing is performed.
8. The computing power load transfer method based on Dijkstra's optimization algorithm according to claim 7, characterized in that, The optimal power transfer path for computing power load is solved iteratively using Dijkstra's optimization algorithm. ,include: The initial distance between the distributed power nodes is set to 0, and the distances between the remaining nodes are set to... ; Starting with distributed power nodes, and aiming to minimize the evaluation index, select the corresponding nodes to establish a priority queue of a minimum binary heap structure. In the priority queue of the minimum binary heap structure, the node-to-node computation of the distributed power source is performed. distance Distributed power nodes to adjacent nodes distance If node and adjacent nodes At least one node is a distributed power node, and > + Then update the adjacent nodes. distance = + And record adjacent nodes The predecessor node is a node If node and adjacent nodes Neither of them are distributed power nodes, and > + Then update the adjacent nodes. distance = + And record adjacent nodes The predecessor node is a node ; The calculation stops when the target load node is dequeued or the queue is empty, and the optimal transfer path Path* is generated based on the recorded predecessor node.
9. The computing power load transfer method based on Dijkstra's optimization algorithm according to claim 8, characterized in that, The optimal transfer path is constrained and validated. If the validation fails, the evaluation index is updated and the optimal transfer path is regenerated. If the validation passes, the optimal transfer path is output. The constraint checks performed on the optimal power transfer path include: node voltage constraints, branch current constraints, and distributed power constraints.
10. A computing power load transfer system based on Dijkstra's optimization algorithm, used to implement the method according to any one of claims 1 to 9, characterized in that, include: The data acquisition module is used to collect real-time data on power distribution network lines and distributed power source operation data. The power transfer planning module is used to establish a weighted topology map of the distribution network containing distributed power sources in order to identify fault areas and computing power loads to be transferred. The power transfer optimization module is used to establish evaluation indicators for the power transfer path by utilizing the total length, total loss, and reliability indicators of the power transfer path; starting from the distributed power node, it determines the optimal power transfer path for the power load with the goal of minimizing the evaluation indicators.
11. A terminal, comprising a processor and a storage medium; characterized in that: The storage medium is used to store instructions; The processor is configured to operate according to the instructions to perform the steps of the method according to any one of claims 1-9.
12. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the steps of the method according to any one of claims 1-9.